Files
aevolution/docs/engineering/VALIDATION.md
T
dingoandClaude Opus 5.5 69e3866fd9 Tell the owner what its engulf and bud did, and why its cell died
M11.8 server half (ADR-0129): v10 diagnostics report the tick's engulf and bud
results in two reserved bytes, and v10 type 27 tells the owner its life ended with
the cause (starved or engulfed). The bridge fills both from the owner's tick
feedback and corpses; v9 views are unchanged. Client presentation stays with the
maintainer.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 07:57:37 -04:00

7461 lines
446 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# Validation report
Recorded 2026-09-19; subsequent gate evidence is appended by date below. Latest
historical verified Linux delivery at the time of the foundation report was **run
1402** on `4c2cc04ac47d37c6191a9ef02d4bbba716a930dc`. Native check history and later
gate-specific results are kept separate. Results distinguish executed checks from
pending Linux/cluster gates.
## M5.7 local publication - 2026-09-26
Added the single-writer local filesystem adapter and manifest v1 under ADR-0048.
The implementation is in `include/evo/world/local_files.hpp` and
`src/world/local_files.cpp`; `schemas/local-world-manifest.md` defines the durable
bytes. The commit point is replacement of the inactive manifest slot after the
referenced payload has been written, flushed, installed and directory-flushed.
Recovery validates the snapshot and optional journal before returning a new store.
FNV is only a corruption check; this is not host-loss durability or the M7 region
ownership system.
Executed on Windows x64 / MSVC 19.44.35207:
- `tools/windows-check.ps1 -Preset release -Action check`: configure/build passed;
47/47 CTests passed with `NDEBUG`.
- `tools/windows-check.ps1 -Preset analysis -Action check`: clang-tidy 19 and
warnings-as-errors build passed; 47/47 CTests passed.
- `tools/windows-check.ps1 -Preset asan -Action check`: build passed; 47/47 CTests
passed. The Windows preset is not Linux ASan+UBSan evidence.
- `tools/windows-check.ps1 -Preset release -Action format`: passed.
- `git diff --check`: passed before commit.
New coverage validates checkpoint/journal/recovery, exact compaction over a published
journal, divergent-state rejection, exclusive writer lock, injected publication
boundaries, partial-write/out-of-space outcomes, corrupt manifest and missing payload.
Before-manifest failures recover the previous state; after replacement, reopen
recovers the new state. Actual power-cut/disk-full behavior, POSIX syscall path,
Linux CI/libFuzzer, and filesystem-controller flush guarantees remain unverified
locally. There is no automatic orphan cleanup, production storage adapter,
Kubernetes deployment or World0 change.
### M5.7 review corrections
The follow-up review found that command high-water alone allowed a divergent
checkpoint to replace an active journal. Checkpoint now compares its canonical bytes
with the fully recovered journal result before compaction. Poisoned handles reject
recovery; reopen resolves ambiguous post-replacement directory-sync failures.
Interruption tests now use a different edited payload and compare recovered canonical
state and debris across the commit point. Deterministic partial-write/out-of-space
injection covers payload and manifest temporaries, including retry, and a directory
sync failure hook verifies poison/reopen behavior. The injection models a partial
write outcome; it is not a physical disk-full or power-cut test.
Revalidated after these corrections on Windows x64 / MSVC 19.44.35207:
- Release configure/build and 47/47 CTests passed.
- clang-tidy analysis build and 47/47 CTests passed.
- ASan build and 47/47 CTests passed.
- clang-format and `git diff --check` passed.
## Tools and host
- Windows NT 10.0.26200, AMD Ryzen 7 9800X3D, 8 cores / 16 logical processors.
- MSVC 19.44.35217.0, CMake 3.31.6-msvc6, Ninja 1.12.1.
- clang-format and clang-tidy 19.1.5 (Visual Studio LLVM tools).
- kubectl 1.31.4 / Kustomize 5.4.2 for offline manifest rendering.
## Executed native checks
| Command | Observed result |
|---|---|
| `./tools/windows-check.ps1 -Preset debug -Action check` | Configure/build passed; 5/5 CTests passed |
| `./tools/windows-check.ps1 -Preset release -Action check` | Configure/build passed; 5/5 CTests passed under NDEBUG |
| `./tools/windows-check.ps1 -Preset analysis -Action check` | clang-tidy/warnings-as-errors build passed; 5/5 CTests passed |
| `./tools/windows-check.ps1 -Preset asan -Action check` | MSVC AddressSanitizer build passed; 5/5 CTests passed; this is NOT UBSan |
| `./tools/windows-check.ps1 -Preset debug -Action format` | Passed with clang-format 19; repeated twice in one shell after helper fix |
| `./build/release/evo-sim --seed 123456 --entities 8 --ticks 1000000 --verify-replay` | Exit 0; replay_verified=true; hash fab22c2e8ae1c379 |
| `./build/release/evo-benchmark` | Exit 0; MSVC 1944 Release, seed 123456, 10000 entities, 1000 ticks, 0.0210666 seconds, hash 3bad6db1612cbd1 |
| Git Bash `bash -n tools/ci/linux-check.sh` and `bash -n tools/ci/publish-image.sh` | Syntax checks passed; no Linux execution implied |
| `kubectl kustomize deploy/kubernetes/dev` | Rendered Job, ServiceAccount and NetworkPolicy in evolution-dev; final render pins the verified run-1402 image |
CTest covers the independently frozen 1000-tick fingerprint d61ad5bf0c928c76,
seeded replay properties, malformed/semantic snapshot rejection, entity lifecycle,
capacity/overflow, telemetry and CLI file continuation/alias protection. The
expected-failure harness confirms tests remain active in Release.
The benchmark is a single uncontrolled wall-clock observation, not a performance
threshold or an MMO capacity estimate. Its initial compiler output accidentally
used hexadecimal; the stream now returns to decimal and reports MSVC 1944.
## Gitea evidence
The maintainer's [run 1399](https://git.aeternalisgames.com/dingo/AEvolution/actions/runs/1399)
on `8ab48d3a5639121693f5881b600482e870b49f05` completed successfully. API job records
confirm verify (configure/build/CTest) and publish (Docker build/push) succeeded
on runner theraborn-runners-runner-0, label ubuntu-22.04. This is evidence for the
original workflow, not the expanded sanitizer/analysis jobs added in this change.
Exact Linux tool versions and expanded-run results appear below; a runner label
is not a compiler version.
## Limitations and next gates
- Local Docker Desktop has no running Linux engine. Container checks must run on
the configured Gitea runner; no local container success is claimed.
- Expanded Linux GCC14/Clang19, ASan/UBSan, TSan and libFuzzer evidence is recorded
below. TSan currently validates a sequential kernel, not future concurrency.
- Sandbox process restrictions stalled native Ninja and blocked Git Bash signal
pipes. Those attempts were stopped; checks reran successfully outside sandbox.
- Repeated VsDevCmd initialization grew PATH beyond cmd's limit. The helper now
reuses the matching x64 environment and deduplicates PATH; repeated checks pass.
- The maintainer reports Argo setup, but this workspace has no Kubernetes context.
No dev Job completion or cluster admission result is recorded yet.
- No biology, network/load, durable server recovery or backup restore exists.
Those remain M2, M7/M8 and later gates, not failing or passing foundation tests.
Update this report with run URL, tested commit, image digest and Job result when
those checks execute. Never overwrite a golden merely to make CI pass.
## Expanded Linux run 1400
[Run 1400](https://git.aeternalisgames.com/dingo/AEvolution/actions/runs/1400),
commit 9e412736d91ad103b7a951f99b811c629c5ac61e: GCC14 Debug and Release, Clang19
analysis and ASan/UBSan each passed 5/5 tests. Formatting passed. The seeded fuzz
smoke completed 10000 cases with no reported failure. CMake 3.28.3, GCC 14.2.0,
Clang/format/tidy 19.1.1, Linux 6.8.0-110-generic x86-64. Million-tick hash matched
Windows (fab22c2e8ae1c379); benchmark hash matched too (3bad6db1612cbd1), with
0.0674131 seconds observed for 10000 entities * 1000 ticks on the CI host.
TSan compiled but all executions failed before main: its runtime could not call
personality(ADDR_NO_RANDOMIZE) inside the Docker build sandbox. The image job was
skipped, as required. This is not recorded as a TSan pass. The follow-up isolates
execution in a container with the narrowly constrained profile from ADR-0020;
its result must be recorded after execution.
## Successful delivery run 1402
[Run 1402](https://git.aeternalisgames.com/dingo/AEvolution/actions/runs/1402),
commit 4c2cc04ac47d37c6191a9ef02d4bbba716a930dc: every job succeeded. The general
checks reused Docker's valid cache for unchanged checked sources; their full
execution also passed in run 1401. TSan executed all five CTests successfully in
2.71 seconds using the scoped profile and writable tmpfs outputs. Run 1401 had
first exposed a non-root tmpfs permission error; setting explicit mode=1777 on
these isolated scratch mounts fixed it. No test was disabled or ignored.
The runtime image built, ran the restricted replay smoke test and was pushed to:
`registry.aeternalisgames.com/aeternalis/aevolution:sha-4c2cc04ac47d37c6191a9ef02d4bbba716a930dc-run-6-1`
Registry digest:
`sha256:8fce2b462c9907f38d3fb832afeb70d0bff068dd832b2171e861186dbcb8e6b9`.
The dev manifest pins this digest. This proves build/test/image delivery; Argo
application mapping and Kubernetes Job completion remain unverified locally.
## M1.5 native ECS experiment
Opt-in target evo-ecs-experiment (EnTT4 pinned in ADR-0021), based on 1c6e6e1 plus
this experiment's changes. MSVC Release contract checks passed, including 72
seed/density/count combinations, hand-calculated arithmetic, canonical equality,
restore continuation, retired-ID rejection, malformed snapshot checks and
allocator balance. The final clang-tidy build/test passed after tightening the
process-boundary diagnostic catch. MSVC ASan passed on the final source.
Baseline Release CTest remained 5/5
and format-check included the experiment. No simulation rule/golden changed.
Commands: configure release/analysis/asan with `-B build/ecs[-analysis|-asan]
-DEVO_BUILD_ECS_EXPERIMENT=ON`, build target evo-ecs-experiment, then
`ctest --test-dir <directory> -R ecs-experiment --output-on-failure`.
Seven measured samples plus warmup are retained in results/ecs-msvc1944.jsonl;
protocol, build observations and scope limits are in ECS_EVALUATION.md. The
Linux experiment CI job is pending its first pushed run; no Linux experiment
success is claimed yet.
## M1.5 Linux evidence
[Run 1404](https://git.aeternalisgames.com/dingo/AEvolution/actions/runs/1404),
source bf7d60b7cd174736d38aea42616064e55e0c5f05: every job passed. The new ci-ecs
stage explicitly prepared the pinned header/license, then built and ran the
experiment under GCC14 Release, Clang19 ASan/UBSan and Clang19 analysis. All three
contract invocations passed. Raw measured GCC samples are retained at
results/ecs-gcc142.jsonl; all 56 fingerprints match corresponding Windows samples.
No production rule or snapshot-format change occurred. This supersedes the
pending Linux experiment note above. The final follow-up edits only record
results and decision scope; production ECS adoption remains deferred.
## M1.6 sequential scheduler (2026-09-19)
Native commands: `tools/windows-check.ps1 -Preset release -Action configure`,
clang-format on changed C++ files, `cmake --build --preset release`,
`ctest --preset release`; then `tools/windows-check.ps1 -Preset analysis -Action check`,
`tools/windows-check.ps1 -Preset asan -Action check` and Release `-Action format`.
All three configurations passed 5/5 CTests; clang-tidy and format-check passed.
Tests execute under NDEBUG and cover six batch permutations across 32 seeds,
an independent direct-lifecycle reference, ordered receipts, restore continuation,
rejected conflicts and malformed admission, the 1,024-command boundary, full entity
capacity and tick/ID exhaustion. Golden `d61ad5bf0c928c76` remains unchanged.
The copy-and-commit implementation allocates before publishing world state and
statically checks noexcept publication/receipt moves. Allocation-failure injection
was not run. No scheduler throughput claim, concurrency validation, durable input
journal or Kubernetes execution is implied.
[Gitea run 1407](https://git.aeternalisgames.com/dingo/AEvolution/actions/runs/1407)
passed all four jobs for source `b620edfd6f8097172acf055402a24620cb078358`:
Linux checks, the ECS experiment, separate TSan, and image smoke/publication.
This closes the pending Linux validation for M1.6. The maintainer's independent
dev image pin from `20e609d` was preserved during rebase. Kustomize renders the
three expected dev resources successfully; no cluster execution is claimed.
## M1.7 spatial queries (2026-09-19)
Release and MSVC ASan passed 5/5 CTests. Tests cover periodic-image reference
comparison (32 seeds × 3 lifecycle/restore phases × 24 queries), hand-checked
seams/corners/inclusive radii, empty/full capacity, invalid centers and maximum
u32 radius. Queries leave snapshots unchanged; existing v1 goldens remain intact.
Native commands use `tools/windows-check.ps1 -Preset <release|analysis|asan>
-Action check` and Release `-Action format`. Initial analysis flagged an unnecessary
test-helper vector copy; the helper now takes a const reference. Final analysis,
Release, ASan and formatting passed after applying the helper's required line
wrapping. Linux validation is recorded below.
`build/release/evo-spatial-benchmark.exe` produced 72 measured records plus
metadata in results/spatial-msvc1944.jsonl. One warmup and three samples per
scenario; 1k/10k/100k entities, 128 queries, uniform/clustered positions, four radii.
Timing includes result allocation, destruction and checksum work, excludes setup;
no performance threshold. Desktop CPU load was not isolated. No cluster changes
are required for this library increment.
[Gitea run 1409](https://git.aeternalisgames.com/dingo/AEvolution/actions/runs/1409)
passed all four jobs for source `9b19509673d88683ee72e093e42534bfe151f928`:
Linux checks, ECS regression, separate TSan and image smoke/publication.
All 72 spatial benchmark checksums and match counts agree with Windows.
Raw Linux measurements are in results/spatial-gcc142.jsonl; 24 native scenarios
also have identical checksums across their three repetitions. No deployment
manifest, rules version, snapshot format or golden was changed.
## M2.1 finite inventory ledger (2026-09-19)
Native Release, analysis and MSVC ASan each passed 6/6 CTests; formatting passed.
Commands: `tools/windows-check.ps1 -Preset <release|analysis|asan> -Action check`
and Release `-Action format`. Initial clang-tidy rejected the Element enum's
oversized base type; changed to u8 and reran all configurations successfully.
The new independent ledger test checks typed energy/mass separation, finite-source
exhaustion, explicit heat transfer, intermediate overdraft, whole-transaction
rollback, retry/duplicate action semantics, invalid reservoir IDs, maximum batch
and reservoir counts, opening-total overflow and exact u64-limit transfers.
32 seeded runs × 31,250 transactions = one million transactions, each checked
against independent reference balances, a replay branch and recomputed per-channel
totals. This is transfer replay, not serialized recovery or an ecosystem run.
Final ledger tests took 0.63s Release and 7.62s ASan locally; not controlled
performance measurements. Existing particle goldens remain unchanged.
Allocation-failure injection, durable ledger recovery and cross-ledger ownership
are not tested or implemented. Rates and conversions are absent, so rounding
residuals cannot occur in this transfer-only scope.
[Gitea run 1411](https://git.aeternalisgames.com/dingo/AEvolution/actions/runs/1411),
source `8fd3591b0e37aaecf00a6410a5926745a031cb6f`, passed
the Linux checks, ECS regression and separate TSan jobs. Each main Linux
configuration passed 6/6 CTests. Ledger-suite times were 3.34s GCC Debug, 0.37s GCC
Release, 2.46s Clang analysis and 8.78s Clang ASan/UBSan; these are CI test
durations, not controlled performance benchmarks. The image smoke/publication
job also passed, completing all four jobs. No deployment image pin was changed.
## M2.2 finite producer process (2026-09-19)
Native Release, clang-tidy and MSVC ASan passed 7/7 CTests; format-check passed.
Commands: `tools/windows-check.ps1 -Preset <release|analysis|asan> -Action check`
and Release `-Action format`. The producer suite compares hand-calculated ticks
and 32 seeds × 1,000 ticks against a small-domain cumulative arithmetic reference,
an independent replay branch and per-tick energy/element conservation audits.
It covers invalid config/ticks with unchanged state, fractional capture/growth,
efficiency endpoints, finite source exhaustion, all four limiting elements,
zero capture, maximum u64 capture with 50% carried conversion, and maximum recipe
and growth-cost multiplication. Particle goldens remain unchanged.
Allocation-failure injection and tick-counter exhaustion were not exercised;
the latter has an explicit guard but no arbitrary-state restore API. Tests use
in-memory copies, not serialized recovery. No photosynthetic chemistry, metabolic
model, database integration, throughput guarantee or deployment is claimed.
[Gitea run 1413](https://git.aeternalisgames.com/dingo/AEvolution/actions/runs/1413)
passed all four jobs for source `af293b8b50cbb506a49dfff10c6cc92307e16aca`:
Linux checks, ECS regression, separate TSan and runtime image smoke/publication.
The initial push was delayed by a tool approval-review usage limit; retry succeeded
after the maintainer resumed work. No deployment image pin was changed.
## M2.3 shared feeding community (2026-09-20)
Native Release, clang-tidy and MSVC ASan each passed 8/8 CTests; formatting passed.
Commands: `tools/windows-check.ps1 -Preset <release|analysis|asan> -Action check`
and Release `-Action format`. The new suite checks hand-calculated capture/feeding,
producer-to-consumer and consumer-to-consumer predation, heat/waste outputs,
terminal actor rejection, full-tick rollback, competing resource requests,
duplicate sequence/consumer admission, per-channel intake caps, invalid actors,
config/capacity limits and u64 assimilation endpoints. 32 seeds × 200 ticks compare
reordered input batches, receipts, full balances and producer progress, with
independently recomputed energy and elemental totals. Standalone producer and
existing particle golden tests remain unchanged and passing.
No serialized recovery, allocation-failure injection, spatial reach/perception,
physical injury, corpse handling or persistent-world integration was tested or
implemented. These are co-located synthetic process fixtures.
[Gitea run 1415](https://git.aeternalisgames.com/dingo/AEvolution/actions/runs/1415)
passed all four jobs for source `b7dd466f57f385d6a8ffccfaa69f639077545587`:
Linux checks, ECS regression, separate TSan and image smoke/publication.
No deployment image pin was changed.
## M2.4 terminal inventory and decomposition (2026-09-20)
Native Release, analysis and MSVC ASan passed 8/8 CTests; formatting passed.
Commands: `tools/windows-check.ps1 -Preset <release|analysis|asan> -Action check`
and Release `-Action format`. Tests preserve corpse inventories on death, suppress
producer activity immediately, reject repeated/invalid terminal events and roll
back death with later invalid feeding. Sixteen seeds × 256 ticks compare a
closed-form depletion reference, reversed terminal-event order, per-tick budgets,
inventories and conservation. Additional cases cover shared corpse/waste limits,
mixed and zero channel caps, u64-maximum inventories, diagnostic rollback, and
returned nutrients funding producer structure on the next tick.
The existing feeding, standalone producer, ledger and particle tests pass unchanged.
This does not validate reaction chemistry, actual death causes, corpse scavenging,
serialized recovery, allocation-failure injection or persistent history. There is
no deployment change.
[Gitea run 1417](https://git.aeternalisgames.com/dingo/AEvolution/actions/runs/1417)
passed all four jobs for source `e17a9a7279d6bde736261c0c33abe46108e1f622`:
Linux checks, ECS regression, separate TSan and image smoke/publication.
## M2.5 funded reproduction (2026-09-21)
Native Release, analysis and MSVC ASan passed 8/8 CTests; formatting passed.
Commands: `tools/windows-check.ps1 -Preset <release|analysis|asan> -Action check`
and debug `-Action format`. The first format run rejected three wrapped test
declarations; clang-format 19.1.5 rewrote them and Release was rebuilt and rerun.
Tests cover hand-calculated endowment and construction overhead, the endowment
floor, per-channel insufficiency with abundant energy, zero and oversized
investments, reserved slots rejected as feeding targets, corpses and ancestry
subjects, the producer rejected as a parent, terminated and unknown parents,
zero/duplicate sequences, two intents naming one parent, oversized batches,
capacity exhaustion rejecting the whole tick, absent identity reuse after a child
dies and decomposes, ancestry surviving both deaths, and construction endpoints of
0 and 1000 including a paid-for zero-endowment child. 32 seeds × 4 ticks × 2 births
compare reversed intent order against balances, producer progress, birth receipts
and independently recomputed conservation. A u64-maximum investment endows exactly
and conserves. Default zero capacity leaves the M2.4 reservoir layout, feeding,
producer, ledger and particle tests unchanged; golden `d61ad5bf0c928c76` is intact.
Review of `c84a701` found one defect (2026-09-22): births validated each parent
against the progressively modified candidate, so a child allocated earlier in the
same batch was already active and could fund a second birth. Two generations could
therefore appear in one tick, which ADR-0028 forbids. Eligibility is now fixed on
entry to the reproduction phase, after terminal facts and feeding, and checked for
every intent before any birth is applied. The contract itself was already correct
and did not change. The regression test admits one intent whose parent is created
by the preceding intent and requires `unborn_actor` with the entire tick unchanged;
it was confirmed to fail against the unguarded implementation and to pass with the
guard. The same pair succeeds across two ticks, recording the grandchild ancestry.
All four native configurations were rerun after the fix.
Allocation-failure injection, serialized recovery of states/parents/origins and
long-run ecosystem behavior were not exercised. This fixture defines no
reproductive strategy, fecundity, mutation, gestation, spatial placement or account
ownership, and no deployment manifest or image pin changed. Linux validation is
pending its first pushed run; no Gitea result is claimed for this increment yet.
## M2.6 measured ecosystem episodes (2026-09-22)
Native Release, analysis and MSVC ASan each passed 11/11 CTests; formatting passed.
Commands: `tools/windows-check.ps1 -Preset <release|analysis|asan> -Action check`
and debug `-Action format`. Formatting rejected 25 wrapped declarations on the
first run; clang-format 19.1.5 rewrote them and Release was rebuilt and rerun.
Three CTest cases are new: `ecosystem-contracts`, `cli-ecosystem` pinned to
fingerprint `2cd4a15f9fbe9949`, and `cli-ecosystem-conflict`, which requires the
CLI to reject `--ecosystem` combined with proof-run options.
The suite checks a solitary founder that only ages, a hand-checked predation
retirement, births exhausting exactly the reserved slots, tick-limit termination
with survivors, configuration rejection, and two episodes interleaved without
observing each other. Across 64 seeds it steps tick by tick, requiring that the
driver never provokes a rejected batch and that conserved totals are identical at
every tick, then replays each seed and requires an identical report. That ensemble
advanced 22,804 ticks with 879 births, 730 predation deaths, 277 lifespan deaths
and extinction in all 64 episodes, episode length 73 to 463 ticks.
`build/release/evo-sim.exe --ecosystem --seed 0 --episodes 3000` advanced
**1,093,007 ticks** in 1.34 s, with 41,138 births, 33,984 predation deaths, 13,154
lifespan deaths, extinction in all 3,000 episodes, episode length 50 to 497 ticks
and fingerprint `130d7725c0dc8b16`. Deaths equal founders plus births exactly
(6,000 + 41,138 = 47,138), so every process died once and no producer was grazed
out: the producer always retains captured energy, so its whole inventory never
reaches zero. The elapsed time is one uncontrolled wall-clock observation, not a
performance threshold.
Review of `eaf58f2` found one defect (2026-09-22): the CLI validated mode options
in one direction only. `--ticks` was accepted and ignored under `--ecosystem`, and
`--episodes`/`--tick-limit` were accepted and ignored without it, so a measurement
script could exit zero while running the other workload. The parser now records
which options were supplied and rejects any that do not belong to the selected
mode, in both directions. Six CLI cases cover this; `cli-ecosystem-rejects-ticks`
and `cli-episodes-require-ecosystem` are the two genuine regressions, while the
entities, save and verify-replay cases were already rejected and add coverage.
The remaining shared options were audited: `--seed` and `--output` are valid in
both modes, and `--load` conflicts were already enforced. CTest is now 16 cases,
all four native configurations were rerun, and `2cd4a15f9fbe9949` is unchanged.
This measures arithmetic conservation and turnover, not ecological plausibility.
Strategy constants, lifespans and the targeting rule are chosen inputs. There is
no maintenance cost, starvation rule, perception, spatial gate, genome or
heritable variation, so lineages differ only by founder strategy. Episode state is
not serialized, allocation failure is not injected, and no population abstraction
exists. Golden `d61ad5bf0c928c76` and every earlier suite are unchanged. No
deployment manifest or image pin changed. Linux validation is pending its first
pushed run; no Gitea result or cross-platform fingerprint agreement is claimed yet.
## M3.1 immutable genome record (2026-09-22)
Native Release, analysis and MSVC ASan each passed 17/17 CTests; formatting
passed. Commands: `tools/windows-check.ps1 -Preset <release|analysis|asan>
-Action check` and debug `-Action format`. Formatting rejected three wrapped
declarations on the first run; clang-format 19.1.5 rewrote them and Release was
rebuilt and rerun. `genome-contracts` is the one new CTest case, covering a new
`evo::biology` library that nothing else links.
Tests cover founder and descendant records, unordered drafts producing byte-equal
canonical records, digests differing for identical capabilities from different
origins, and every ancestry rule: zero identity, zero origin, zero development
version, a zero parent, a self parent, two parents and a repeated parent. Schema
rules cover an empty record, each of the three required capabilities removed in
turn, a duplicate kind, an unknown kind, an unknown prerequisite bit, and both
inclusive boundaries of all seven declared ranges accepted with one below and one
above each rejected. Graph rules cover a declared prerequisite naming an absent
capability, self, mutual and three-node cycles, a valid ordering using the same
edges without the closing one, and a cycle closed by a mandated rather than a
declared edge.
Encoding tests cover the round trip, byte-exact re-encoding, truncation at every
length from zero to one byte short, one trailing byte, an oversized buffer, a
corrupted magic, and every single-byte corruption of the full record. Three cases
reseal the checksum after editing so the later checks are reached rather than
masked: a rewritten genome ID is caught by the digest, a future schema version
fails closed, and non-canonical record order is caught as a duplicate kind. A
parent count beyond the reserved room and a capability count beyond the schema
are both refused before the corresponding region is read.
Coverage limitation: in schema 1 every mandated capability is also a required
one, so the presence half of the mandate check is reachable only through declared
prerequisites, which is tested. The mandated edges do take part in the cycle
closure, which is tested. A later schema that mandates a capability without
requiring it would exercise the remaining branch; this is recorded rather than
covered by a test that could not fail.
Review of `8584eb1` found two defects (2026-09-22). First, `Genome` had implicit
move operations, so moving one emptied the source's capabilities while leaving
its identity, origin and digest intact: a record that still reported ID 7 but
whose encoded bytes no longer decoded. That contradicted the documented invariant
directly. The type is now copy-only, which suppresses the implicit moves so a
move decays to a copy, and the invariant holds for every operation. The new move
tests were confirmed to fail without the fix: the Debug and ASan builds exit 1 at
runtime, and the Release build does not compile at all, because MSVC proves the
moved-from comparison always throws and rejects the unreachable code under
warnings-as-errors. clang-tidy then flagged the deliberate decaying moves in the
tests as `performance-move-const-arg`; those two lines carry a NOLINT explaining
that the decay is the contract under test.
Second, the encoding tests compared the current encoder against the current
decoder, so a coordinated change to layout or digest would have passed while
breaking existing v1 records. A frozen 83-byte fixture now holds one schema 1
founder record. Every structural byte was verified by hand against
schemas/genome-v1.md before being frozen: magic, both versions, ID 7, origin 1,
zero parents, three records, kinds 0/1/5 with parameters 400/120/5000, and the
`56 + 9N` length. The digest `9b2cdedef16ac87f` and the trailing checksum are
computed values frozen alongside them. The fixture is asserted in both
directions: the encoder must still produce those bytes, and the decoder must
still read them into the expected record. It is a golden and must not be edited
to make a test pass.
No genome affects any simulation, and nothing links the new library except its
own tests. The M2 process fixtures still duplicate founder strategies and do not
read the record. Development, phenotype derivation, mutation, recombination,
multi-parent heredity, migration between schema versions and any persistent
genome archive are not implemented. The encoding is a local record format, not a
network protocol or durable storage, and neither the digest nor the checksum is
an authenticity claim. Golden `d61ad5bf0c928c76` and the ecosystem fingerprint
`2cd4a15f9fbe9949` are unchanged. No deployment manifest or image pin changed.
Linux validation is pending its first pushed run; no Gitea result is claimed yet.
## M3.2 development and phenotype (2026-09-22)
Native Release, analysis and MSVC ASan each passed 18/18 CTests; formatting
passed. Commands: `tools/windows-check.ps1 -Preset <release|analysis|asan>
-Action check` and debug `-Action format`. `development-contracts` is the one new
CTest case, in the existing `evo::biology` library, which still nothing else
links. No existing suite changed.
The fixture genome uses a target size of 8,000,000 fg, which is an exact 20
micrometre cube at the declared density, so every derived value is hand-calculable
and is asserted as a literal: structural mass 8,000,000, extent 20, surface 2,400,
membrane mass 28,800,000, total 36,800,000 and maintenance 160,000 picojoules per
tick. Half-built, damaged and maximum-size cases are hand-calculated the same way,
including the 1e9 fg body that is an exact 100 micrometre cube with an uptake rate
of exactly the reference surface, and the 999,999,999 fg body one cubic micrometre
short of it, whose extent is 99.
Determinism is checked over 45 growth and temperature combinations, each derived
twice and compared, with more growth never yielding less structure. Damage is
checked to change the phenotype while leaving the genome's encoded bytes and
content digest identical, which is the M3.2 risk stated in the roadmap. Capability
prerequisites are checked both ways: without the capability a component is zero at
any size, and with it uptake scales with surface while movement speed does not.
Thermal shortfall is checked inside tolerance, symmetrically outside it, and
without the capability, and is confirmed to change no other derived value.
Unsupported rule versions and out-of-range growth fail closed through both entry
points.
Plans are checked for the full funded step, for the exact largest affordable step
one femtogram short of it, for no growth with no material, for a finished organism,
for clamping the last step to the target rather than overshooting, and for never
repairing damage. The declared flooring is pinned rather than hidden: a body whose
mass floors to zero advances its growth counter for no material, which grants no
structure because none exists at that size either way.
Coverage limitation: the overflow guards are unreachable through the public API at
schema 1 ranges, because the largest admissible genome produces a total mass of
about 7.2e11 femtograms, eleven orders of magnitude below the u64 limit. They are
kept as guards for later schema ranges and are recorded here rather than covered
by a test that could not fail.
No organism develops, nothing links the library except its own tests, and no
inventory moves: a plan reserves nothing and pays for nothing. The transfer half
of ApplyDevelopment, energetic budgets, repair, ageing, starvation, mutation and
morphology are not implemented. Golden `d61ad5bf0c928c76`, ecosystem fingerprint
`2cd4a15f9fbe9949` and the frozen genome fixture are unchanged. No deployment
manifest or image pin changed. Linux validation is pending its first pushed run.
## M3.3 metabolic budgets (2026-09-23)
Native Release, analysis and MSVC ASan each passed 19/19 CTests; formatting
passed. Commands: `tools/windows-check.ps1 -Preset <release|analysis|asan>
-Action check` and debug `-Action format`. `metabolism-contracts` is the one new
CTest case. `evo_biology` now links `evo::accounting` and `evo_processes` links
`evo::biology`; no existing suite changed and no existing fixture reads genomes.
Every budget is asserted as a hand-calculated literal against the same 20
micrometre cube fixture as M3.2: body 24,160,000 / 2,640,000 / 3,280,000 /
6,720,000 femtograms across C/N/P/other, maintenance 160,000 and movement 184,000
picojoules per tick, intake 240,000 picojoules and 2400/720/240/1440 femtograms,
and a reproduction investment of 11,040,000 femtograms priced at 441,600
picojoules. Construction is checked to compose each compartment separately rather
than by one blended ratio, and to agree exactly with what `plan_development`
reports as material and energy.
The non-uniformity the increment exists to demonstrate is tested directly.
Doubling membrane thickness raises body matter to 65,600,000 femtograms,
maintenance to 304,000 picojoules and the movement cost, shifts the carbon share
upward because membrane is the carbon-rich compartment, and leaves intake limit,
speed, assimilation and structural mass identical; the mass and maintenance ratios
are asserted to differ, so no single multiplier could produce this. Growth is
tested the same way: uptake follows surface while maintenance follows mass, and
their ratios differ.
Mass splits are checked to conserve at four awkward target sizes and four growth
fractions, where every share floors and the last channel absorbs the remainder.
An absent capability is checked to yield a zero budget rather than a default, and
unknown rule versions and out-of-range growth fail closed.
The genome-to-ledger path is exercised end to end: a derived `ConsumerConfig`
drives a real `FeedingCommunity` tick, the ledger totals are identical before and
after, and a bite one femtogram past the derived cap is refused by the existing
M2.3 contract. The opening inventory is checked to be exactly the caller's.
Review of `be71d4f` found one defect (2026-09-23): construction composed the mass
difference rather than differencing the two composed bodies, which discards the
rounding the existing body already carries. At target mass 1001 growing from 10 to
60, it funded C25 N7 P2 other16: the right total, but nitrogen and phosphorus one
short each and other two over, so spending the budget could not produce the body
the same rules derive. Construction is now the per-channel difference between the
composed bodies, which funds C25 N8 P3 other14 for that case.
Because the remainder rule is not monotone per channel, that difference can be
negative on a channel, so construction became a transition with an intake and a
release rather than a single requirement: a body growing from 13 to 14 femtograms
acquires one carbon and one nitrogen while freeing one from other. The release is
a rounding artifact bounded by a few femtograms, not excretion.
A follow-up review found the affordability half of the same problem: a transition
could require more of an element than was available, because the check compared
net mass. A step growing a body from 13 to 14 femtograms nets one femtogram but
acquires a carbon and a nitrogen while freeing one of other, and released other
pays for neither. The environment now states availability per ledger channel
rather than as one pool, since a pool cannot say that carbon is plentiful while
nitrogen is not. `plan_development` keeps a total-mass bound, sound because net
never exceeds gross but not an authority, and `metabolic_budget` narrows it to the
largest step whose own intake fits channel by channel and reports the growth that
step funds. That revises ADR-0031 and is recorded there.
The affordability tests place a single femtogram in each channel in turn and
require that no step needing a different element is funded, that nothing is
acquired beyond what is present, and that the mass-only plan would have admitted
the step; a carbon and a nitrogen together then fund it exactly. Across six target
sizes by seven growth fractions by four stock levels, every reported intake is
checked to fit the availability channel by channel and never to exceed the step
the plan proposed. These tests were confirmed to fail with the per-channel check
removed.
A third review found the availability bound itself could wrap: summing four
unrestricted u64 channels with 2^62 in each reaches exactly 2^64, so the total
read as zero and both planning and budgeting refused growth that a billion
femtograms per channel funded. The sum now saturates, which is also the right
answer since no step can need more than u64. Regressions in both suites use that
exact 2^62 case and a single saturated channel, and require the same step a
billion per channel funds; both were confirmed to fail without saturation. The
equivalent helper in `metabolism.cpp` is applied only to inventories derived from
bounded phenotype masses, never to caller input, so it is left unsaturated rather
than given a branch no test could reach.
The invariant `held + intake - release == reached` is now asserted per channel for
the reviewer's exact case, for the releasing case, across six target sizes by
seven growth fractions by four material budgets, and across a forty-step walk from
nothing to full growth where the accumulated intake and release are checked
against the whole history. The net of every transition is also asserted to equal
what `plan_development` charges material for. These tests were confirmed to fail
against the previous implementation.
Coverage limitation: as in M3.2, the overflow guards are unreachable through the
public API at schema 1 ranges and are recorded rather than covered by a test that
could not fail. The small integer helpers are duplicated between `development.cpp`
and `metabolism.cpp` rather than shared through an internal header.
No organism spends any budget: nothing consumes maintenance, movement or
construction, and no starvation, death or repair rule exists. The constants are
chosen for legible magnitudes, not balance. Golden `d61ad5bf0c928c76`, ecosystem
fingerprint `2cd4a15f9fbe9949` and the frozen genome fixture are unchanged. No
deployment manifest or image pin changed. Linux validation is pending its first
pushed run; no Gitea result is claimed for any increment since `b01a824`.
## M3.4 seeded mutation candidates (2026-09-23)
Native Release, analysis and MSVC ASan each passed 20/20 CTests; formatting
passed. Commands: `tools/windows-check.ps1 -Preset <release|analysis|asan>
-Action check` and debug `-Action format`. `mutation-contracts` is the one new
CTest case, in the existing `evo::biology` library. No existing suite changed.
The eligible set for the founder fixture is pinned by hand rather than observed:
nine operators, derived from the schema by the rules that make them eligible.
Energy conversion and membrane can adjust either way, movement, ingestion,
reproduction and thermal tolerance can be acquired, target size can adjust up but
not down because its 9,999,999 step would fall below the minimum from 8,000,000,
and no required capability can be shed. Over 2,000 seeds the offers are exactly
that set and nothing else.
Uniformity is checked to within a fifth of each operator's exact share, about six
standard deviations for 8,000 offers across nine operators. That test was
confirmed to fail when the draw was replaced by taking the first eligible
operators, which is the shape a biased offering would take. The rejection
sampling that removes modulo bias is correct by construction rather than
test-verified: over nine bins with 64-bit draws the bias it removes is about one
part in 10^18 and no test at this scale could observe it. That is recorded rather
than claimed as covered.
Stability is checked both ways: the same context always returns the same
candidates, while a new event index, seed, stream or growth state returns a
different draw. Selection is checked to refuse an operator that was never offered,
a moved event index, a changed growth state, a tampered offer and a different
parent genome, each with its own error. Boundary operators are checked to be
absent rather than clamped at both ends of a range, and a capability another one
declares as a prerequisite is checked to be unsheddable while the dependent
capability itself remains sheddable. Every offered candidate is selected in turn
and its successor checked to validate, to record parent 7, and to survive a
decode; the reported maintenance and mass are checked to equal what
`metabolic_budget` charges that successor.
Review of `5497f4d` found one defect (2026-09-23): selection established a
stale context by comparing regenerated candidate lists, which cannot establish it.
Four of nine operators is only 126 distinct lists, so different contexts draw the
same list routinely, and an offer from event 1 was accepted under event 239. An
offer now records the parent's identity and digest and the context itself, and
selection compares those directly before anything else. Because they then
determine the draw, a list that still differs means the offer was altered, which
is now its own error rather than a context having moved.
The regression searches for a context whose draw collides with the offer's,
asserts that such a collision exists rather than assuming it, and requires that
selection still rejects it. It was confirmed to fail against the previous
implementation. Rewriting the offer's recorded context to name the colliding one
is checked to be accepted, and should be: those candidates genuinely are what that
context draws, so the result is a legitimate offer for it rather than a forgery.
Editing a candidate is checked to be caught as an altered offer.
No organism mutates and no fixture calls any of this. Candidate count, context
window, mutation cost as a game rule, inherited versus acquired expression, NPC
mutation and recombination remain open and are not settled here. Golden
`d61ad5bf0c928c76`, ecosystem fingerprint `2cd4a15f9fbe9949` and the frozen genome
fixture are unchanged. No deployment manifest or image pin changed. Linux
validation is pending; no Gitea result is claimed since `b01a824`.
## M3.5 lineage retention (2026-09-23)
Native Release, analysis and MSVC ASan each passed 21/21 CTests; formatting
passed. Commands: `tools/windows-check.ps1 -Preset <release|asan|analysis>
-Action check` and debug `-Action format`. `lineage-contracts` is the one new
CTest case, in the existing `evo::biology` library. No existing suite changed.
Retention is asserted as a property of the type, not only of its behaviour: four
concepts check that the archive offers no `erase`, `remove`, `clear` or `prune`,
with a std::map positive control so the checks are known to be live rather than
vacuously true. An earlier attempt used bare requires-expressions, which MSVC
diagnoses eagerly for a non-dependent type, and a std::vector control, which was
wrong because `vector::erase` takes an iterator rather than a key; the concept
correctly reported false and the control was corrected rather than the assertion
weakened.
A branching fixture of five genomes over three generations is used throughout.
Tests cover a genome outliving both organisms that carried it with its record,
ancestry, children and organism terminal ticks intact; ancestry and descendants
resolving through a genome whose only carrier died; common ancestors for
siblings, cousins, a genome against itself and two separate founders; refusal of
a duplicate identity, a child admitted before its parent and a full archive;
death being terminal against repeat, reversal and rebirth; and absence being
reported for every lineage that has no living carrier, including one that gains a
carrier again afterwards. The capacity case admits the full 4,096 genomes as one
chain, checks that the refusal leaves every earlier record intact, and walks the
4,096-deep ancestry.
The implementation was changed twice under clang-tidy rather than suppressed.
`std::map` members made the archive's move potentially throwing, because MSVC's
map move constructor allocates a sentinel node; sorted vectors replaced them,
which also matches the rest of the codebase. A wrapper struct holding a `Genome`
by value then had the same problem, because ADR-0030 made `Genome` copy-only, so
genomes are now stored directly and identified by their own `id()`. The redundant
per-genome carrier tally was dropped in the same change: carriers are counted from
the organism records, so there is no second tally that could drift.
Review of `8020347` found one defect (2026-09-23): the retention claim covered
methods named like removal and nothing else, so `archive = LineageArchive{}`
discarded every record, and assigning a copy taken before a death restored that
organism's living state. Both forms of assignment are now deleted, which also
makes the archive unswappable, and the copy constructor is declared so the
implicit moves are suppressed and a move decays to a copy. Copying stays available
because a clone removes nothing and GENOME_MODEL expects cloned-world rehearsals.
The interface tests now cover assignment, replacement from a temporary and
swapping as well as the named methods, each with a std::vector positive control,
and check at runtime that a copy taken before a death does not disturb the
original's terminal record and that a moved-from archive still holds everything.
All four static assertions were confirmed to fail with the deleted assignments
removed.
Linux CI on `29ce255` reported one build blocker (2026-09-23), the first Linux
result recorded since `b01a824`. GCC Release passed 21/21, then the Clang 19
analysis build failed: a bare `{1, 2, 3, 4, 5}` deduces `initializer_list<int>`
and the sign-changing conversion to each `std::uint64_t` element trips
`-Wconversion` under `-Werror`. The literals are now unsigned.
MSVC `/W4 /WX` does not diagnose this, and GCC's `-Wconversion` does not warn when
a constant value is known to fit, so neither the local checks nor the GCC job
could have caught it; only Clang did. Attempting to reproduce Clang's diagnostic
locally through clang-tidy failed, because the warning flags are not honoured in
its cl driver mode and no clang driver is installed on this host. The whole tree
was instead audited by hand for the pattern: every other range-for over a braced
literal list either deduces through `auto` or already carries suffixes matching
its declared type, and this was the only instance. That audit is a one-off, not a
check that will catch the next one; the standing gap is that no Clang or GCC
compile happens on this host before a push.
Nothing is serialized, so no history survives a process, and no fixture calls any
of this. Terminal extinction is not claimed anywhere, durable retention and
migration are M7 and M8, and NPC mutation remains undecided. Golden
`d61ad5bf0c928c76`, ecosystem fingerprint `2cd4a15f9fbe9949` and the frozen genome
fixture are unchanged. No deployment manifest or image pin changed. Linux
validation is pending; no Gitea result is claimed since `b01a824`.
## M3.6 genome ecology ensembles
Executed on this Windows host (2026-09-23): Release configure, build and `ctest`
33/33; the analysis preset (clang-tidy 19, warnings as errors) built clean; the
ASan preset configured, built and ran 33/33; `clang-format` 19.1.5 reported no
violations after the new files were formatted. All four checks ran and passed on
the committed tree. No Clang or GCC compile happens on this host, so the standing
gap recorded for `29ce255` is unchanged and Linux CI remains the first Clang and
GCC result for this increment.
`evo-ensemble-tests` covers eight properties: that a derived strategy carries
exactly ADR-0032's budget quantities and invents no margin, that divergence is
reproducible and refuses its bounds rather than truncating, that conservation
holds at every tick of every cell, that a run is reproducible and a different seed
is not, that the archive ends holding exactly the founders and every organism that
lived with no living carriers, that the two seed halves agree within the stated
band, that each gradient axis alone changes the outcome, and that a thermally
tolerant strategy keeps more scheduled life at the extremes. `evo-sim --ensemble`
is pinned by fingerprint `c2b96c86101be628`, confirmed stable across three runs
and distinct from a different seed's `8f88e72cf2715eac`.
Three defects were found and fixed during implementation, all by measurement
rather than by review. `EnsembleReport` and `CellReport` were default-initialized
rather than value-initialized: both are aggregates whose vector members give them
a non-trivial default constructor, which leaves the scalar counters indeterminate,
and the episode total read as garbage. Both are now `{}`.
The other two are findings about the model, and are recorded in ADR-0035 rather
than tuned away. The header claimed the resource level entered both the derived
strategies and the episode's nutrient pool. It does not enter the derivation at
all: founders are derived fully grown, so their construction transition is empty
and ADR-0031's per-channel `available` cannot reach intake, assimilation or
reproduction investment. The comment was corrected rather than the code, because
the code was right and the claim was false.
The resource axis was then measured to be decorative at the original constants and
non-monotonic once it was not. A ten-point sweep at the original fixture returned
41 to 45 organisms across three orders of magnitude of nutrients with identical
tick totals and eight extinctions everywhere: every birth came from the opening
endowment, because a body of 353,600,000 fg needed 73,776,000 fg of carbon to
reproduce while feeding supplied 8,000 fg per tick before assimilation, or roughly
23,000 ticks against a 60-tick lifespan. Rescaling so a birth can be earned
required pinning the investment fraction to the genome schema's floor of one per
thousand, because intake scales with surface and investment with body mass, so
the ticks needed to fund a birth fall only as the cube root of mass. That a
reproduction is affordable only at the schema floor is a finding about the
ADR-0031/0032 cost model and is now the second priority in STATUS.md.
After rescaling, a sweep of 200 to 200,000 femtograms gave 45, 45, 45, 51, 61, 81,
44, 44, 42 and 42 organisms: a peak near 8,000 and a fall away on both sides. The
mechanism is ADR-0026's feeding rule that a target may be bitten by only one
consumer per tick, which caps community income at one producer bite however large
the pool; scarcity empties the producer, consumers bite each other, and that
concentrates mass into individuals that reach the threshold, while abundance
spreads the same income too thin to fund a birth. The gradient test therefore
asserts only that an axis moves the outcome, never a direction. An earlier draft
of that test asserted `rich_organisms > lean_organisms`; that assertion held at
the two chosen cells but implied a monotonicity the sweep disproves, so it was
replaced rather than satisfied by picking two cells that flattered it.
Coverage limits. The grid is four cells and the tests use eight seeds per cell, so
every number describes this fixture and nothing wider. The split-half band
(`high * 2 <= low * 3 + 2`) is a chosen tolerance, not a derived one. No organism
mutates during an episode, so nothing here exercises in-episode divergence and NPC
mutation stays open. The thermal penalty lives in the harness and no contract
knows about it, so no contract test covers it. Nothing is serialized.
The eleven new CLI tests are new-behaviour tests for a new mode, not regressions
of a fixed defect; only the M2.6 bidirectional mode check they extend was ever a
regression. They were confirmed to fail the run, which is what `WILL_FAIL`
asserts, but each one was not separately confirmed to fail through the branch it
names, so a rejection arriving from a different branch would still pass. Golden `d61ad5bf0c928c76`, ecosystem fingerprint `2cd4a15f9fbe9949` and the
frozen genome fixture are unchanged, and ADR-0029's episode driver was not
touched. No deployment manifest or image pin changed. Linux validation is pending;
no Gitea result is claimed since `b01a824`.
Review of `89ad95b` found three defects (2026-09-24), each now fixed with a
regression confirmed to fail against the unfixed code.
The harness recorded a death for every organism still alive when an episode
stopped. A one-tick run therefore reported zero extinctions while writing terminal
death records for founders that had not died, and every such organism's watched
time was added to the same total as a completed lifetime. Ending observation is
not death, and ADR-0034 makes a recorded death terminal, so there was nothing for
the archive to be terminal about. Survivors now stay living, are counted in
`survivors`, and their watched time accumulates in `observed_ticks`, which is a
lower bound on a lifetime the run never saw end. `deaths` and `lived_ticks` cover
only organisms observed to die, and the two are never added together. The run and
archive test now checks that deaths plus survivors equals every organism recorded
and that a lineage is absent exactly when all its carriers were seen to die.
A draw that produced a genome without the reproduction capability derived a zero
reproduction investment, which ADR-0029's driver refused outright, so
`--ensemble --seed 10` exited with invalid_arguments; seeds 12 and 15 were
confirmed to fail the same way and seed 11 to succeed. Filtering such variants out
before sampling was rejected as the worse repair: the measured set would then have
been the fertile subset chosen by a guard, and an inability to reproduce is a
result rather than a reason for exclusion. The driver now accepts the strategy and
proposes no birth for it, since an empty investment would otherwise clear both
reproduction bounds trivially and request a birth transferring nothing. ADR-0029
is amended rather than silently contradicted, and ADR-0035's claim that the driver
does not change was corrected to say what did.
Only configurations the driver previously rejected outright behave differently, so
the M2.6 pin was expected to hold and was checked rather than assumed:
`--ecosystem --seed 0 --episodes 32` still reports `2cd4a15f9fbe9949`, and the
ecosystem, feeding and producer suites pass unchanged. The ecosystem test that
asserted a zero investment is rejected now asserts the opposite fact, that such an
episode runs and reports no births.
A half's thermal shortfall was assigned rather than accumulated, so a field
documented as a total across the half's episodes carried one episode's reading.
Two episodes at 15,000 millikelvin reported 15,000 rather than 30,000. It now
accumulates per episode. The tolerance test's expectations were per-episode
readings and are now stated against the episode count, with an assertion that the
count exceeds one so a sum and a single reading cannot look alike.
`cli-ensemble` was repinned from `c2b96c86101be628` to `5ec59d390909b526`,
confirmed stable across three runs and distinct from seed 7's `090b83efb85ae477`.
This is a change in reported facts, not a test accommodation: survivors are no
longer counted as deaths, the outcome record gained `deaths`, `survivors` and
`observed_ticks`, and the shortfall is now a sum, so the previous value describes
a report this build does not produce. The rationale is recorded beside the pin.
Each fix was confirmed to be load-bearing by reverting it alone and rebuilding.
Restoring the survivor-as-death behaviour failed the archive test at the survivor
count and, when that test was run second, the dedicated regression at `deaths ==
0`. Restoring the zero-investment rejection failed the sterile-strategy regression
at `EcosystemEpisode::create`. Restoring the assignment of thermal shortfall
failed the summation regression. One of my own expectations was wrong rather than
the code: a one-tick episode's survivors include organisms born during that tick,
watched for none of it, so observed time equals one tick per founder rather than
one per survivor. The test now says that.
Coverage limits unchanged from the increment above, with two additions. Censored
lifetimes mean a run's mean lifetime is not computable from `lived_ticks` alone
and no test asserts one. Across a run the archive now ends holding living
carriers, which is accurate about what was observed and is not a claim that
anything persists; nothing is serialized.
## M4.1 cohort state and ownership
Executed on this Windows host (2026-09-24): Release configure, build and `ctest`
34/34; the analysis preset (clang-tidy 19, warnings as errors) built clean; the
ASan preset configured, built and ran 34/34; `clang-format` 19.1.5 reported no
violations. All four checks ran and passed on the committed tree. No Clang or GCC
compile happens on this host, so Linux CI remains the first result for the new
`evo::ecology` library.
`evo-cohort-tests` covers six properties: that a cohort keeps exactly the fields
it was given, that totals equal the summed classes, that two cohorts with
identical totals but different structure are distinguishable, that every
construction invariant is rejected by its own named error, that one habitat and
genome holds one record, and that only the recorded owner under a current epoch
may replace it. A sixth checks that a cohort names a genome the archive resolves
ancestry for without storing that ancestry.
Every guard was confirmed load-bearing rather than assumed. Each of the nine was
removed in turn and the suite rebuilt and rerun: the duplicate, stale epoch, wrong
owner, unowned reserve, residual bound, class ordering, empty class and class
range guards each produced a failing run, and removing the overflow guard fails to
compile, because the running totals then have no reader and the Release build
treats that warning as an error. An earlier attempt at this sweep reported all
nine as not load-bearing; that result was false and was discarded. The harness
wrote to `/tmp`, which Git Bash and PowerShell resolve to different directories,
so the mutation script was never found and the source was never modified. The
sweep was rerun against an absolute path.
Two defects in my own test code were found and fixed before the suite was trusted.
The single-argument `CHECK` macro split on the comma inside a braced initializer
such as `Owner{11, 2}`, which is a compile error rather than a silent one, and the
macro is now variadic. More seriously, the rejection tests originally asserted on
`.error()` alone, which is undefined when the call unexpectedly succeeded and would
have let a removed guard pass quietly; they now check the failure first through a
`refused` helper. A regex rewrite applied to make that change spanned lines and
corrupted several positive assertions, so the test file was rewritten rather than
patched.
Coverage limits. Nothing here is serialized, so `schemas/population-v1.md` pins
field meaning and invariants and there is no encoder, no round trip and no
migration to test. The approximation bound is carried and never computed or
checked, so no test asserts that aggregation is faithful; that is M4.5.
Conservation between a cohort and the rest of the world is not asserted, because
this type holds an inventory without participating in any transfer. Ownership is
tested as a single-process invariant only: epochs make a stale writer fail locally
and nothing here exercises fencing between workers, which is M7 and M10. Negative
reserves are not representable at all, since every quantity is unsigned, so what
the suite rejects instead is a declared state that would require one, and two
static assertions record that the underlying types are unsigned. `max_cohorts` is
enforced but not exercised at its limit, since installing 4,096 records would add
nothing to what the duplicate and lookup tests already establish.
No existing contract changed. The proof golden `d61ad5bf0c928c76`, the ecosystem
fingerprint `2cd4a15f9fbe9949`, the ensemble fingerprint `5ec59d390909b526` and
the frozen genome fixture are all unchanged, and the other 33 tests pass
untouched. No deployment manifest or image pin changed. Linux validation is
pending; no Gitea result is claimed since `b01a824`.
Review of `1c8b292` found two defects (2026-09-24), both fixed with regressions
confirmed to fail against the unfixed code.
Replacement checked the ownership epoch and nothing that ordered two writes made
inside one. A record installed at tick 1,000 holding 50 picojoules could be
replaced by its own earlier tick-900 record holding 100, restoring inventory the
installed record had already spent, because within an epoch the two were
interchangeable. A cohort now carries a revision that strictly increases within an
epoch, and a repeated revision is refused as well as an older one, since a write
has to be strictly later rather than merely not earlier. A newer epoch may restart
the series, because re-acquiring ownership begins a new one. The interval boundary
is checked independently of both, so neither advancing a revision nor re-acquiring
ownership is a way to rewind simulated time. A later revision at the same boundary
stays legitimate: that is a correction, not a rewind.
The implicit move emptied a cohort's age classes while keeping its habitat, genome
and owner, so a moved-from record was the invalid cohort `create` refuses to build
and the registry accepted through a move what it rejects through construction. The
type is now copy-only, which suppresses the implicit moves so a move decays to a
copy. ADR-0030 made the genome copy-only for exactly this reason in M3.1; the
precedent existed and was not applied when this type was written.
Both fixes were confirmed load-bearing by reverting each alone. Removing the
revision check fails the new ordering regression, as does removing the backward
boundary check. Removing the copy declarations fails to compile in Release, and
not because of the static assertion: with that assertion also removed, MSVC proves
the code after `source.classes().size() == 3` unreachable, having established that
a moved-from vector is empty and the check therefore always throws. That is a
compile-time proof of the assertion rather than an execution of it, so the runtime
path was checked separately under the ASan preset, where the same mutation exits 1.
Coverage limits are unchanged from the increment above, with one addition. The
revision is a caller-supplied counter, not one the registry issues, so nothing
prevents a caller from choosing revisions that are unrelated to its own write
order; what is enforced is only that the registry refuses a record that is not
strictly later than the one it holds. Issuing revisions, and fencing between
workers, remain M7 and M10.
No existing contract changed. The proof golden `d61ad5bf0c928c76`, the ecosystem
fingerprint `2cd4a15f9fbe9949`, the ensemble fingerprint `5ec59d390909b526` and
the frozen genome fixture are unchanged. Linux validation is pending; no Gitea
result is claimed since `b01a824`.
## M4.2 collapse to cohort
Executed on this Windows host (2026-09-24): Release configure, build and `ctest`
35/35; the analysis preset (clang-tidy 19, warnings as errors) built clean; the
ASan preset configured, built and ran 35/35; `clang-format` 19.1.5 reported no
violations. All four checks ran and passed on the committed tree. No Clang or GCC
compile happens on this host, so Linux CI remains the first such result.
`evo-collapse-tests` covers seven properties: that a plan totals its members
exactly and moves each member's whole reserve, that ages and the per-class reserve
distribution survive the move without redistribution, that every ineligible state
and every malformed request refuses the whole plan by name, that a committed
collapse leaves the ledger's conserved total unchanged while emptying every member
reservoir and crediting the destination, that every rejection leaves both the
ledger and the registry exactly as they were, that one plan cannot be committed
twice, and that a moved-from plan is still whole.
All ten guards were confirmed load-bearing by removing each in turn and rebuilding:
the ineligible, foreign, duplicate, self-destination, age classification, stale
sample, occupied destination, empty class and commit ordering guards, and the
copy-only declaration. Two needed a second pass. The first age mutation did not
compile, which is not evidence about the test, so it was rewritten to make the
class predicate always match, which places every member in the first class and is
detected at runtime. Removing the copy declarations fails to compile in Release
for the reason recorded for the cohort; with the static assertion also removed the
same mutation was rerun under ASan, where it exits 1, so the runtime assertions
are load-bearing independently of the compile-time proof.
The commit ordering mutation is worth naming: moving the ledger transaction ahead
of the destination check is detected, because a rejected conversion then leaves
biomass in a reservoir nothing owns while the members are already empty. That is
the dual-representation failure this milestone is named after, and the ordering is
what prevents it rather than a comment saying it does not happen.
One expectation of my own was wrong rather than the code. A replay of a committed
plan was expected to report `destination_occupied`, but the staleness check ran
first and reported `stale_plan`, which is equally true. Both are pure checks made
before anything moves, so safety was unaffected and only the diagnosis differed;
the destination check now runs first, because "this conversion already happened"
says more about a replay than "the world moved". The ordering rationale is
recorded beside it.
`accounting::ReservoirId` and `accounting::Transfer` gained the defaulted
comparisons that every other value type in the project already carries, so a
plan's legs can be compared. This is additive, no behaviour depends on it, and
ADR-0037's status was corrected to say so rather than claim no M1 change.
Coverage limits. Completeness of the offered member set is not checkable here and
is not checked: nothing establishes that the caller offered every member it had.
Emptying a reservoir is not withdrawing the right to use it, since the ledger has
no removal, so a later transfer could credit an emptied reservoir and no test
asserts otherwise. The before-and-after property is demonstrated across rejections
inside one process; there is no persistence, so a crash between the ledger
transaction and the registry install is outside what an in-memory registry can
show, and durable conversion sequences, escrow and replay stay M7 and M8. Plans
are bounded to 256 members by the ledger's transaction limit and nothing
coordinates several conversions. The ledger-failure and destination-unavailable
branches are reachable only past bounds the tests do not build, so they are
refused-by-construction rather than exercised.
No existing behaviour changed. The proof golden `d61ad5bf0c928c76`, the ecosystem
fingerprint `2cd4a15f9fbe9949`, the ensemble fingerprint `5ec59d390909b526` and
the frozen genome fixture are unchanged, and the other 34 tests pass untouched.
No deployment manifest or image pin changed. Linux validation is pending; no Gitea
result is claimed since `b01a824`.
Review of `1a59b63` found one defect (2026-09-24), now fixed with a regression
confirmed to fail against the unfixed code.
Commit proved the destination reservoir existed and that the registry slot was
free, but never that the reservoir was empty. A conversion into an occupied
reservoir therefore succeeded and mixed inventory it had never sampled into the
cohort's backing store, leaving the cohort's recorded reserve understating what
backed it and the surplus owned by nobody. The reported case was reproduced
against the unfixed code rather than assumed: gathering into reservoir 5, which
the fixture already seeds with seven picojoules, commits successfully and leaves
the ledger holding 67 against a cohort recording 60.
A destination holding anything is now refused. The success test additionally
asserts that the destination balance equals the installed cohort's reserve, so a
cohort's reserve is the whole of its reservoir rather than a part of it, and the
new regression asserts that the refusal leaves the ledger balances, the ledger
action counter and the registry all exactly as they were, then that emptying the
reservoir first makes the same plan committable.
The check runs after the staleness loop rather than beside the other destination
checks. Both run before anything moves, so this is a reporting choice and not a
safety one: replaying a committed plan into a fresh registry would be refused by
either, and that the members no longer hold what was sampled is the more useful
answer than that the reservoir is occupied. The ordering rationale is recorded
beside the check.
The guard was confirmed load-bearing by removing it alone, which fails the new
regression. An earlier attempt to add that mutation reported the guard as not
load-bearing; the result was void and discarded, because a Bash heredoc had
mangled the escaped newlines in the mutation script and the script failed to
parse, so nothing was ever mutated. It is the second time this environment quirk
has produced a false negative in a mutation sweep, and both times the printed
output rather than the exit status is what revealed it.
## M4.3 interval accounting
Executed on this Windows host (2026-09-24): Release configure, build and `ctest`
36/36; the analysis preset (clang-tidy 19, warnings as errors) built clean; the
ASan preset configured, built and ran 36/36; `clang-format` 19.1.5 reported no
violations. All four checks ran and passed on the committed tree. No Clang or GCC
compile happens on this host, so Linux CI remains the first such result.
`evo-population-tests` covers ten properties, and `evo-cohort-tests` gained two
for the amendments. An interval with no flux advances the clock and submits no
ledger transaction. Births change the population and leave the reserve untouched.
Death, predation, immigration and emigration each move biomass once to the named
counterparty with the conserved total unchanged. A loss exceeding the population
or any one inventory channel is refused rather than clamped, including at the
boundary of a single millionth of a member, while taking exactly everything is
vacancy rather than refusal. A population below one member stays a population and
its record stays installed. A declared result that does not match the fluxes is
refused in both directions, for counts and for biomass. Empty and backward
intervals, a foreign owner and a repeated revision are refused. A commit against a
record that has moved on, or a reservoir that no longer holds what was counted,
changes nothing in either the ledger or the registry, and one plan cannot be
committed twice.
The seeded ensemble runs twenty-four seeds of forty proposed intervals each,
deliberately proposing losses unbounded relative to the cohort so that some must
be refused. Every accepted interval is checked for the conserved total being
unchanged and for the cohort's reservoir equalling its recorded reserve. The test
asserts that both the accepted and the refused path actually occurred, so a change
that made everything refuse, or everything pass, fails rather than looking clean.
All fourteen guards were confirmed load-bearing by removing each in turn and
rerunning: the two no-clamp guards, the two mismatch guards, the interval, owner,
revision and counterparty guards, the three commit staleness guards, and the three
amendments to the cohort record. The two clamp mutations are the ones worth
naming, because they are the defect rather than a missing check: replacing the
refusals with saturating subtraction, which is what a clamp is, is detected.
Two of my own expectations were wrong rather than the code. The escrow arithmetic
in the migration test was miscalculated: the reservoir receives two hundred matter
and returns one hundred, so from seven it stands at a hundred and seven, not a
hundred and ninety-seven. And the dwindling-population test shed all but one
femtogram of the reserve while declaring a result that held none, which is a real
`biomass_mismatch`; the test now sheds the whole reserve. That second one surfaced
a genuine tension worth recording: a class reduced to a fraction of a member may
hold no reserve, because ADR-0036 gives reserve to members and a fraction is not
one yet, so that biomass has to go somewhere in the same interval. It is recorded
in ADR-0038's costs rather than worked around.
ADR-0036 was amended twice, and both were forced by this milestone rather than
chosen. Threading the reservoir through each request instead of binding it to the
record would have let a caller advance one cohort against another's store, and a
balance comparison cannot distinguish two equal reserves, so the binding is a
correctness fix and not tidiness. Vacancy is required because a decline to exactly
zero was otherwise unrepresentable and the only way through would have been the
clamp this milestone exists to refuse. `Cohort::vacant` is a separate factory
precisely so that an empty class list in a draft stays an error.
Coverage limits. Closure is checked and stability is not: an interval whose
arithmetic balances and whose ecology is nonsense is accepted, and no test asserts
otherwise, because comparing matched ensembles is M4.5. The declared age structure
is not verified to be what ageing would produce, since that needs the within-class
distribution the record discards; nothing here checks that members did not get
younger. Emigration escrows into a reservoir nothing settles with. Ownership
remains a single-process invariant. The overflow and ledger-failure branches are
reachable only past bounds the tests do not build, so they are refused by
construction rather than exercised. Nothing is serialized.
The proof golden `d61ad5bf0c928c76`, the ecosystem fingerprint
`2cd4a15f9fbe9949`, the ensemble fingerprint `5ec59d390909b526` and the frozen
genome fixture are unchanged, and the other 33 tests pass untouched. No deployment
manifest or image pin changed. Linux validation is pending; no Gitea result is
claimed since `b01a824`.
## M4.4 materialization
Executed on this Windows host (2026-09-24): Release configure, build and `ctest`
37/37; the analysis preset (clang-tidy 19, warnings as errors) built clean; the
ASan preset configured, built and ran 37/37; `clang-format` 19.1.5 reported no
violations. All four checks ran and passed on the committed tree. No Clang or GCC
compile happens on this host, so Linux CI remains the first such result.
`evo-materialize-tests` covers ten properties: that a drawn member receives the
floor share while the rounding remainder stays with the class, that a class drawn
empty hands over everything it held, that neither a residual nor a member that is
not there can be drawn, that taking everyone leaves a vacant record rather than an
extinction, that ages come from the top of a class and are stable when regathered,
that a round trip is exact at one member per class, that a coarse class conserves
the total while smoothing individuals, that repeated oscillation never pays, that
malformed draws and stale views are refused without changing anything, and that a
moved-from plan is whole.
The oscillation test runs twelve collapse and materialize cycles over three
members of unequal reserve and asserts three things every cycle: the ledger's
conserved total never moves, no member ever holds more than the largest starting
member held, and no member is younger than the corresponding member of the
previous cycle. Those three together are what stops a round trip being a way to
heal or to reset age. It also ends by asserting every member has been pushed to a
class ceiling, which records the cost of the conservative age rule rather than
leaving it as a claim.
Eleven guards were confirmed load-bearing by removing each in turn and rerunning.
Two of them are the named risk rather than ordinary validation, and both are
detected: rounding the share up instead of down, which creates biomass, and
drawing ages from the bottom of the class instead of the top, which returns
members younger than they were and makes oscillation profitable.
The free-healing risk has two sources and only one is closed. The arithmetic
source is prevented outright and tested. The other is the declared resolution
itself: within a class members are indistinguishable, so a coarse class heals a
starving member at a well-provisioned member's expense while conserving the total
exactly. That is not fixable by accounting and is asserted rather than avoided,
with a companion test showing a round trip is exact when each class holds one
member. Reading conservation of the total as evidence that any individual is right
would be a mistake this pair of tests exists to prevent.
Coverage limits. Materialization is not claimed to be the inverse of collapse and
no test asserts that it is beyond the one-member-per-class case. Nothing checks
that a proposed draw is ecologically sensible. A full draw yields at most one unit
per member more than a partial one, which is an ordering artifact bounded by the
class count and is not separately tested at every count. Injury, fertility and
incubation are not reconstructed because a cohort never carried them. Ownership
remains a single-process invariant, emigration escrow is still unsettled, and
nothing is serialized. The overflow and ledger-failure branches are reachable only
past bounds the tests do not build.
No existing behaviour changed. The proof golden `d61ad5bf0c928c76`, the ecosystem
fingerprint `2cd4a15f9fbe9949`, the ensemble fingerprint `5ec59d390909b526` and
the frozen genome fixture are unchanged, and the other 36 tests pass untouched. No
deployment manifest or image pin changed. Linux validation is pending; no Gitea
result is claimed since `b01a824`.
Review of `380d609` found two defects (2026-09-24), both fixed with regressions
confirmed to fail against the unfixed code.
Materialization treated an empty reservoir as an available one. A vacant cohort, or
one simply holding nothing, still owns its store, so a draw could credit that store
while the owning record went on recording nothing: the same inventory in two places,
reached from the opposite direction to the destination defect found in `1a59b63`.
Destinations are now checked against the registry's ownership as well as their
balance, and the registry answers whether a reservoir is used rather than callers
inferring availability from what it contains.
The same hole existed on the collapse side in a worse form, and was found by
checking whether the reported defect had a sibling rather than by being reported.
Nothing verified that the destination reservoir was unowned, so the shared-reservoir
rule added in M4.3 would have refused the install only after the ledger transfer had
already run, leaving the members emptied and their biomass in a reservoir nothing
accounted for. That is exactly the stranding the M4.2 ordering was chosen to prevent,
defeated by a check that ran too late. Whichever registry step will follow a transfer
is now proved to succeed before it, through new `may_install` and `may_replace`
queries that `install` and `replace` themselves delegate to, so there is one rule
rather than a restatement that could drift.
The second finding was a functional gap the oscillation test had been hiding by
discarding its registry every cycle. A full draw leaves a vacant record, which
ADR-0036 retains because absence is not extinction, and collapse then refused that
habitat and genome as occupied. A full materialization was therefore a one-way door
out of a habitat: the lineage that had been there could never return. Collapse now
repopulates a vacant record in place, continuing its ownership and revision series
rather than restarting them, and a record that still holds a population is still
refused, because merging two silently is not a conversion. The oscillation test keeps
one registry for all twelve cycles and asserts the revision continues across each
repopulation, so the round trip is exercised against the world it left behind.
Four of the five new guards were confirmed load-bearing by removing each alone: the
ownership check on both sides, the collapse pre-check ordering, and the repopulation
path. The fifth is not, and is recorded rather than presented as tested. The
`may_replace` pre-check in `commit_materialize` is unreachable given the checks
before it: the installed record has already been proved equal to the one the plan
was built from, and the plan's result carries the same habitat, genome, reservoir and
tick with a strictly later revision, so the replacement cannot be refused. Removing
it changes nothing the tests can see. It is kept because it is the structurally
correct place for the guard and its sibling on the collapse path is genuinely
reachable, not because it is exercised.
An environment note for the record: a Bash heredoc containing apostrophes failed to
parse, so the larger test edits in this increment were written to a scratchpad file
and run from there. That is the third distinct way heredocs have mangled content in
this project, after backslash escapes and `/tmp` resolving differently between
shells.
## M4.5 resolution comparison
Executed on this Windows host (2026-09-24): Release configure, build and `ctest`
38/38; the analysis preset (clang-tidy 19, warnings as errors) built clean; the ASan
preset configured, built and ran 38/38; `clang-format` 19.1.5 reported no
violations. All four checks ran and passed on the committed tree. No Clang or GCC
compile happens on this host, so Linux CI remains the first such result.
The increment adds no library code apart from one defect fix described below. One
rule is written once and applied at both resolutions; the individual run gives it a
member at a time and the aggregate run gives it a class average and multiplies by
the count. The aggregate run sub-steps at one tick, so interval-length error is not
confounded with averaging error, and interval-length error is therefore unmeasured.
The control is the part that makes the rest mean anything. With one member per class
a class average is that member, so the two runs must agree exactly; the test checks
population and every channel at every tick over forty ticks, and checks the class
count stays inside the sixty-four-class bound. Introducing a one-unit drift into the
aggregate upkeep is caught by it, so the instrument was verified rather than assumed.
Without this control a measured divergence would be indistinguishable from a
transcription mistake in one of the two runs.
Measured, and reported rather than smoothed into a tolerance. Class widths ten and
fifty: the aggregate population survives ten thousand ticks, with population gaps of
250 and 375 parts per thousand and energy gaps of 24 and 43. Class widths
twenty-five, one hundred and two hundred: the aggregate population reaches zero while
the individual run sustains seventy-five members. That is a false extinction caused
by resolution alone. The numbers are in
`docs/engineering/results/resolution-equivalence-msvc1944.jsonl`.
The collapse is not monotone in class width: twenty-five destroys the population and
fifty sustains it. An earlier version of the test asserted that coarser is never
better, and the measurement refuted it; the assertion was removed rather than the
widths being chosen to make it hold, and the non-monotonicity is now asserted
instead, because it is surprising and because the warning written around it could
otherwise become wrong without failing. Five widths locate the effect and do not map
it; no mechanism beyond whole-class starvation is claimed.
Conservation was exact at every tick on both sides of all five runs, including the
three whose populations died out. Four of five conserve perfectly and three are
ecologically worthless, which is the clearest available demonstration of
ECOLOGY_MODEL's point that conservation is insufficient ecological evidence.
Two flaws in my own experiment were found and fixed before any number was believed.
The aggregate model originally shifted class bounds every tick and opened newborn
classes at the full declared width, which produced overlapping classes: the model
being measured was not one a cohort record could represent, and the first numbers
measured were therefore of a model the library could not hold. Newborns now extend
the youngest class down while it is narrower than the width, keeping the structure
contiguous and ascending, and every step of the aggregate run is now checked to
construct a valid `Cohort`. The earlier figures were discarded.
The comparison also found a real defect in M4.3. `plan_advance` checked affordability
with arrivals added before losses were taken, but emitted the ledger legs with the
outflows first. The ledger applies legs in sequence, so an interval affordable as a
whole underflowed leg by leg and was refused at commit with `ledger_failure`: a
fifty-tick interval shedding 10,480 energy against 6,950 held while drawing 13,080
in reproduced it exactly. Arrivals are now the first leg, matching the order the
arithmetic was validated in. The fix was confirmed load-bearing by restoring the old
order, which fails both suites, and a dedicated regression in `evo-population-tests`
covers an interval affordable only with its arrivals. One existing test had seeded
its escrow assuming the old order and was corrected, with its comment; that comment
had asserted the old ordering was what made the interval fundable, which was true
and is now true of the opposite order.
Coverage limits. One genome, so the mixed-genome diversity case ECOLOGY_MODEL also
asks for is untested. One model, one scarce axis, one initial distribution and one
capacity, so every number describes this fixture. The control runs over tens of
ticks because one class per age exhausts the class bound, while the measurement runs
over ten thousand: the instrument is checked at a shorter horizon than it is used at.
The asserted envelope bounds are round numbers above the measurement with room for
platform variation, so they are a regression guard and not a theoretical limit.
Interval-length error, crash behaviour at conversion boundaries beyond one process,
distribution across owners and persistence are all out of scope here.
The proof golden `d61ad5bf0c928c76`, the ecosystem fingerprint `2cd4a15f9fbe9949`,
the ensemble fingerprint `5ec59d390909b526` and the frozen genome fixture are
unchanged. No deployment manifest or image pin changed. Linux validation is pending;
no Gitea result is claimed since `b01a824`.
Review of `b4c3da5` found one defect (2026-09-25), now fixed with the controls
extended to catch it.
The two runs allocated contested breeding slots in different orders. The individual
run iterated members in insertion order; the aggregate run walked classes as stored,
which is youngest first. When eligible parents outnumber free slots that decides
which parents reproduce, and it does so for a reason that has nothing to do with
averaging, so the published divergence was not attributable to resolution. The
control never reached the capacity boundary and so could not have seen it, which is
the part that made ADR-0040's central claim unsupported rather than merely
imprecise.
The allocation order is now part of the stated rule, oldest first, honoured at both
resolutions: classes partition ages, so the oldest class holds the oldest members,
and ordering inside a class remains something only the individual run can see, which
is a resolution effect rather than an implementation difference.
A second control was added at the capacity boundary, where twelve eligible parents
contend for one free slot. Both controls now compare the whole distribution of
members rather than censuses. That change matters on its own: whoever breeds hands
the same investment to a child either way, so every total agrees while the
distribution does not, and a census-only control cannot see a different parent
having bred. The reported probe diverging only by tick 44 is exactly that effect.
The attribution was then verified by mutation rather than argued. Restoring the
individual run's insertion order fails the new control; restoring the aggregate
run's youngest-first order fails it; and the earlier one-unit drift in the aggregate
upkeep still fails the first control. Since either allocation order alone fails,
the divergence reported in the results file is attributable to averaging.
One of my own attempts at the control was wrong and was caught by the same mutation
sweep. The contested fixture first gave ages that descended with the index, on the
reasoning that this made insertion order differ from age order; it does the
opposite, since descending ages make insertion order identical to oldest-first, and
the individual-side mutation passed. Ages now ascend with the index, so oldest-first
disagrees with both the insertion order and the stored class order, and each
mutation alone is caught. Had the sweep not been run, the new control would have
looked like it established attribution while covering only half of it.
The measurements were regenerated and are unchanged: the same five widths, the same
two regimes, the same gaps. That is worth stating plainly rather than presenting as
confirmation, because the finding was about what the numbers could be attributed to
and not about their values.
Coverage limit added. Attribution rests on the controls covering every stage where
the two runs could order things differently. Breeding allocation is the only
order-sensitive stage, and the environment is sized so intake never depletes it,
which the fixture asserts rather than assumes. A future stage introducing another
order-sensitive decision would need its own control before these numbers could be
trusted again.
## M4.6 local validation - 2026-09-26
Base 55a728a plus working M4.6 changes. Integrated the prior agent's untracked
`tests/fixtures/resolution_model.hpp`; no production library changes or format
version changes. Existing contested-parent and round-trip controls remain active.
- Native MSVC 19.44 Release configure/build and CTest: 39/39 passed.
- Native analysis (clang-tidy 19, including fixture headers): build and 39/39 passed.
Initial analysis found a throwing error-reporting boundary in the benchmark;
replaced catch-handler stream writes with stdio diagnostics and reran successfully.
- Native AddressSanitizer: build and 39/39 passed. Windows ASan does not cover UBSan.
- clang-format 19 format-check passed, now including tests/*.hpp. Diff whitespace
check passed.
- Full Release benchmark: 15 scenarios, one warmup plus three measured repetitions
per representation. Raw results and host/method caveats are in
[AGGREGATE_BUDGET.md](AGGREGATE_BUDGET.md). All scenarios conserved every channel;
none qualified across all scarcity inputs under the proposed research gate.
- Linux CI, UBSan, target-server profiling and production deployment were not run.
No World0 resources or user source documents were changed.
The final exception-handler-only adjustment was compiled in analysis; Release and
ASan benchmark targets were then rebuilt and their resolution tests rerun. No
performance or model algorithm changed after the archived timing run.
## M5.1 local voxel contract - 2026-09-26
Added ADR-0042, the local voxel contract and dependency-free `evo::world`. No
existing simulation golden, codec or runtime dependency changed. The earlier
uncommitted M5-M9 roadmap expansion is preserved in the working tree.
Executed on Windows x64 / MSVC 19.44 with the repository helper:
- `tools/windows-check.ps1 -Preset release -Action check`: configure/build,
warnings-as-errors and 40/40 CTest passed.
- `tools/windows-check.ps1 -Preset analysis -Action check`: clang-tidy 19 build
and 40/40 CTest passed.
- `tools/windows-check.ps1 -Preset asan -Action check`: AddressSanitizer build
and 40/40 CTest passed. Native Windows ASan is not UBSan coverage.
- `tools/windows-check.ps1 -Preset release -Action format`: passed.
- `git diff --check`: passed.
New Release-active tests enumerate independent signed chunk intervals, all 299,520
logical cell ordinals across edges 8/16/32/64, envelope corners, each-axis invalid
bounds including INT64_MIN/MAX and UINT32_MAX, metric coordinates and all valid
positive fill numerators. Typed rejection is checked before reading expected values.
No allocation/terrain-generation/performance claim is made for the address envelope.
Linux GCC/Clang CI and UBSan have not been run for this increment. No container,
Argo or World0 deployment occurred. M5.2 storage benchmarks remain the next task.
## M5.2 compact chunk comparison - 2026-09-26
Added `evo-chunk-layout`, Release-active CTest checks, ADR-0043 and the
[measured report](CHUNK_LAYOUT_EVALUATION.md). No production world API, persisted
format or simulation golden changed.
Executed on Windows x64 / MSVC 19.44:
- `tools/windows-check.ps1 -Preset release -Action check`: 41/41 passed.
- `tools/windows-check.ps1 -Preset analysis -Action check`: clang-tidy 19 and
41/41 passed after fixing a multiplication-width warning and reserving edit input.
- `tools/windows-check.ps1 -Preset asan -Action check`: 41/41 passed.
- `tools/windows-check.ps1 -Preset release -Action format`: passed.
- Release `evo-chunk-layout`: completed; archived 180 measured samples after warmups.
Independently checked matching initial/edited hashes across all layouts and repeats
for each of 20 edge/fixture pairs. Full per-cell oracle checks also run internally.
- `git diff --check`: passed.
Memory figures are owned capacities, not RSS or true allocation peaks; encoding
sizes are estimates, not a codec. See the report for excluded buffers and timing
limits. Native ASan does not cover UBSan. Linux CI, R730xd benchmarking and Kubernetes
execution were not run locally. No Argo/World0 deployment occurred. Next: M5.3.
## M5.3 deterministic abiotic generation - 2026-09-26
Implemented ADR-0044 in evo::world with a versioned inert material table and pure
point/chunk queries. The fixture creates no biological state. No existing simulation
version, golden or persistence codec changed.
Executed on Windows x64 / MSVC 19.44:
- `tools/windows-check.ps1 -Preset release -Action check`: 42/42 CTest passed.
- `tools/windows-check.ps1 -Preset analysis -Action check`: clang-tidy 19 build and
42/42 CTest passed.
- `tools/windows-check.ps1 -Preset asan -Action check`: 42/42 CTest passed.
- `tools/windows-check.ps1 -Preset release -Action format`: passed.
- `python tools/world_generation_reference.py`: computed the 12 frozen fixture
fingerprints independently before C++ execution; all match in CTest.
- `git diff --check`: passed.
New tests cover 262144 point/chunk comparisons across signed seams, query order,
stratigraphy, known/unknown materials, version/edge rejection and extreme addresses.
Fingerprint bytes are explicitly little endian logical fields. Linux CI and UBSan
were not run locally; no cross-platform execution result is claimed. No deployment
or World0 state change occurred. Next: M5.4 bounded sparse chunk ownership.
## M5.4 bounded chunk ownership - 2026-09-26
Added CompactCells, a fixed-slot ChunkStore, revision-validated read tokens and
conservative dirty pinning under ADR-0045. No cell edit, persistence, distributed
lease or clean acknowledgement API was introduced. Existing goldens remain unchanged.
Executed on Windows x64 / MSVC 19.44:
- `tools/windows-check.ps1 -Preset release -Action check`: 43/43 CTest passed.
- `tools/windows-check.ps1 -Preset analysis -Action check`: clang-tidy 19 build and
43/43 CTest passed.
- `tools/windows-check.ps1 -Preset asan -Action check`: 43/43 CTest passed.
- `tools/windows-check.ps1 -Preset release -Action format`: passed.
- `git diff --check`: passed.
The initial test oracle mixed unsigned local offsets with a negative integer literal;
corrected to explicit int64 arithmetic. Analysis required explicit optional checks
at reads/eviction and identified an integer-width expression in the test. One narrow
use-after-move suppression exercises the explicitly guaranteed empty-void source.
Final checks above ran after those corrections.
Coverage includes palette counts through 256 then dense fallback, full logical cell
equality, material/fill distinctions, move validity, extreme/far addresses, exact
accounted budget boundaries, failed-load state preservation, slot pressure, clean
regeneration, stale/foreign tokens and dirty eviction refusal. Counter exhaustion is
guarded but not injected; no allocation-failure fault injector is claimed. Memory
accounting excludes allocator/runtime overhead and is not an RSS cap.
Linux CI/UBSan and target-host performance have not been run locally. No deployment
or World0 change occurred. Next: M5.5 accounted transactional edits.
## M5.5 transactional excavation - 2026-09-26
Added bounded atomic excavation, finite per-material debris volume, full-payload
retry identity and a 64-command replay ring under ADR-0046. Existing biology/world
base goldens and persisted formats were not changed. Formatter coverage now includes
private headers under src, including the shared chunk-state definitions.
Executed on Windows x64 / MSVC 19.44:
- `tools/windows-check.ps1 -Preset release -Action check`: 44/44 passed, including
the final late-source-failure test and nonthrowing commit static assertions.
- `tools/windows-check.ps1 -Preset analysis -Action check`: clang-tidy 19 build and
44/44 passed. After the final targeted additions, rebuilt analysis and reran
`ctest --test-dir build/analysis -R world-edit-contracts --output-on-failure`: passed.
- `tools/windows-check.ps1 -Preset asan -Action check`: 44/44 passed. After those
additions, rebuilt ASan and reran the same targeted test in build/asan: passed.
- `tools/windows-check.ps1 -Preset release -Action format`: passed.
- `git diff --check`: passed.
Coverage includes seam-crossing excavation, four chunks/64 cuts, partial fills,
rock/regolith volume conservation, all-or-nothing capacity/memory/source failures,
missing/stale revisions, duplicate/changed requests, malformed bounds and retention
expiry. Late source failure verifies an earlier prepared chunk is not published.
No allocator failure injection, revision exhaustion injection or crash recovery is
claimed. The source/replacement commit operations have compile-time nonthrowing guards.
Linux CI and UBSan were not run locally. These are typed volume transfers, not a
validated density/mass/chemistry model. No deployment or World0 modification occurred.
Next: M5.6 canonical snapshots/edit journals preserving geometry, debris and retry state.
## M5.6 canonical world snapshots and edit journals - 2026-09-26
Added ADR-0047, byte schemas, canonical snapshot/restore and detached atomic journal
replay. State includes base versions, identity, membership, revisions, geometry,
debris, successful edit count and exact bounded retry records. No existing golden
or biological simulation format changed. New snapshot v1 has an independently encoded
132-byte empty fixture from tools/world_snapshot_reference.py.
Executed on Windows x64 / MSVC 19.44:
- Release configure/build and 46/46 CTest passed on the final source/tests.
- Analysis configure/build (clang-tidy 19) and 46/46 CTest passed after explicit
size/iterator conversion fixes. Final history-corruption additions rebuilt and
passed both tests selected by `ctest --test-dir build/analysis -R world-codec
--output-on-failure`.
- ASan configure/build and 46/46 CTest passed. After final size/iterator/golden and
history-corruption additions, rebuilt ASan and reran both world-codec tests: passed.
- `tools/windows-check.ps1 -Preset release -Action format`: passed.
- `python tools/world_snapshot_reference.py`: generated independent golden bytes;
encoder equality and decoder acceptance pass in CTest.
- `git diff --check`: passed.
New coverage includes 65-edit history rollover, deterministic continuation, historical
retry receipts, exact re-encoding, snapshot+journal replay, wrong predecessor/final
state, every fixture truncation, oversized/trailing data, version/identity/budget
rejection, recomputed-checksum geometry/debris/receipt/revision/history corruption,
and a deterministic 128-input semantic fuzz smoke. The fuzz harness mutates valid
fixtures and repairs checksums so validation below the envelope is exercised.
Linux libFuzzer/UBSan jobs are wired for 10000 cases but were not executed locally.
Native Windows ASan is not UBSan. Full-size throughput, allocation-failure injection
and crash-atomic files remain unverified/later gates. Codecs write no files, clear no
dirty flags and issue no durable acknowledgements. No Argo/World0 operation occurred.
Next: M5.7 local snapshot/journal publication, compaction and recovery.
## M5.7 review and M5.8 bounded world proof - 2026-09-26
Reviewed a2a12fdc696c005bd3feb496f6efcdc58c28262c against the three prior findings.
Committed-journal compaction compares exact recovered canonical state; poisoned
handles refuse recovery; interruption tests use changed payloads and include partial
write failures. No new blocking finding identified in that correction's scope.
M5.8 adds ADR-0049, evo-world-proof and world-process-restart CTest. It launches fresh
prepare/compact/verify processes for three workloads, checks all cells and conserved
debris independently, compares canonical bytes, enforces frozen hashes and checks
clean regeneration. Existing formats and simulation goldens are unchanged.
Executed on Windows x64 / MSVC 19.44:
- tools/windows-check.ps1 -Preset release -Action check: 48/48 passed, including the
final source with frozen hashes and nonthrowing boundary diagnostics.
- tools/windows-check.ps1 -Preset analysis -Action check: clang-tidy 19 and 48/48
passed. Initial widening-conversion and exception-boundary diagnostics were fixed.
- tools/windows-check.ps1 -Preset asan -Action check: 48/48 passed.
- tools/windows-check.ps1 -Preset release -Action format: passed.
- Five independent Release process workflows passed; medians/ranges and exact
accounting/payload sizes are in WORLD_PROOF.md. Measurements preceded only the
equivalent size_t constant and error-reporting correction; no success-path
workload, snapshot or algorithm changed.
- git diff --check: passed.
CTest integrates the same frozen expectations into existing Linux CI, but this
turn did not execute Linux, UBSan, physical power-cut, target-server or cluster
checks. ASan is not UBSan. Process restarts are graceful; M5.7 fault injection
remains separate coverage. No production durability or planet capacity claim.
No Argo or World0 operation occurred. Next: M6.1 units and environmental coupling.
## M6.1 environmental units and coupling - 2026-09-27
Added ADR-0050, ENVIRONMENTAL_UNITS.md, checked local habitat geometry, exact mass
concentration/water saturation and rational pJ/tick integration helpers. No field
storage, density converter, source scheduler, gameplay or snapshot format changed.
Terrain volume still cannot be credited as biological mass.
Executed on Windows x64 / MSVC 19.44:
- `tools/windows-check.ps1 -Preset release -Action check`: 49/49 passed.
- `tools/windows-check.ps1 -Preset analysis -Action check`: clang-tidy 19 and
49/49 passed.
- `tools/windows-check.ps1 -Preset asan -Action check`: 49/49 passed.
- After strengthening independent balance sums, open-boundary and rollback tests,
reconfigured each preset, rebuilt `evo-environment-units-tests`, and ran
`ctest --test-dir build/<preset> -R '^environment-units-contracts$' --output-on-failure`:
passed in Release, analysis and ASan. Final code-only adjustment was an explicit
cstddef include; CTest registration was grouped with other world tests.
- `tools/windows-check.ps1 -Preset release -Action format`: passed.
- `git diff --check`: passed.
Coverage includes each invalid box axis, zero/maximum inputs, exact dimensional
examples, concentration/type separation, dry/saturated/excess water, multiplication
and addition overflow, invalid denominators/residuals, and 78,660 repeated-addition
oracle comparisons of split versus combined intervals. The closed/open accounting
example uses a finite boundary reservoir, independently sums balances and verifies
rejected overdraft leaves all balances unchanged. Documentation's large terrain
mass example was checked using exact BigInteger division (1e21 / 65535).
Linux/UBSan and target-cluster checks were not executed locally; existing CI includes
the new CTest. There is no new concurrency requiring TSan work. Helpers do not enforce
owner identity, interval deduplication or residual persistence; those remain explicit
M6.2-6.7 obligations. No deployment or World0 operation occurred. Next: M6.2.
## M6.2 optional environmental fields - 2026-09-27
Added ADR-0051, FieldStore, field snapshot v1 and its byte schema. Optional water,
N/P, organic matter/energy, incident-energy and heat channels have explicit finite
openings. Material defaults specify water capacity only. Uniform baselines with
optional 64-site pages preserve absence versus zero. Same-channel local transfers
stage replacements before commit and persist a strictly increasing command watermark.
There is no resource setter, site regeneration, cross-compartment or organism credit.
Executed on Windows x64 / MSVC 19.44:
- `tools/windows-check.ps1 -Preset release -Action check`: 50/50 passed on final code.
- `tools/windows-check.ps1 -Preset analysis -Action check`: clang-tidy 19 and
50/50 passed on final code.
- `tools/windows-check.ps1 -Preset asan -Action check`: 50/50 passed on final code.
- `tools/windows-check.ps1 -Preset release -Action format`: passed.
- `git diff --check`: passed.
Initial build diagnostics corrected explicit aggregate initialization and bit widths.
The hostile fixed-underlying-type enum test has a narrowly scoped analyzer suppression
with rationale; actual channel validation remains exercised.
Tests cover uniform/sparse/varying memory, page seams and release, exhausted sources,
source/capacity/budget/stale-command rollback, absent/present-zero restore, cross-field
conserved-channel overflow, independent dense transfer continuation and all ten fields
at 4096 sites. Maximum logical snapshot is 630880 bytes. A separately constructed
170-byte absent-field fixture checks canonical schema bytes. Every truncation of a
260-byte fixture, invalid/old versions, identity, malformed/resealed mappings and
quantities, oversized snapshots and 256 repaired-checksum mutations are exercised.
Memory observations are in ENVIRONMENTAL_FIELDS.md; accounted bytes are not RSS.
Existing terrain/simulation formats and goldens are unchanged. Field snapshot v1
is a buffer codec, not LocalWorldFiles publication or combined terrain/biology
durability. Rate sources and their interval/residual persistence are not enabled.
Linux/UBSan, target-host performance, allocator-failure injection and cluster recovery
were not executed locally. Existing Linux CTest CI includes the new test. No deployment
or World0 operation occurred. Next: M6.3 read-only environment views.
## TSan world-proof scratch mount correction - 2026-09-27
The supplied Linux TSan log failed world-process-restart at CMake MAKE_DIRECTORY,
before launching any proof phase. Its build-local output path had no writable mount
in the read-only test container; Testing and cli-integration already had mounts.
Added the matching 16 MiB, mode-1777, noexec/nosuid tmpfs for
`/source/build/tsan/world-proof`. Container user, rootfs, network, capabilities and
seccomp restrictions are unchanged. Documented all test scratch mounts in deploy/README.md.
Executed `ctest --test-dir build/release -R '^world-process-restart$' --output-on-failure`
after configuring the Release preset: passed on Windows. `git diff --check` passed.
The workflow mount path was checked against the CMake test's TEST_DIR. No C++ changes
or new sanitizer claims. Local Docker CLI exists, but docker version could not connect
because the dockerDesktopLinuxEngine named pipe is absent. Consequently the corrected
Linux TSan container invocation requires confirmation in Gitea CI; it was not run
locally. No application deployment or World0 changes.
## M6.3 read-only environment views - 2026-09-27
Added ADR-0052 and a bounded canonical footprint query with typed mass/energy,
exact concentration/moisture, habitat resolution, explicit caller tick, field
watermark and sampled chunk revisions. FieldStore gained read-only identity and
capacity accessors. Views own copied samples and retain private non-owning store/
chunk tokens for freshness checks. Existing formats, versions and goldens are unchanged.
Executed on Windows x64 / MSVC 19.44:
- `tools/windows-check.ps1 -Preset release -Action check`: 51/51 passed on final code.
- `tools/windows-check.ps1 -Preset analysis -Action check`: clang-tidy 19 and
51/51 passed. Removed an ineffective move of a trivially copyable sample after
the initial analysis diagnostic.
- `tools/windows-check.ps1 -Preset asan -Action check`: 51/51 passed on final code.
- `tools/windows-check.ps1 -Preset release -Action format`: passed.
- `git diff --check`: passed.
Tests cover negative/positive chunk seams, input permutation, missing site/chunk,
absent versus zero/unrequested fields, all moisture statuses, invalid masks and
footprints, exact quantities/ratios, maximum 256-site footprint with all ten fields,
and unchanged terrain/field snapshots after queries. Freshness checks cover changed
tick, restored foreign stores, successful versus rejected field transfers, global
field invalidation, unrelated terrain activity, sampled dirty marking, excavation
and eviction/reload. The query never loads missing chunks or authorizes consumption.
Linux/UBSan/TSan and cluster execution were not run locally; existing CI includes
the new CTest without additional scratch mounts (this test writes no files).
Host tick/store pairing and single-owner execution are explicit preconditions.
ChunkStore does not yet bind a world identity, so coherent field/terrain pairing
cannot be independently authenticated by this view. Combined ownership, reservation,
atomic inventory transfer and retry outcomes remain M6.4. No deployment or World0
operation occurred.
## M6.4 atomic environmental transfers - 2026-09-27
Executed on Windows x64 with the installed Visual Studio toolchain:
- `tools/windows-check.ps1 -Preset release -Action check`: build and 52/52 CTests passed (NDEBUG).
- `tools/windows-check.ps1 -Preset analysis -Action check`: clang-tidy/warnings-as-errors build and 52/52 CTests passed.
- `tools/windows-check.ps1 -Preset asan -Action check`: AddressSanitizer build and 52/52 CTests passed.
- `tools/windows-check.ps1 -Preset release -Action format`: passed.
New environment-transfers-contracts exercises last-stock contention, fresh versus
stale/foreign views, exact retries, changed-payload refusal, expired command IDs,
command exhaustion, late-leg rollback, account and water capacities, admission
overflow, separate owner/field allocation budgets, and 100 batches checked against
independent expected balances across all ten compartments. No golden changed.
This evidence covers local in-memory atomicity. Coupled snapshots, allocation-failure
injection, process crash recovery and biological lifecycle integration are not claimed.
Linux/UBSan/TSan were not run locally; their existing Gitea jobs remain required CI
evidence. No cluster workload or World0 state was changed.
## M6.5 prerequisite review - 2026-09-27
Documentation-only review identified contradictory source-persistence sequencing in
ADRs 0050/0051 and 0053. ADR-0054 proposes a resolution; it is not accepted or
implemented. `git diff --check` passed. No source, public API, test, schema or build
configuration changed, so builds/sanitizers were not rerun for this proposal. The
M6.4 validation above remains historical evidence, not M6.5 completion evidence.
## M6.5a coupled resource codec - 2026-09-27
ADR-0054 accepted by the maintainer. Added resource envelope v1 without changing
nested terrain/field formats or existing simulation goldens. Rates are still disabled;
source configuration/residuals and biology join the later M6.5c fixture envelope.
Executed Windows checks:
- Release check: 52/52 CTests passed with NDEBUG.
- Analysis check: clang-tidy/warnings-as-errors build and 52/52 CTests passed.
- ASan check: 52/52 CTests passed.
- After tightening restore to require all mapped chunks already in the snapshot,
Release and analysis rebuilt and environment-transfers-contracts passed again;
the ASan full run includes this final adjustment.
- Release format target passed.
Codec tests cover empty history, 66 transfers crossing the retry-ring boundary,
canonical restored bytes, exact/conflicting/expired retries, equivalent continuation,
all truncation lengths, checksum errors, recomputed-checksum invalid version/counts/
amounts, identity mismatch and each budget. Initial compilation caught missing codec
error enum entries; corrected before the passing runs above.
No Linux/UBSan/TSan, disk crash, allocation-failure injection or cluster validation
was performed locally. Buffer round trips are not durable publication. M6.5b-d remain.
## M6.5b process contract - 2026-09-27
Documentation-only segment: accepted ADR-0055 and ENVIRONMENTAL_PROCESS.md define
versioned fixture conversions and the narrow process/world ownership boundary.
Inspected M2 producer/feeding implementations, their tests and ADRs, M3 metabolic
budget contract, and M6 resource APIs before specifying the adapter.
Executed independent PowerShell arithmetic checks for the documented fractional
capture/efficiency sequence, nutrient-limited growth and per-channel feeding split;
all passed. `git diff --check` passed. These are specification checks, not runtime
integration tests. No C++/CMake/schema change; builds, CTest and sanitizers were not
rerun for this documentation segment. M6.5c/d must supply executable evidence for
rollback, source exhaustion, temperature limits, lifecycle and restored replay.
## M6.5c persisted environmental community - 2026-09-27
Implemented the one-habitat/two-actor fixture and envelope v1, preserving source
identity/rate, residuals, last funded interval, lifecycle, resources and 64 exact
input/report receipts. Restore replays receipts from a historical anchor, validates
reports and compares the final state. Scalar candidate staging commits without
allocation; the world layer checks capacities and conserved totals before publication.
Executed Windows x64 checks:
- Release check: 53/53 CTests passed, NDEBUG enabled.
- Analysis check: clang-tidy/warnings-as-errors build and 53/53 CTests passed.
- ASan check: 53/53 CTests passed.
- After final tight-budget and continuous/restored comparison additions, analysis
ran all 53 tests again; Release and ASan rebuilt and the two environmental
transfer/community tests passed again.
- Release format check passed. No existing golden changed.
Coverage includes hand-calculated capture/growth/feeding, source disabled/unfunded,
frozen residuals, inclusive temperature gates, terminal producer, whole-tick rollback,
exact/conflicting/expired retries, eight deterministic input sequences through 90
ticks, continuous versus restored bytes/reports, every truncation length, corruption,
recomputed-checksum invalid config/anchor/report/count/state, identity and budgets.
M6.5d remains the broader adversarial, long-run and cost qualification gate.
The supplied Gitea log shows GCC 14 Release rejecting transfer_codec.cpp with
-Werror=free-nonheap-object in the vector reserve/push_back path. Changed that writer
and the new process writer to exactly sized buffers, checked writes and final cursor
validation. No warning suppression or resource format change. The next Gitea run must
confirm Linux: Docker's Linux engine was unavailable locally; WSL Ubuntu exists but
has no g++/g++-14, CMake or Ninja. No toolchain installation was attempted. No local
Linux/UBSan/TSan or cluster result is claimed.
Initial native static analysis flagged an unspecified enum base and integer widening
in a size constant; both corrected before the passing analysis runs. No disk/host-loss
recovery, biological maintenance, reproduction or player/offline progression added.
## Environment-view optional guards - 2026-09-27
The supplied Linux clang-tidy-19 log rejected five optional field accesses in
`tests/environment_views_tests.cpp` with bugprone-unchecked-optional-access. Three
had no presence check; two relied on CHECK, whose throwing helper is not recognized
as a control-flow guard by this analysis. All five now use required_field, which
explicitly throws on absence before returning a reference. The rule remains enabled;
production code, formats, expected values and NDEBUG behavior are unchanged.
Executed native analysis, Release and ASan builds plus the targeted
`environment-views-contracts` CTest in each preset; all passed. Release format check
and git diff --check passed. Full suites were not repeated for this test-only fix.
Linux clang-tidy/libstdc++ confirmation remains with the next Gitea run; native
analysis is not claimed as a reproduction of the Linux diagnostics.
## M6.5d environmental qualification - 2026-09-27
Added the environmental-qualification CTest and optional `--long` driver. It tests
admission/capacity/overflow boundaries, each missing recipe resource, efficiency
endpoints, consumer temperature gates, recomputed-checksum mutations, a u64-max
trace and deterministic paired continuous/restored simulations. No runtime change.
See [ENVIRONMENTAL_QUALIFICATION.md](ENVIRONMENTAL_QUALIFICATION.md) for detailed
coverage, exact commands, long-run output, machine and measured memory/timing.
Executed Windows checks:
- Release configure/build/check: 54/54 CTests passed.
- Analysis configure/build/check: clang-tidy/warnings-as-errors and 54/54 passed.
- ASan configure/build/check: 54/54 passed, including tick exhaustion.
- After adding the final tick-exhaustion case, Release and analysis rebuilt and
environmental-qualification passed again. A constant-expression integer-width
diagnostic in the new test was corrected before that final analysis run.
- Release format and git diff --check passed.
- Final Release `evo-environmental-qualification.exe --long`: 16 x 65536 logical
ticks and paired continuous branches passed; 2097152 advances, 329231 bites,
matching snapshots and conserved totals. Reported fingerprint 93a9f9b296caac36.
Native M6.5 qualification is complete; no new Linux/UBSan/TSan or R730xd/cluster
run was available locally. CI must confirm the committed test on its toolchains.
No source-rule, schema, golden or deployment change.
## M6.6 retained corpse resources - 2026-09-27
Rules/envelope v2 now support trusted terminal facts for either fixture actor,
retained corpse inventory and bounded tagged decay. This is an in-memory process
checkpoint, not durable region publication or gameplay mortality. Rule v1 and its
canonical bytes are preserved; no golden was replaced. Config/input extensions
are rejected under v1 rather than silently omitted.
New tests cover pending death plus feeding rollback, terminal actor rejection,
shared per-channel caps in ID order, exact transfer destinations, partial decay,
zero caps, water headroom, restored duplicate/conflicting/stale requests, permanent
terminal state, consumer-corpse nutrients feeding a surviving producer next tick,
every v2 byte truncation, and checksum-repaired invalid lifecycle/history fields.
Eight 90-tick cases compare retained balances against independent cumulative-quota
formulas, with restore at ticks 17, 65 and 90 and exact conservation every tick.
Commands: tools/windows-check.ps1 -Preset release/analysis/asan -Action check;
-Preset release -Action format; git diff --check. The initial suites found the
older negative test's rules=2 was now valid; it now rejects unsupported rules=3.
Final results are recorded below after all processes complete.
Release long qualification again ran 16 x 65536 logical ticks, 2097152 paired
advances, and produced unchanged v1 fingerprint 93a9f9b296caac36. Snapshot size
remains 30615 bytes in that v1 case. Owner size is now 20152 bytes (formerly13400),
including bounded extension receipts; budget admission uses sizeof dynamically.
No new Linux/UBSan/TSan or R730xd measurement is claimed. Linux CI remains a
separate verification step; local Docker Linux engine/toolchain is unavailable.
Final native validation: Release 54/54 (3.66s), clang-tidy analysis 54/54 (13.37s),
ASan 54/54 (46.72s), format check and git diff --check passed. M6.6 is natively
qualified; the next gate is M6.7 interval scheduling. No deployment was changed.
## M6.7 local environmental intervals - 2026-09-27
ADR-0057 and ENVIRONMENTAL_SCHEDULING.md define the stateless interval adapter.
Admission validates world/frame, fixed quantum, completed process/field revision,
ordered endpoints, at most 64 ticks and an exact input count. Each process tick is
atomic; execution failure explicitly returns the committed prefix and failing tick.
No new scheduler state, snapshot format, biology rule, dependency or deployment.
Tests compare 130-tick streams under rules 1 and 2 against direct execution using
batch widths 1/3/17/64, flattened report equality and final canonical snapshots.
Every batch pauses without mutation and restores before continuing. Admission
cases include foreign/stale boundaries, quantum mismatch, missing/extra input,
65-tick backlog, u64-max backlog and reversed intervals. Failure at first, middle
and final tick preserves precisely the committed prefix, then restores and resumes
to the uninterrupted result. The existing valid u64-max checkpoint now also tests
scheduler pause and wrap refusal. Partial success is intentionally not a whole-batch
transaction; callers must inspect failure and committed count.
Executed tools/windows-check.ps1 -Preset release/analysis/asan -Action check:
Release 54/54 (6.44s); clang-tidy analysis 54/54 (14.20s); ASan 54/54.
After making the optional failure access explicitly guarded, rebuilt Release and
reran both environmental CTests: 2/2 passed. Analysis/ASan include that final guard.
Release format-check and git diff --check passed. Tests retain NDEBUG checks.
Linux/UBSan/TSan CI confirmation remains pending; native validation does not claim
a Linux or target-server run. Next is M6.8 integrated ecological qualification.
## M6.8 integrated environmental qualification - 2026-09-27
Extended the existing qualification executable; no runtime, schema or golden
changed. Complete evidence and reproduction commands are in
[ENVIRONMENTAL_SLICE_QUALIFICATION.md](ENVIRONMENTAL_SLICE_QUALIFICATION.md).
- tools/windows-check.ps1 -Preset release -Action check: 54/54 passed (4.06s).
- -Preset analysis -Action check: clang-tidy and 54/54 passed (18.57s).
- -Preset asan -Action check: 54/54 passed (60.83s); qualification itself 12.83s.
- -Preset release -Action format and git diff --check passed.
- Release evo-environmental-qualification.exe --long passed: existing v1 long
fingerprint unchanged, plus 5120 v2 episodes/655360 logical ticks/1310720 paired
advances. Full-cycle assertions, checked E/C/N/P/other totals, separate water,
residual ranges, stable resource residency and restored retry equality passed.
- Three integrated execution/snapshot/restore cost samples were collected on the
developer machine. Concurrent analysis/ASan work is disclosed in the report.
M6 is natively qualified for the bounded synthetic scope. Linux CI/UBSan/TSan,
R730xd measurements and persistent server recovery are not claimed. The local
Linux toolchain/Docker engine remains unavailable. Next: M7.1 durability contract.
## M7.1 region durability contract - 2026-09-27
Documentation-only gate: ADR-0058, REGION_DURABILITY.md and logical region-commit
schema. Reviewed existing M6 process/resource codecs and scheduling contracts,
LineageArchive public API and retention tests, individual genome codec, separate
population/particle interfaces, persistence architecture and ADRs 0013/0016.
The audit identifies complete M6 state and missing archive/binding/host state.
Full checkpoint batches are selected; semantic events alone are not recovery.
Verified new contract relative links (2 resolved) and all eight ADR sections;
git diff --check passed. Reviewed the crash matrix against upload, fenced DB
publication, ambiguous commit, ACK, recovery and clone boundaries. Those rows
are future executable acceptance tests, not fault-injection results from this gate.
No C++/CMake/schema codec implementation changed, so build/sanitizer suites were
not rerun. M6.8's recorded native checks remain the latest executable evidence.
No database, object storage, Kubernetes access or deployment was performed.
M7.1 design gate is complete. M7.2 must evaluate/pin dependencies and exercise real
migrations; M7.4 must implement additional ancestry state closure before M7.9 can
claim lineage recovery. No retained-history or World0 requirement was relaxed.
## M7.2 PostgreSQL bootstrap - 2026-09-28
- Windows Release configure/build/CTest passed 54/54 (24.17s); format passed.
- Built checksum-verified PostgreSQL/libpq 18.6 from source in WSL Ubuntu.
- GCC14 Release, clang19 analysis (clang-tidy warnings-as-errors), and clang19
ASan/UBSan optional adapter builds and real private-database integration passed.
Tests cover parameter separation/NULL, rollback, duplicate SQLSTATE, u64 range
and fractional rejection, runtime-role denial, deadline closure, result bounds,
backend termination, clean/repeated migrations, failed migration rollback,
unknown versions and changed checksums. Expected SQL errors are negative tests.
- Fixed analysis enum-size warning and migration rejection exit status during
qualification; reran all three final configurations successfully.
- clang-format19 dry-run passed for all three new C++ files; bash -n passed for
preparation, migration, integration and publish scripts. ldd found all psql
dependencies: libpq, libc/libm and OpenSSL (matching tooling image packages).
- Docker Linux engine is unavailable locally. Container build/smoke and Gitea
remote checks are configured, not claimed as executed. No cluster, external DB,
TLS endpoint or Kubernetes Secret was accessed. TSan was not rerun: adapter has
one caller-owned I/O thread and introduces no concurrent execution.
The adapter is separate from the headless kernel. Region fencing, publication,
ambiguous durable commit reconciliation and crash recovery are later M7 gates.
Only clean-to-v1 and exact-v1 restart are supported migration paths at this gate.
## M7.3 fenced region ownership - 2026-09-28
- Windows MSVC Release configure/build/CTest passed 55/55, including the new
portable `region-lease-contracts` test. Repository clang-format 19 check passed.
- WSL Ubuntu (GCC 14.2.0; clang 19.1.1; PostgreSQL/libpq 18.6 source build):
GCC14 Release, clang19 analysis (clang-tidy warnings-as-errors) and clang19
ASan/UBSan builds of `evo-postgres-tests` each passed `tools/ci/postgres-check.sh`
against a fresh private cluster. The first analysis build flagged two
unchecked optional accesses and two unreserved vectors. These were fixed and
all three configurations were rerun after formatting.
- Real-database coverage, as the runtime login rather than SET ROLE:
- Denied (42501): table read/write, provisioning, the internal check function,
and object creation. A hostile search_path is ignored.
- Busy, forged-holder, stale-epoch and released-lease rejections; epoch
non-reuse.
- 1..64 tick delta; predecessor mismatch; exact-retry detection via the head.
- 1 s database-time expiry and successor acquisition; paused stale writer fenced
locally and by the database.
- A writer queued 1.3 s on the lease row lock is rejected after expiry.
- A 100 ms renewal deadline under lock (unknown) fences locally before a
successor acquires.
- 64 two-thread acquisition rounds with exactly one winner and consecutive
epochs; 32 racing head advances with exactly one winner.
- Epoch and sequence exhaustion.
- Migrations tested: clean->v2, repeated v2 no-op, v1->v2 upgrade, and failed
0001 rollback. Superuser and missing runtime roles are refused with no schema
left. Unknown v3, changed v2 checksum and changed v1 checksum are refused.
- Mutation sensitivity (scratch copies only): replacing post-lock
`clock_timestamp()` with `now()`, removing the lease check from head advance,
and disabling the busy check each made the suite fail at the expected test.
- Local Unix-socket latency, 200 samples per operation in each final run:
- p50 about 0.61-0.74 ms and p99 about 0.8-1.15 ms for acquire, renew,
advance and release.
- Max up to 2.8 ms under analysis/ASan builds.
These are local numbers, not remote TLS or cluster measurements.
- Not executed: Docker/Gitea CI, remote TLS endpoints, Kubernetes Secrets, real pod
kill, real network partition (a lock wait plus client deadline stands in for
it), VM freeze and TSan. The contention threads each own a separate
connection; libpq is uninstrumented.
### M7.3 review fixes - 2026-09-28
- P1: an attribute-only runtime role check let a login inheriting
`pg_write_all_data` update `evolution.region_leases` directly. The CI script
reproduces that UPDATE first. `apply.sql` now audits effective privileges on
every run (the v2 migration file and its checksum are unchanged). Rejected, each
with a rollback:
- a runtime group inheriting `pg_write_all_data` on a fresh database (no
schema left);
- that grant on the login, and on the group WITH INHERIT FALSE;
- a PUBLIC table grant, EXECUTE on the internal check function, and a
CREATEDB login.
The restored configuration passes again.
- P2: `valid(LeasePolicy)` now bounds each term by the duration before any
arithmetic. `acquired` validates the duration before computing a deadline.
`renewal_unknown` ignores durations above 60 s and fences on non-positive ones.
New tests use `milliseconds::max()` and `min()` for every field. Built against
the previous source under clang19 UBSan, they reproduce the reported signed
overflow (`9223372036854775807 + 2000`).
- Reran Windows CTest 55/55, the repository format check, and GCC14 Release,
clang-tidy analysis and ASan/UBSan private-PostgreSQL suites: all passed.
## M7.4a checkpoint publication - 2026-09-28
- Windows MSVC Release configure/build/CTest passed 56/56, including the new
portable `region-checkpoint-contracts`. It covers:
- golden byte offsets and round trips;
- malformed and unsupported manifests, including u32 values above 255 that
must not truncate onto supported enums;
- chain rules;
- payload identity, tick, quantum, corruption and budget checks;
- 24-tick continuation equality after restore.
The repository clang-format 19 check passed.
- WSL Ubuntu, with GCC 14.2.0, clang 19.1.1, PostgreSQL/libpq 18.6 and OpenSSL
3.0.13-0ubuntu3.15, running as uid 1000 so permission faults were really
exercised. GCC14 Release, clang19 analysis (clang-tidy warnings-as-errors) and
clang19 ASan/UBSan each passed `tools/ci/postgres-check.sh`.
- Analysis initially flagged `performance-enum-size` on the u32-based
profile/durability enums. They are now u8 with explicit decoding of values
above 255 as unknown. All three configurations were rerun.
- New real-database/filesystem coverage:
- SHA-256 vectors (NIST "", "abc" and the 448-bit message) and agreement with
PostgreSQL `sha256()`.
- Object store: rejects root symlinks, creates 0700 namespaces, puts are
idempotent, never overwrites differing content, ignores temporary leftovers,
and detects byte flips, symlinks, truncation and growth. Size bounds and
write-permission failure are also covered.
- Genesis provisioning and duplicate refusal; three batches (16/64/16 ticks);
a 65-tick batch refused before upload.
- Recovery from a new store handle is byte-identical and continues 44 ticks
identically to an uninterrupted reference.
- The build allowlist is enforced by the host and by the database.
- Upload failure leaves the head unchanged. A fenced publication leaves a
verified orphan, and recovery ignores it.
- Missing and corrupt payloads quarantine; restoring the object recovers.
- Owner edits to a manifest (23514) and head repointing (23503) are refused.
With the constraint dropped, recovery still detects the edit, and the
constraint is restored afterwards.
- Unknown regions are never created.
- Lease tests were ported to manifest publication. They additionally cover
malformed manifests, a quantum change, an unretained build, a future rules
version and a globally reused commit ID.
- Migrations: clean→v3, repeated no-op, v1→v3. v2 with existing heads fails
closed and stays at v2. The audit rejects EXECUTE on provisioning and SELECT on
commits. Tampered v2/v3 checksums and an unknown v4 are refused.
- Mutations, run on scratch copies with the source restored afterwards: removing
the predecessor-digest check, the lease check, the version/quantum continuity
check or the retained-build check in SQL, or the host manifest re-digest, each
failed at the expected assertion.
- Local fixture measurements: payloads of 22775 and 37367 bytes.
- Release publication took 3.6-4.8 ms per batch; analysis 5.5-6.7 ms; ASan
about 10.5 ms.
- Manifest publish p50 was about 0.74-0.91 ms.
These are local socket and local disk only.
- Not executed: Docker/Gitea CI, a remote object store or PVC, power-loss fsync
semantics, read-back verification fault injection, process kill, and TSan.
## M7.4b lineage checkpoint profile - 2026-09-28
- Windows MSVC Release CTest passed 57/57, including the new `lineage-region-contracts`
and extended lineage and region-checkpoint contracts. The repository clang-format
19 check passed.
- WSL Ubuntu (GCC 14.2.0, clang 19.1.1, PostgreSQL 18.6, OpenSSL 3.0.13):
- The Linux `checks` build set (separate build dirs) passed CTest 57/57 in each
of GCC14 Debug, GCC14 Release, clang19 analysis and clang19 ASan/UBSan.
- The PostgreSQL integration suite passed 2× Release, 2× analysis and 5×
ASan/UBSan.
- Archive codec coverage:
- canonical encoding regardless of admission order, including a child with a
smaller ID than its parent;
- checksum, magic, version, count cap, length and organism-count framing;
- genome damage, unknown genome, invalid alive/died forms, death before birth,
reordered organisms (non-canonical) and a child encoded before its parent;
- a full 4096-genome/65536-organism round trip at 2,543,648 bytes.
- `LineageRegion` coverage:
- closure refusals: missing, identical or dead binding; low high-waters; stray
living organism; future birth; archive death while the community is active;
- the death recorded at tick 20 in the same advance; a refused tick changes
nothing;
- restore plus 36 continuation ticks identical to the original;
- container, binding, envelope and archive damage refused.
- Profile 2 coverage:
- the 4,268,144-byte cap is asserted in C++ and enforced by SQL (the maximum is
accepted, +1 refused); the profile-1 1 MiB+1 is refused;
- mismatched profile/history is EVL06;
- lineage publish (death included), recover and continue, and a second
publication;
- the environmental path refuses a lineage predecessor before upload;
- the database refuses a profile switch;
- a host of the other profile reports unsupported.
- Migrations: clean→v4; v1→v4; v3 with a committed genesis upgraded to v4 with
the commit retained; v2 heads still block. Tampered v2/v3/v4 checksums and an
unknown v5 are refused.
- Mutations (restored afterwards), each failing at the expected line: removing
the SQL profile/history continuity check (store test 392), the living-set
closure check (lineage region 58), or death recording (lineage region 78).
- Flake found and fixed. The ASan PostgreSQL suite failed intermittently: 1 of 3,
then 1 of 5 runs, at the queued-writer and expired-renewal assertions.
- Cause: a 4-minute probe measured WSL2 stepping CLOCK_REALTIME back 470-596 ms
every ~32 s (3.66 s total, roughly 1.5% versus CLOCK_MONOTONIC).
- PostgreSQL `clock_timestamp()` follows that wall clock, so a 1.2 s local sleep
could be under 1 s of database time.
- Tests now poll `clock_timestamp() >= expires_at`, holding the row lock where
relevant. Afterwards the suite ran 5/5 under ASan.
- The fencing logic itself was unchanged; the requirement is recorded in
ADR-0060.
- Local measurements, Release:
- manifest publish p50 0.72 ms and p99 1.24 ms;
- checkpoint publication 3.5-5.0 ms for 23-37 KiB payloads.
- Not executed: Docker/Gitea CI, remote storage, process kill, reproduction within
a batch (the fixture has none), durable history rows (M7.5) and TSan.
## M7.5 durable batch outcomes - 2026-09-28
- Windows MSVC Release CTest passed 58/58, including the new `region-batch-contracts`:
- recorder sealing at a death or at 64 ticks, early flush, confirm matching, and
a refused tick recording nothing;
- trace round trip, size (19,552 bytes at 64 ticks), range, contiguity and
header binding;
- decision order, uniqueness, range, and refusal of unknown values.
The manifest v2 golden layout, v1 decoding and batch-evidence structural rules
were also covered.
- WSL Linux: the `checks` build set (GCC14 Debug and Release, clang19 analysis,
clang19 ASan/UBSan) passed CTest 58/58 each, and the repository format check
passed.
- The first PostgreSQL clang-tidy build flagged a returned const-reference
parameter and two unchecked optional accesses; all three were fixed. Final
PostgreSQL suites: Release 2 runs, analysis 2, ASan/UBSan 4, all passing.
- Real-database coverage:
- Durable death at tick 20, with decisions and outbox rows in the same commit.
- Outbox: redelivered, acknowledged true then false, then empty; invalid limit
and unknown rows refused.
- A lost response retried with identical bytes returns the same receipt and
records no second death; resolution says committed.
- A changed manifest or changed decisions conflict (EVL07/22023). Other regions
and unknown IDs resolve as conflict and absent.
- A repeated death is EVL08. A forged same-size block, a partial block and a
reordered block are 22023. A gap is EVL03. None of these moved the head or
decisions.
- A 300 ms client timeout while queued on the lease row became unknown; once the
lock cleared the server committed, resolution returned committed and the retry
was idempotent.
- A terminated orphan request resolved as absent, and the same bytes were
retried and committed.
- A missing predecessor payload is `predecessor`. State advanced outside the
batch fails replay. Batches from the wrong predecessor and unsealed batches are
refused before upload.
- Recovery reconciles decisions. An owner-inserted death the state does not
show quarantines with `history`, and cannot be deleted (EVL09).
- Commit and decision UPDATE and DELETE are EVL09 even for the owner; the
defense-in-depth test disables and restores the trigger explicitly.
- Lineage profile: a death is recorded with the archived organism id and
reconciled against the archive.
- Migrations: clean→v5, v1→v5 and v3→v5 with an existing v1-manifest commit;
tampered checksums for v2 through v5 and an unknown v6 are refused.
- Mutations (scratch SQL copies; host source restored and rerun green), each
caught: removing idempotent retry, outcome finality, the decision-digest check,
decision recording, or recovery reconciliation.
- Local measurements, including predecessor replay: Release 6.6-8.1 ms per batch;
analysis about 20-23 ms; ASan about 19-21 ms. Manifest publish p50 about
0.75-0.86 ms.
- Not executed: Docker/Gitea CI, remote object store, process kill, a broker
consumer, an injected failure between commit insert and statement end, and TSan.
### M7.5 review fixes - 2026-09-28
- P1 object durability:
- `put()` now re-fsyncs the file and the directory before reusing an existing
object, on both the found-existing and the lost link-race paths.
- Namespace opens always sync the parent directory.
- A test-only `ObjectStoreFaults` hook injects directory-sync failures and
counts successful syncs.
- Tests: a namespace-creation sync failure, then a reopen still failing, then
success with exactly 2 syncs. An install sync failure leaves the object
visible but unacknowledged, three retries while syncs fail all return io with
no extra syncs, then success with +1 sync; every reuse adds a sync.
- Publication with 1, 1 and 2 injected failures returns `upload`/io and sends
no manifest (head stays 0), then publishes once syncs succeed.
- Reverting the reuse sync or the always-sync namespace behavior makes these
tests fail.
- P1 pre-v5 history:
- `pre_v5_guard.sql` (EVL10) runs before 0005 in every upgrade path. 0005 and
its checksum are unchanged.
- CI builds a real v4 database: v1 genesis, a lease, and a sequence-1
publication through the v4 function. The v5 upgrade is refused with the
expected message; history stays 1..4, both commits remain, and no decision
table is created. Replacing the guard with a no-op makes that check fail.
- A committed death published without any decision (the state an accepted
legacy upgrade would produce) recovers as `history`, quarantined and never
served.
- Reran GCC14 Release (3 runs), clang-tidy analysis (1) and ASan/UBSan (2)
PostgreSQL suites, all passing, and the repository format check. The changes
are Linux-only persistence code, SQL and tests; the portable targets and
Windows are unchanged.
## M7.6 region host - 2026-09-28
- Windows MSVC Release CTest passed 58/58; this change touches only the Linux
PostgreSQL CMake section and Linux sources. The repository clang-format 19 check
passed.
- WSL Linux: GCC14 Release, clang19 analysis (clang-tidy) and clang19 ASan/UBSan
builds of `evo-postgres-tests` and `evo-server` passed the full PostgreSQL suite.
Final runs: Release 3, analysis 1, ASan 3 across both verification passes.
- The first analysis build flagged unchecked optional accesses and ignored release
results in the host and server, plus two test issues. These were fixed with
explicit local checks rather than suppressions, then rebuilt and rerun in all
three configurations.
- Host scenarios (real PostgreSQL, scaled scheduling clock, 4 s test lease):
- Admission: not_ready before recovery, 64-input cap (full), duplicate and
invalid refusals, byte accounting.
- Ticks 1–19 are provisional with no acks. Tick 20's death pauses and commits
in the same step: 20 acks and one outcome (consumer, organism 11, sequence 1).
The next batch flushed by interval at tick 41.
- Starvation: debt reached 10 without invented ticks, then catch-up of 8 per step.
- Storage outage: the sealed batch waits with nothing acknowledged and debt
growing, then commits when storage recovers.
- Lease row locked: the commit becomes unknown, the host is not ready and
refuses admission, and resolution adopts the receipt after release.
- Renewals blocked past the local deadline: fence, all 8 unconfirmed inputs
reported `fenced`, then re-acquisition at a newer epoch from the committed head.
- Drain (environmental profile): backlog dropped as drained, 6 pending ticks
committed, lease released, and a successor acquired immediately at epoch 2.
- Unrestorable build and foreign frame quarantine; an invalid configuration is
refused.
- Process tests (real `evo-server` via posix_spawn, stdin and stdout pipes):
- SIGTERM after 80 inputs: exit 0, 80 acks, head 80 at sequence 8, lease
released, readiness file removed, max_queued 64. The outcome line followed
tick 20's final ack.
- SIGKILL during running: every acked tick ≤ the committed head (11/11) and the
lease was left held. The successor was ready after 3.94–3.97 s at epoch 2 with
committed=11, then drained cleanly.
- Mutations (restored and rerun green), each caught: acknowledging at simulation
time, not committing a sealed death, reporting ready while resolving, and
draining without releasing the lease.
- Not executed: Docker/Gitea CI, TSan (the host is single-threaded; libpq is
uninstrumented), remote PostgreSQL/TLS, a production 20 s lease kill test,
durable client deduplication, and a network gateway.
## M7.7 recovery before readiness - 2026-09-28
- WSL Linux: GCC14 Release, clang19 analysis (clang-tidy, clean on the first
build) and clang19 ASan/UBSan builds of `evo-postgres-tests` and `evo-server`
passed the full PostgreSQL suite, including the M7.7 recovery tests. Runs:
Release 2, analysis 1, ASan 2. The repository format check passed after
reordering one fixture include.
- Windows was not rerun. This change touches only Linux persistence and host
code, SQL, tests and docs; M7.6's Windows run (58/58) covers the portable
targets, which are unchanged.
- Kill matrix (real SIGKILL via `--crash-at`, inputs 1..40 with a death at tick
20, 5 s flush):
| Crash point | Head (seq/tick) | Acked up to | Decisions | Objects |
|---|---|---|---|---|
| before-transaction:1 | 0/0 | none | 0 | 3 (verified orphans) |
| before-upload:2 | 1/20 | 20 | 1 | 3 |
| before-transaction:2 | 1/20 | 20 | 1 | 5 (orphans) |
| after-commit:2 | 2/40 | 20 (lost acks are the safe direction) | 1 | 5 |
| after-ack:2 | 2/40 | 40 | 1 | 5 |
The lease stayed held after every kill. Each successor became ready at epoch 2
from the committed head (with tail verification), then drained and released.
- Resumed stale owner: the owner was SIGSTOPped for 4.5 s (4 s lease). The
successor became ready at epoch 2 with committed=10 and processed ticks 11–20.
After SIGCONT the old owner printed `ready 0`, rejected new inputs as
`not_ready`, acknowledged nothing beyond tick 10, and exited 0 on SIGTERM
without touching the successor's chain.
- Damaged heads: a flipped head payload byte and a missing head trace each made
`evo-server` exit 4 (quarantined) without ever printing `ready 1`, and the head
was unchanged. After repair a successor became ready at epoch 4.
- Tail verification (in-process): a missing or corrupt head trace and a missing
predecessor payload are `tail`, and recovery succeeds again after repair. A
published trace of different inputs over the same range (a host bypass the
database cannot detect) is `tail`.
- Mutations (restored and rerun green): skipping tail verification made the
quarantine test time out because the server became ready. Dropping replay
byte-equality failed the forged-trace test (line 262).
- Not executed: Docker/Gitea CI, the production 20 s lease, host loss or VM
freeze, multi-host network partitions, or verification of links older than
the head.
## M7.8 operations endpoints - 2026-09-28
- Windows MSVC Release CTest passed 59/59, including the new portable
`ops-contracts`: exposition format, HELP/TYPE per family, the fixed 30-series
set, bounded labels, one-hot phase and non-wrapping lag. The exposition is
3485 bytes typically and 3836 at worst-case values (limit 8192).
- WSL Linux: the `checks` build set (GCC14 Debug and Release, clang19 analysis,
clang19 ASan/UBSan) passed CTest 59/59 each.
- The PostgreSQL suites (GCC Release, clang-tidy analysis, ASan/UBSan) passed
after fixing the ops test's optional-access and reserve findings. The earlier
PostgreSQL analysis runs used a stale binary and are not counted. The
repository format check passed after formatting the server.
- Process probe tests (real `evo-server`, loopback, ephemeral port):
- Serving: `/livez` 200 `live`, `/readyz` 200 `ready`. Metrics showed ready 1,
epoch 1, lease remaining above 0, capacities 64/65536 and the running phase.
After 30 inputs: committed tick 30, lag 0, 30 acknowledgements, checkpoint
age under 60 s.
- HTTP bounds: POST 405, unknown path and query string 404, malformed 400,
over 4 KiB 431. Eight idle connections do not block probes; a ninth is
closed within 500 ms; idle ones close at their deadline.
- Database unreachable: `/livez` 200, `/readyz` 503 `not ready: acquiring`,
ready 0, store failures above 0, never `ready 1`, exit 0 on SIGTERM.
- Stale process (SIGSTOP 4.5 s, then SIGCONT): `/livez` 200, `/readyz` 503,
ready 0, fences_total 1, lease remaining 0. The successor answers 200 at
epoch 2.
- Mutations (restored and rerun green): readiness that returns 200 whenever live
was caught. Raising the connection bound to 1000 was not caught at first,
because the check did not distinguish refusal from deadline expiry. The test
now requires the refusal within 500 ms and catches it.
- Image: the Docker engine was not running locally (Docker Desktop stopped), so
the `server-runtime` build and `tools/ci/server-image-smoke.sh` have not
executed. A filesystem emulation (binary plus the pinned libpq, `env -i`)
showed usage exit 2 with no settings and all libraries resolved. It also found
and fixed a bug: without arguments the server had reached object-store opening
instead of refusing with usage. The binary keeps a build-tree RUNPATH, which is
harmless in the image.
- The manifest example parses (Deployment and NetworkPolicy, three probes) and
the workflow YAML parses; `bash -n` passed for the smoke script.
- Not executed: the container build and smoke, the Gitea job, cluster apply,
Prometheus scraping, and TSan.
## M7.9a region clones and retained history - 2026-09-29
- Windows MSVC Release CTest passed 59/59. `lineage-region-contracts` now covers
rebinding: only the new identity restores, continuation matches, and rebinding
back is byte-exact. `region-checkpoint-contracts` covers the clone-genesis
manifest rules.
- WSL Linux `checks` set (GCC14 Debug and Release, clang19 analysis, clang19
ASan/UBSan) passed CTest 59/59 each. The format check passed.
- The PostgreSQL suites passed in GCC Release, clang-tidy analysis and ASan/UBSan.
These runs used a second database, `evo_clone`, with its own runtime group and
login. clang-tidy first flagged branch-clone in the tool's option parsing and
optional access in the new tests; both were fixed, then the suites were rebuilt
and rerun.
- Harness additions:
- A populated v6 database (a v2 genesis) upgrades in place to v7. The replaced
head constraint is `region_heads_published_tick`, and the v7 parser reads the
stored genesis.
- Upgrade paths now end at 1..7.
- Version and checksum drift checks now include v7.
- Clone tests (real `evo-server` and `evo-world-clone` processes):
- Trajectory: world 7 was run to tick 30, with a death at tick 20, then cloned
to world 8 in `evo_clone`. The clone recovered under the clone runtime at
tick 30, with the inherited death and retained ancestry, and the audit found
one clone link naming the source cut. Both worlds then ran inputs 31..90; after
rebinding the clone to world 7, the snapshots were byte-identical.
- Isolation:
- A same-world target was refused as `identity`.
- A world that already holds an original region was refused (22023).
- A forged origin was refused (22023).
- The runtime role cannot provision a clone (42501).
- A clone world refuses an original region.
- Origins are immutable (EVL09).
- The source session is read-only (25006, even through SECURITY DEFINER).
- A clone lease publishing a successor to the source is fenced.
- The clone runtime login cannot reach the source's functions (42501).
- The source head and decisions were unchanged afterwards.
- Failure: a SIGKILL before the target transaction left the source unchanged,
the target unprovisioned with one orphan object, and a target server that
quarantined (exit 4) without becoming ready. The tool refused a same
database or same user (exit 2) and a same world (exit 4). The retry
succeeded and reused the same object. A source with a damaged head payload
was refused (exit 3), and its target stayed untouched.
- Retention: an owner was SIGKILLed after acknowledging tick 100, and its
successor continued to tick 200. The audit verified every link, and the tool
printed the same result. A lower `max_commits` returned `bounds`. Removing the
first link's trace made the audit report `tail` at sequence 1 while head
recovery still succeeded; restoring the trace made the audit pass again.
- Environmental profile: a clone inherited the tick-20 death. Moving that
durable row to tick 19 (triggers disabled as break-glass) made recovery
report `history` and the audit report `history` at sequence 0. Moving it back
made recovery pass.
- Mutations (each restored and rebuilt afterwards). Each was caught by its
intended check:
- dropping the isolated-world refusal;
- dropping the origin check;
- a read-write source session;
- skipping the inherited-outcome digest;
- skipping the audit's per-link tail verification.
The first attempt at the inherited-outcome mutation did not compile, so its
stale-binary run was discarded and the mutation was rerun.
- Not executed: clones across separate PostgreSQL clusters or hosts, Docker, the
Gitea CI, and cluster runs.
## M7.9b upgrade barrier and durability qualification - 2026-09-29
- Every M7.9b change is in the Linux PostgreSQL build or in docs. Windows was not
rerun; its 59/59 from M7.9a still covers the portable code, which is unchanged.
- WSL Linux `checks` set (GCC14 Debug and Release, clang19 analysis, clang19
ASan/UBSan) passed CTest 59/59 each. The format check passed.
- The PostgreSQL suites passed in GCC Release, clang-tidy analysis and ASan/UBSan.
After the mutation run, one host test was corrected: `unrestorable_head_quarantines`
now retains its build, so the quarantine it checks again comes from recovery
rather than acquisition. Release and analysis were rebuilt and rerun green after
that change. The ASan build compiled with it, but its full run predates it.
- Harness: upgrade paths now end at 1..8. The populated v6 database upgrades to v8.
Version and checksum drift checks include v8.
- Migration and barrier test (database `evo_online`, created at v7 from the
checked-in files):
- A pre-v8 writer with a build-less lease published ticks 1..20 (with the
death).
- An `apply.sql` run with an invalid v8 checksum failed. It left history 1..7,
no `retired_at` column and no four-argument acquire. The writer renewed and
published ticks 21..40.
- The real migration then reached v8 while the writer held its lease, and the
writer published ticks 41..60.
- Barrier refusals: the runtime cannot call it (42501), and retiring the last
current build is refused (22023).
- After the barrier (1 lease fenced), the old writer's publication and renewal
were fenced. Build-less acquisition returned EVL06, and acquisition with b1
returned unsupported.
- A b2 lease publishing a link that names b1 was refused as unsupported.
- A b2 `evo-server` became ready at epoch 3 from committed tick 60 and
acknowledged through tick 90. Its final head is built by b2.
- The audit passes with both builds retained, and reports unsupported at
sequence 0 with only b2.
- A b1 `evo-server` quarantined (exit 4) without becoming ready, and the head
was unchanged.
- Qualification: see engineering/DURABILITY_QUALIFICATION.md. SIGKILL to ready
took 19.5–19.6 s under the production lease in all three builds (the 60 s
target is asserted). Acknowledgement latency p50 was about 549 ms and p99
about 1048 ms. No acknowledged input was lost, and one durable death was
recorded across five owners.
- Mutations (each restored and rebuilt). Each was caught:
- no retired-build trigger;
- no build check in four-argument acquisition;
- no barrier check in build-less acquisition;
- a barrier that does not fence holders;
- a host that retries an unsupported acquisition instead of quarantining.
The migration test caught this one after the host-test correction.
- Not executed: Docker, the Gitea CI, cluster runs, a separate database host, and
host or storage loss (M8).
## M7.6–M7.9 review fixes (schema v9) - 2026-09-29
Review `.cache/reviews/M7_6-M7_9_REVIEW.md` found three issues. All three are
fixed:
- **[P1] Build retirement could race acquisition.** Schema v9 applies one
upgrade-lock order across the barrier, acquisitions, lease checks, provisioning
and the retired-build trigger.
- **[P2] Clone-world classification could race.** Provisioning now takes a
per-world lock and checks classification after it.
- **[P2] The operations listener accepted without bound.** Each `serve()` now
makes at most 4 accept attempts.
Evidence:
- WSL Linux `checks` set (Debug, Release, analysis, ASan/UBSan): CTest 59/59
each. The format check passed.
- PostgreSQL ASan/UBSan: full suite passed. Analysis and Release were rerun green
after fixing two clang-tidy findings in the new tests. Upgrade paths now end at
1..9.
- Race tests (database `evo_race`): each holds a transaction open, confirms the
contender is blocked on a lock via `pg_stat_activity`, then commits.
- A build-less acquisition queued behind a barrier is refused (EVL06).
- A build-aware acquisition of the build being retired is refused, and the
region stays unheld.
- The next build acquires.
- Racing barriers for the last two current builds: one returns 22023 and one
build stays current.
- Different-source clones into one world, and an original region racing a
clone: the second is refused (22023).
- Same-source clones into different regions both succeed.
- The final origins are exactly `2/1<10, 4/1<10, 4/2<10`.
- Accept budget:
- With 16 queued connections, successive `serve()` calls give 4 open, then 8
open with 0 refused, then 8 open with 4 refused.
- Refusals never exceed 4 per call under replenished connections.
- The first accepted probe got 200.
- A real server acknowledged 60 ticks under continuous connect/close churn
without losing readiness.
- Mutations, each reverting one fix (restored and rebuilt). Each was caught:
- an unlocked build-aware acquire;
- an unlocked legacy acquire;
- an unlocked barrier;
- no per-world lock;
- an accept budget of 1000.
- Qualification was unchanged: SIGKILL to ready took 19.55–19.60 s.
- Windows was not rerun. All changes are Linux PostgreSQL, server or doc files.
## M8.1a inventory preparation - 2026-09-29
- No cluster access was available: there is no kubeconfig on the workstation.
Nothing in the cluster was read or changed, and the `aevolution-db` Secret was
not accessed. Only the key names the maintainer supplied are recorded.
- `tools/inventory/collect-cluster.sh` passed `bash -n`. It has not been run
against a cluster. It only reads, and it lists Secrets through the server-side
table, so no data is fetched.
- `deploy/kubernetes/inventory/disk-probe.yaml.example` parses as a PVC,
NetworkPolicy and Job. Its script was run locally in WSL2 on ext4 as a
baseline:
- 4 KiB O_DSYNC writes: mean 1169 µs;
- 40 KiB object publication: mean 4646 µs, max 7983 µs;
- 256 MiB fdatasync: 365 ms.
It has not been applied in any cluster.
- The qualification now records the server's resource profile from /proc. Over
three owners in Release: peak RSS 9,940 KiB, and 60 ms CPU over 16.3 s of
serving. The full PostgreSQL Release suite passed with it. The analysis build
compiled clean, and the format check passed.
## M8.1b cluster inventory - 2026-09-29
- The maintainer ran `tools/inventory/collect-cluster.sh`. It made read requests
only, and listed Secrets as name, type and key count. The output is in
`C:\tmp\cluster` on the workstation and is not committed. The results are
summarized in docs/infrastructure/INVENTORY.md:
- RKE2 v1.32.5 with one untainted control plane (6 vCPU / 11.7 GiB) and three
workers (8 vCPU / 31.3 GiB).
- No kubelet reservations, quotas or LimitRanges.
- Storage is `local-path` only (reclaim Delete, no CSI, no snapshots).
- Canal/Calico, ingress-nginx, MetalLB and cert-manager with Let's Encrypt.
- `aevolution-dev` has no automated sync.
- No PostgreSQL 18.6 in the cluster (Gitea runs 17.6).
- The collector ran as cluster-admin. A read-only identity is an M8.2 item.
- `deploy/kubernetes/inventory/disk-probe-local-path.yaml` parses as one
NetworkPolicy and three PVC/Job pairs, pinned to worker-1..3. It is outside
every Argo path and has **not yet been applied**, because this workstation has
no kubeconfig. The maintainer approved running it. Its results go in
INVENTORY.md.
## M8.1c storage probe and corrections - 2026-09-29
- The maintainer applied `deploy/kubernetes/inventory/disk-probe-local-path.yaml`
by hand on the cluster, and all three Jobs completed. The resources were then
deleted.
- worker-1: 1906 µs dsync, 28.7/60.6 ms object publish, 1003 ms for 256 MiB.
- worker-2: 2511 µs, 36.0/73.4 ms, 2122 ms.
- worker-3: 2239 µs, 32.1/61.0 ms, 2339 ms. Its root filesystem is 87 % full.
Every `local-path` volume is on the node's root LVM ext4 volume. The figures are
provisional until the planned RAID change.
- Corrections from the maintainer:
- Host RAM was increased by about 500 GB, so the earlier "about 20 GiB left"
estimate is withdrawn.
- AEvolution's PostgreSQL is a separate VM outside Kubernetes, reached through
`aevolution-db`.
- Collection is manual, with no automation planned.
- No credential values were read. The 18.6 version, role and TLS requirements are
enforced by the adapter and migrations when M8.3 connects.
## M8.2 environment isolation - 2026-09-29
- `tools/ci/deploy-policy-check.py` (WSL, Python 3 with PyYAML): 35 objects in 27
files, 0 failures.
- Nine mutations were each caught, then restored:
- World0 auto-sync;
- the World0 cascade finalizer;
- the policy set to failurePolicy Ignore;
- the policy no longer covering Secrets;
- an egress hole in default-deny;
- the World0 protected label removed;
- the cleanup tool accepting world0 (rc 127, it reached kubectl);
- service-account token automount;
- a wildcard stage destination.
- `kubectl kustomize` (client v1.31.4) renders the dev, stage and World0 platform
overlays (9 objects each), and still renders the live `deploy/kubernetes/dev`
path unchanged.
- `bash -n` passes for both ops scripts.
- The CI `deploy` suite (a Docker `ci-deploy` stage) has not run yet; Docker is
not running locally.
- Not executed: applying anything to the cluster, the admission policy's CEL
compilation (it needs an API server; the runbook validates it with a server
dry run first), and `tools/ops/verify-isolation.sh`.
## M8.2 live verification - 2026-09-29
The maintainer applied the dev, stage and World0 platform overlays, the World0
admission policy and the stage and World0 AppProjects, then ran
`tools/ops/verify-isolation.sh` against the cluster. The output is kept locally in
`k8s-script-results.md`, which is gitignored.
- **Guardrails:** 24 PASS. All three namespaces have their labels (World0 is
protected), quota, limit range, default-deny and DNS policies, and three service
accounts.
- **RBAC negatives:** 144 PASS. None of the nine environment service accounts can
get or list secrets, create pods, delete PVCs or create rolebindings in any of the
three namespaces, and none can delete namespaces.
- **Admission:** 3 PASS, 2 SKIP. The policy is bound. Server dry runs deleting the
World0 namespace and a World0 service account were both denied, naming the
break-glass group. The secret and PVC cases were skipped because none exist in
World0 yet.
- **Cleanup:** 4 PASS. The tool refused `world0`, `evolution-world0`, `World0` and
the empty name.
- **Projects:** 3 FAIL, all a script defect. The project jsonpath concatenated the
destinations with the cluster whitelist, which prints `[]` when empty. The
destinations themselves were exactly `evolution-<env>`, with an empty whitelist.
The script now checks destinations and cluster kinds separately and accepts
`[]`. The corrected script has not been re-run on the cluster.
- **WARN:** the `default` AppProject allows every namespace. The maintainer
accepted this (docs/BACKLOG.md).
## M8.3 dev/stage stores - 2026-09-29
- WSL Linux `checks` set: CTest 59/59 in each configuration. The format check
passed.
- PostgreSQL suites: Release and ASan/UBSan passed with the new
`region_store_tests`. Analysis passed after one clang-tidy fix (an int
multiplication widened in the MiB report).
- Store tests:
- An unreachable reserve refuses new objects with `full`, leaving no files. An
identical existing object is still reused.
- A lineage batch through a full store fails as `upload`/`full`, and the head is
unchanged.
- RLIMIT_FSIZE makes a real write fail with `io`, leaving no object or
temporary.
- `check-store`:
- its first run creates the probe and its second verifies it;
- a missing root exits 4 without creating it;
- a full volume exits 4;
- an owner login exits 3 (tables READABLE);
- the runtime login reports `version=180006 ... ssl=false` (local socket), and
`tables=denied functions=reachable`.
- An `evo-server` whose reserve cannot be met simulated but acknowledged
nothing. Its drain exited 3, and the head was unchanged.
- Mutations (restored and rebuilt), each caught:
- removing the reserve check;
- `check-store` always succeeding.
- `tools/ci/deploy-policy-check.py`: 41 objects, 0 failures. Four new-rule
mutations were caught:
- a Delete class;
- a platform claim on `local-path`;
- a `PGPASSWORD` key reference;
- a whole-Secret `envFrom`.
- Kustomize renders the dev and stage overlays with the `aevolution-objects`
claim; World0 has none.
- The server image now contains `evo-world-clone`, and its smoke script checks the
tool's usage exit. It has not been built locally, because Docker is not running.
- Not executed: anything in the cluster, the store Jobs, retention, rebind and
rotation evidence, and the image build.
## M8.3 service-file connections - 2026-09-29
- The maintainer's dev Secrets have the keys `ca.crt`, `pg_service.conf`
(section `[evolution]`) and `pgpass`, with no `user` key. The values were
pasted into the conversation, so both passwords must be rotated. That rotation
is also the M8.3 credential-rotation evidence step. The values are not recorded
anywhere in the repository.
- `read_pg_service` and `--pg-service` were added to `evo-server` and
`evo-world-clone`.
- Parser tests: a valid file with another section and CRLF lines parses. The
parser refuses:
- `password`, `passfile` and `options` keys;
- `sslmode=require`;
- a repeated key or a repeated section;
- a non-numeric timeout;
- malformed lines;
- a missing section;
- a file over 4 KiB;
- a missing file.
`sslmode=disable` is accepted only for a local socket.
- `check-store --pg-service` against the local cluster verified the probe and the
privilege boundary.
- The store Jobs now mount each login Secret whole (service file, password file,
CA) and no longer read `aevolution-db`. The migrate Job's `PGSERVICE` is now
`evolution`; the previous `evolution-migrator` did not match the maintainer's
file.
- PostgreSQL Release and ASan/UBSan passed. The analysis build passed after one
loop copy was fixed. Deploy policy: 0 failures.
## M8.3 Argo delivery - 2026-09-29
- The platform layer is now `deploy/kubernetes/platform/cluster`, delivered by the
`aevolution-platform` Application. It renders 29 objects, all marked
`Prune=false,Delete=false`: 3 namespaces, 3 quotas, 3 limit ranges, 6 network
policies, 9 service accounts, 2 claims, the storage class, the policy and its
binding.
- The store Jobs are kustomize components used as Argo hooks. The dev overlay
renders into `evolution-dev` with PreSync and PostSync hooks; its digests and
role are still markers.
- `deploy-policy-check.py`: 43 objects in 41 files, 0 failures. Five new mutations
were each caught:
- platform prune enabled;
- a missing Delete guard;
- an unfinished stores overlay included in the live dev path;
- an extra cluster kind in the platform project;
- the admission policy no longer covering the platform app.
The unfinished-path rule first missed the live dev Application, which exists
here only as an `.example`. It now covers every Application whose path is
resolved.
- The digests the maintainer pasted are local image IDs from the build step, not
registry digests. The registry digest is printed by the Publish step.
- Not executed: bootstrapping the platform Application, or any Argo sync.
## M8.4 region host - 2026-09-29
- Windows MSVC Release CTest 59/59, including the new opening code.
`lineage-region-contracts` proves `development_opening(7)` is byte-identical
to the fixture, and that world 100 differs only by identity.
- WSL Linux `checks` set: 59/59 in each configuration. The format check passed.
- PostgreSQL suites: Release, analysis and ASan/UBSan passed. Analysis passed after
one optional-access fix in `provision`.
- Provisioning test:
- A runtime login cannot provision.
- The owner provisions `world=7 region=86 sequence=0 tick=0`, and a rerun reports
`already-provisioned`.
- A server then became ready and acknowledged ticks 1–5, after which the tool
reported `sequence=1 tick=5` and left the region untouched.
- Mutation: removing the already-provisioned early return was caught.
- `deploy-policy-check.py`: 50 objects, 0 failures. Four new mutations were each
caught:
- a build identity differing from the image digest;
- a RollingUpdate strategy;
- a CPU limit;
- a grace period shorter than the drain.
The checker first misread overlay patches as whole Deployments; it now checks
only the base.
- Kustomize renders dev and stage in order: egress policy (−2), migration hook
(−1), provisioning hook (0), Deployment and Service (1), then the store check
(PostSync).
- `region-drill.sh` passed `bash -n`, and refuses World0 before kubectl (also
enforced by the policy check).
- Not executed: the in-cluster region sync and the drill.
## M8.5 observability - 2026-09-30
- `promtool` 3.13.3 and `amtool` 0.34.1 came from the official release tarballs,
with SHA-256 verified against the published sums: `b349c732…` and `265b9d1e…`.
- The Prometheus configuration syntax is valid, and all 16 rules are valid.
- The alert injection tests passed, as did the Alertmanager `check-config`.
- Rule mutations: five were tried, and three were caught at first.
- **NotReady firing after 15 s:** missed, because the forced-deletion test only
evaluated after recovery. The test now also checks during the dip.
- **ObjectsFull weakened to half the reserve:** missed, because the injected
free space was below both thresholds. The test now uses 900 MB.
- After these fixes all five were caught: the lease threshold, NotReady too
eager, debt counted without inputs, a weakened ObjectsFull, and `refused`
drops counted.
- Windows MSVC CTest 59/59, including the `ops-contracts` update to 32 series.
- WSL Linux `checks` set: 59/59 in each configuration. The format check passed.
- PostgreSQL suites: Release, analysis and ASan/UBSan passed. The ops process
test checks the reserve gauge (1 GiB) and that free space is nonzero.
- `deploy-policy-check.py`: 67 objects, 0 failures. New rules:
- monitoring images must be digest-pinned;
- monitoring pods mount no token;
- no RBAC objects are allowed;
- the monitoring project is scoped.
- Kustomize renders the platform bundle (four namespaces), the monitoring app
(2 Deployments, 2 ConfigMaps, 2 Services, 2 policies, 1 claim), and dev and
stage with the scrape policy.
- The CI `ci-deploy` stage now copies `promtool` and `amtool` from the pinned
images and runs `tools/ci/monitoring-check.sh`. It has not run locally, because
Docker is not running.
- Not executed: anything in the cluster (monitoring sync, target health, the
`EvoRegionDown` injection), alert delivery (no receiver), and the node log
rotation check.
## M8.4 and M8.5 in-cluster results - 2026-09-30
- `tools/ops/region-drill.sh dev`, run by the maintainer (idle region, world 100):
- start: epoch 3, committed tick 0;
- rollout restart: the replacement was Ready in **3010 ms**, at epoch 4 with
committed tick 0 (Recreate drain released the lease);
- forced deletion: the replacement was Ready in **19268 ms**, at epoch 5 with
committed tick 0 (waiting out the 20 s lease);
- PASS.
- M8.5: the maintainer reports the new server digest pinned and the monitoring
project, platform, monitoring app and dev/stage applications synced. Target
health and the `EvoRegionDown` injection results were not recorded here.
- Image upgrade to `edb39f…`: completed with `REGION_RETAINED=4af72099…`, so the
existing dev and stage heads, written by the earlier build, stay recoverable
after the upgrade.
## M8.6 synthetic load and fault drills - 2026-09-30
- `region_load_tests` (real PostgreSQL and real `evo-server`, fast lease):
- A synthetic owner acknowledged contiguously from tick 1 past tick 45, with
deterministic `SYN1` IDs. The consumer's death was emitted as an outcome.
- After a SIGKILL mid-stream, every acknowledgement was at or below the
committed head, and there was exactly one durable decision.
- The successor started at epoch 2 and acknowledged exactly committed + 1 onward,
with no repeated death.
- With synthetic inputs enabled, a line typed on stdin was ignored.
- `lineage-region-contracts`: `development_input(t)` equals the fixture's
`region_input(t, true)` for ticks 1–200.
- Mutation: resuming the driver at committed + 2 was caught by the contiguity check.
- Windows MSVC CTest 59/59. WSL Linux `checks` set: 59/59 in each configuration.
PostgreSQL suites passed in Release, analysis and ASan/UBSan. The format check
passed.
- `deploy-policy-check.py`: 67 objects, 0 failures. New rules: `REGION_INPUTS`
must be empty in Git, and `fault-drill.sh` must refuse World0 before kubectl.
- `fault-drill.sh` passed `bash -n`. It refuses `world0` and unknown scenarios
(exit 2) without calling kubectl.
- Not executed: the in-cluster drill (kills, full, empty, db-outage).
### M8.6 first in-cluster drill (dev, run by the maintainer) - 2026-09-30
`fault-drill.sh dev all`, then `fault-drill.sh dev db-outage`:
- **kills: PASS.** RTO 20 147, 19 777 and 20 222 ms (lease-bound, as ADR-0073
expects). The dead pods' highest acknowledgements (209, 419, 629) equalled the
heads their successors recovered: no acknowledgement was lost.
- **full: PASS, with a finding.** No acknowledgement while full, and recovery from
the committed head afterwards. But `store_failures` reached 4 247 in about 30 s:
the host retried a refused publication on every step.
- **empty: PASS, with an evidence gap.** No pod became Ready, and restoring the
volume recovered committed 894, past the last acknowledgement (888). The script
found no `phase quarantined` line, because it read only the previous container's
log.
- **db-outage: FAIL, not yet diagnosed.** The host stayed Ready for the whole
10-minute window. Nothing shows that the database was actually unreachable from
the pod's node. The prompt did not name the node or the rule to apply, so the
block may have targeted another address.
Fixes (this segment):
- `RegionHost` backs off commit retries after a store failure: 100 ms, doubling to
a 2 s cap, and reset by the next confirmed commit. `region_host_tests` now
expects exactly 2 attempts in the first 200 ms and 5 in 2 s of a store outage,
and a commit within 23 steps after the store heals. Mutation: removing the
backoff gate was caught.
- `fault-drill.sh` empty: reports the restart count and last exit code (4 means
quarantined), keeps the current and previous logs, and prints their tails.
- `fault-drill.sh` db-outage: prints the pod's node, the node's InternalIP, and
the exact `iptables` insert and delete rules to run on the database VM.
- Checks run: the format check; WSL `checks` 59/59 in debug, release, analysis
and ASan; the PostgreSQL suites in analysis and ASan/UBSan; the mutation. Windows
MSVC CTest 59/59 (the server is not part of the Windows build).
`fault-drill.sh` passed `bash -n`.
- Pending: a new image, then re-running the full, empty and db-outage scenarios.
## M8.7 backup and restore - 2026-09-30
- `tools/ci/backup-check.sh` (inside `postgres-check.sh`; real PostgreSQL 18.6, a
real `evo-server` under synthetic load) passed in the Release, analysis and
ASan/UBSan PostgreSQL builds:
- An online backup taken while the server committed recorded head tick 60, at or
past every acknowledgement before it (60). The source then moved on to tick 101.
- `verify-backup.sh --isolate 301` on a second, throwaway cluster, as fresh
roles: the restore audited 13 commits to tick 60 and matched the MANIFEST. The
clone into world 301 resumed at `epoch=1 committed=60`, acknowledged to tick
82, and drained with exit 0.
- Refused bundles, each leaving no schema and no objects behind:
- corrupt object, unlisted file, damaged dump: exit 4 at the checksum step;
- missing payload object with consistent checksums: exit 3,
`audit failed missing_payload sequence=0`;
- MANIFEST head changed: exit 3, heads differ.
- A second restore into a database that already holds a world was refused
(exit 4), and the first restore was left unchanged.
- The MANIFEST names no credentials.
- Mutations, each caught by `backup-check.sh`:
- restore without the history audit: the missing object was accepted (exit 0);
- a failed restore that keeps its files: restored objects were left behind;
- `pg_dump` outside the exported snapshot (2 s later): the restored heads
differed from the MANIFEST, and verification failed.
- The PostgreSQL migration, rollback and permission suites passed in all three
builds. `deploy-policy-check.py`: 67 objects, 0 failures.
`cluster-backup.sh` refuses World0 before kubectl (policy rule).
- `bash -n` passed on every new script. shellcheck is not installed in WSL, so it
did not run.
- Not executed:
- the postgres-tools image build with `pg_dump` (CI builds it on push);
- `cluster-backup.sh` against dev or stage;
- any restore of in-cluster data.
- The permanent destination, encryption, schedule and WAL archiving are not
implemented; they are pending the maintainer's choice of destination.
## M8.8 World0 gate and host-loss rehearsal - 2026-09-30
- `tools/ci/dr-check.sh` (inside `postgres-check.sh`; real PostgreSQL 18.6 and real
`evo-server` processes) passed in the Release, analysis and ASan/UBSan
PostgreSQL builds. The Release run:
- The old host served world 400 past the tick-20 death.
- An online backup cut at tick 40, and the old host went on acknowledging, to
tick 110.
- An in-place restore into a replacement database, with the old host still
running, audited 9 commits and printed `REWIND ... restored at tick 40`. It
left no lease held and passed the runtime rights audit.
- The new host resumed at `phase running epoch=2 committed=40`, and its first
acknowledgement was tick 41.
- `fence-database.sh` retired build b1…, leaving `current 0` and `held 0`, and
printed the final head (`final 400 1 22 110 1`). The still-running old host
dropped its queued inputs as `fenced` and exited 4 (quarantined). The old head
stayed at 110.
- A later host on the fenced database logged `phase quarantined` and exited 4,
never ready and with no acknowledgement.
- The restored world's audit passed (16 commits to tick 73, one outcome). There
was exactly one durable decision, and the new host printed no `outcome 20`.
- Measured rewind: 70 acknowledged ticks (110 − 40).
- The ASan run measured 115 and 75.
- Mutations, each caught by `dr-check.sh`:
- leases not fenced on restore: "a restored lease is still held";
- fence without the barrier: "fence evidence";
- in-place restore without grants: the runtime rights audit failed, and the
restore was removed.
- Deploy policy negatives, each rejected: a Job in the World0 path; the World0
Application with a resolved path; a `region/world0` overlay pinning a digest
that is not stage's. The unmodified tree: 67 objects, 0 failures.
- Found and fixed during validation: `dr-check.sh`'s EXIT trap ran `kill -9` on
exited processes under `set -e`, which made the script exit 1 after passing
(the kill now ends in `|| true`).
- The backup check still passes with grants kept in the dump.
- Not executed:
- any in-cluster host-loss restore or fence;
- a second failure domain, which does not exist;
- any World0 deployment or promotion, which is blocked by design.
## M8.9 upgrade rehearsal and node plan - 2026-09-30
- `region_upgrade_tests` (real evo-server and evo-world-clone processes) passed in
the Release, analysis and ASan/UBSan PostgreSQL builds:
- Current build b1 and candidate b2 ran on clones of one cut, at 20×
acceleration, from tick 30 to tick 150: 120 ticks, 6 s of region time. Wall
times were 306/307 ms (Release), 322/323 ms (analysis) and 299/299 ms (ASan).
- `compare` passed: `compared tick=150 outcomes=1`.
- A temperature-shifted input stream gave `compare differs state` (exit 6).
- A clone stopped at tick 100 gave `compare differs ticks` (exit 6).
- Comparing a world with itself, and `--accelerate` or `--synthetic-until`
without the driver, were usage errors (exit 2).
- `tools/ci/upgrade-check.sh` passed in all three builds:
- A failing migration (migration lock, `ALTER TABLE region_commits`, locks held
3 s, then an error) left schema v9 and no new column. b1 went on
acknowledging, and the real migrate was a no-op.
- b2 waited while b1 held the lease. `upgrade_barrier(b1)` fenced one lease, and
b1 exited 4.
- b2 resumed at `epoch=2 committed=135`, at b1's last acknowledgement (135), and
continued with tick 136. The ASan run: 137.
- A late b1 host exited 4 without becoming ready.
- The history audit across both builds passed: 24 commits to tick 167.
- Mutations, each caught:
- `compare` without the state comparison: upgrade test line 201 (the diverged
world passed);
- `--synthetic-until` ignored: line 165;
- `--accelerate` as a no-op: line 184 (over 4 s).
- The WSL `checks` set: 59/59 in debug, release, analysis and ASan. clang-tidy
(analysis builds) and the format check were clean.
- Deploy policy: 67 objects, 0 failures. Negatives, each rejected: an
objects-volume pod without the stores selector; a required compute affinity;
`--accelerate` in the region Deployment. `upgrade-barrier.sh` and
`scheduling-check.sh` refuse World0 before kubectl.
- Found during validation: genesis seed 0xF1 collided with a checkpoint-store test
(fixture provisioning failed), and was replaced by the unused 0xD7. clang-tidy
flagged duplicated option branches, which were merged.
- Not executed:
- `upgrade-barrier.sh` or `scheduling-check.sh` against dev or stage;
- scheduling onto a second labelled node;
- the Windows MSVC build (no Windows target changed; the server and PostgreSQL
code are Linux-only).
## M8.1–M8.9 review fixes - 2026-09-30
Review: `.cache/reviews/M8_REVIEW_680e7e5-5b482ad.md` (one P1, five P2).
- **P1 restore ownership.** `backup-check.sh` passed in the Release, analysis and
ASan/UBSan PostgreSQL builds, including three new cases:
- A schema created after the checks but before `pg_restore`: exit 4 ("pg_restore
failed; the database was not changed"). The competing table survived, and only
this restore's own objects directory was removed.
- A second concurrent restore of the same database: exit 4 ("another restore
holds this database"). It left no directory, and the first restore was
verified.
- While the first restore held the database, a lease acquisition there ended in
`lock timeout`. An occupied objects path was refused and left untouched.
- The reviewer's `restore-race.sh` now ends with `remaining_evolution_schemas=1`
(`schema=no, objects=yes`).
- **P2 barrier Ready-pod selection.** `tools/ops/lib/pods.sh` uses a plain
jsonpath and needs exactly one Ready pod that is not being deleted.
`ops-lib-check.sh` passed its fixtures: one pod, not Ready, being deleted, no
condition, Ready beside a deleting pod, two Ready pods, and no pods. Not run:
the jsonpath against a live API server (no kubectl or Go in this environment).
- **P2 barrier Secret access.** The generated pod has `fsGroup: 10001`. The deploy
policy check now renders the backup and barrier pods with `--print-manifest` and
checks: non-root 10001 with fsGroup 10001, no token, read-only root, no
escalation, Secret files 0440, and egress admission of the component. Not run: a
disposable pod reading the Secret in the cluster.
- **P2 World0 claimRef.** A new admission validation denies removing a World0
volume's claim reference, or changing its namespace or name, without
break-glass. The policy check requires the rule, and `verify-isolation.sh`
dry-runs the removal when a World0 PV exists. Not run: no admission test cluster
here, and no World0 PV exists.
- **P2 CI change selection.** `ci-paths.sh` compares against the event's base (the
push's `before`, or the PR base, fetched if missing). It runs on manual
dispatch, local runs and any uncertainty, and the path lists now include the
check scripts. `ci-paths-check.sh` passed 11 fixtures, and the deploy suite runs
it first. The reviewer's `ci-range.sh` now runs in both cases. Not run: a real
Gitea push, so the event-context mapping is unverified.
- **P2 drill verdicts.** Every kill RTO is asserted against `RTO_BUDGET_MS`
(60 000). The empty scenario needs positive quarantine evidence; INCONCLUSIVE and
other exits fail. Acknowledgements are read from the dead or drained owner's
finalized log. The fixtures passed. Not run: the drill itself.
- Mutations, each caught:
- dropping a schema that was not installed: `backup-check.sh` failed;
- removing `fsGroup`: the policy check failed;
- accepting several Ready pods, or treating never-ran as PASS: the
`ops-lib-check.sh` fixtures failed;
- falling back to `HEAD^` when there is no base: the `ci-paths-check.sh` fixtures
failed.
- The host-loss and upgrade rehearsals, and the PostgreSQL migration and
permission suites, still pass in all three builds. The deploy policy check: 67
objects, 0 failures.
- ROADMAP corrected: M8-dev/stage is implemented and locally qualified, but **not
passed** until the live gates listed there are run.
## Review 3a34ae3 follow-up: durable restore gate - 2026-10-01
Review: `.cache/reviews/REVIEW_3a34ae3.md`. P1: losing the auxiliary claim session
released authority before verification. P2: a nonzero `pg_restore` exit was taken
as proof of rollback.
- `restore-world.sh` no longer relies on an auxiliary lock session. It works in
three steps, as the ADR-0074 amendment describes:
- an install transaction holds the dump, a claim row with the invocation's
nonce, fenced leases and retired builds;
- verification follows;
- a nonce-checked publish transaction re-enables exactly the held builds.
Ambiguous outcomes are resolved from the claim row. An unresolvable one is
OUTCOME UNKNOWN (exit 5): the target stays unservable, and
`--abandon NONCE --objects DIR` resolves it later.
- `backup-check.sh` passed in the Release, analysis and ASan/UBSan PostgreSQL
builds, with these cases:
- a competing schema before install: the rollback was confirmed and the
competitor's table survived;
- two concurrent restores: the first installer was verified and published; the
second rolled back, confirmed, and left the winner intact;
- after install, `acquire_region_lease` was refused ("build not current") both
before and after `pg_ctl restart` mid-restore; the restore then verified, and
one current build was restored;
- a lost install acknowledgement: "install outcome resolved: committed", then
verified;
- an unreachable resolution: exit 5, both schemas and the objects kept,
acquisition refused, `--abandon` with a wrong nonce removed nothing, the right
nonce removed everything, and a retry restore succeeded;
- an occupied objects path: refused and untouched;
- damaged bundles: refused, with no `evolution` or `evolution_restore` left.
- The reviewer's claim-loss reproduction, adapted to the renamed hook
(`after-install`), finds no lock to terminate. The runtime acquisition inside it
failed with `build not current` (exit 1), and the restore then published.
- `dr-check.sh` (in-place restore) and `upgrade-check.sh` passed in all three
builds.
- Mutations:
- install without retiring builds: caught ("authority was available before
publication");
- resolution ignoring a committed claim: caught (exit 5 instead of verified);
- cleanup ignoring the claim's nonce: NOT exercised (the edit pattern failed to
apply). The abandon path also checks the nonce first, so that defense is two
layers deep and only the outer layer is tested.
- Not run: a real network partition around COMMIT. The lost acknowledgement is an
injected test mode (`EVO_RESTORE_TEST_INSTALL_ACK`).
## M9.1 scoped native dependency probe — 2026-10-01
Decision: ADR-0077. SDL 3.4.16 shared window/events/2D renderer, explicitly
prepared from the official archive with SHA-256 validation. No simulation API,
world format, server image or deployed workload changed.
Executed:
- Windows x64 MSVC 19.44: fresh `client-probe` configure/build and 2/2 labeled
CTests passed. Native standalone probes reported `video=windows` with both
`renderer=direct3d11` (automatic) and explicitly requested `renderer=vulkan`;
synthetic key delivery and rectangle/background pixel readback passed.
Developer environment required explicit compiler and Windows SDK tool paths.
- WSL Ubuntu, GCC 14 Release: configured the same preset in
`build/client-probe-linux` with `SDL_X11=ON`, `SDL_WAYLAND=OFF`; build and 2/2
client CTests passed. `xvfb-run -a env SDL_VIDEO_DRIVER=x11
./build/client-probe-linux/evo-client-probe --renderer software` passed.
X11 development dependencies are reproduced in Docker's `ci-client` stage.
- clang-tidy 19 standalone against `experiments/client/main.cpp`, C++23 and the
prepared SDL include directory: no owned-code diagnostics. Initial use of the
GCC compilation database failed on GCC module options; rerun used explicit
Clang arguments. Repository `format-check` passed.
- Clang 19 ASan/UBSan build of the owned probe, linked to the Release SDL shared
library: dummy video/software renderer passed. SDL itself was not instrumented.
**Additional X11 sanitizer run failed LeakSanitizer:** 50,016 bytes in 912
allocations remained at exit; resolved SDL call sites included
`X11_InitKeyboard` and `X11_GetGlobalContentScale`, with underlying unloaded
library frames unresolved. This is not a proven application leak or a clean
native sanitizer result. Do not suppress or declare it fixed. M9.2 must reproduce
with symbolized native libraries and establish repeated window-lifecycle memory
behavior before desktop qualification.
Diagnostic rerun with `LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libX11.so.6`
passed with leak detection enabled. Keeping libX11 loaded changes the result,
suggesting library unload/static allocation reachability; it does not establish
that repeated native lifecycles are leak-free. No suppression was added.
- Headless Release, client disabled: configured/built `evo-sim`; seed 123456,
eight entities, 1,000 ticks and `--verify-replay` passed, hash `d61ad5bf0c928c76`.
- Offline cached archive preparation passed; deliberately invalid archive was
rejected with checksum mismatch before extraction. CI selector regression:
all 12 fixtures passed, including an SDL pin change selecting the client gate.
- Artifact sizes (not comparable performance measurements): Windows executable
14,336 bytes + SDL3.dll 1,998,848; Linux X11 executable 17,624 bytes + shared SDL
3,304,808. Build configurations/backend sets differ. No throughput or cold-build
benchmark claim is made.
Limits: synthetic keyboard events only; hidden Windows window and virtual Linux
X11 display. No human visual inspection, real input/focus/DPI, Wayland or Linux
GPU qualification. A Linux offscreen GLES attempt failed because the environment
could not initialize OpenGL/GLES. No full headless CTest rerun, TSan, Docker build,
remote Gitea run or Kubernetes action was performed for this gate. The new client
CI job performs analysis, dummy pixel/error tests and an Xvfb window probe;
remote execution remains pending. M9.1 selects a scoped dependency, not a
production-ready renderer or packaged client.
## M9.2 native application and renderer boundary — 2026-10-01
Implemented ADR-0078, opt-in `evo-client`, bounded immutable presentation frames,
private main-thread SDL ownership, logical/drawable size separation, focus and
visibility/minimize handling, recoverable reset and terminal device-loss policy.
No camera, gameplay, protocol, simulation-state or deployment changes.
Executed:
- Fresh Windows Release: `tools/windows-check.ps1 -Preset client -Action check`
configured and built with MSVC 19.44; **3/3 client CTests passed** under NDEBUG.
Also ran `build/client/evo-client-lifecycle-tests.exe` with native Windows video,
then `evo-client.exe --frames 3` (default renderer) and `--renderer vulkan
--frames 3`; all returned success. Earlier combined probe/client configuration
passed 5/5 labeled tests.
- Linux GCC 14 Release in `build/client-probe-linux`: **5/5 combined probe/client
CTests passed**, as did native X11 lifecycle checks under Xvfb. The new suite executes 32
window/app destroy/recreate rounds, plus partial renderer initialization failure
and hidden-window/quit behavior. It covers copied-frame independence, finite
geometry/primitive limits, foreign events, focus, resize/scale queries,
non-unit anisotropic scale pixel readback, hide/minimize/resume, renderer
reset/loss and terminal close/error behavior.
- Clang 19 Debug ASan/UBSan: `cmake --preset asan -B build/client-asan
-DEVO_BUILD_CLIENT=ON -DCMAKE_CXX_COMPILER=clang++-19
-DCMAKE_C_COMPILER=clang-19 -DSDL_X11=ON -DSDL_WAYLAND=OFF`, built application
and lifecycle tests, **3/3 CTests passed**. Xvfb/X11 lifecycle executable also
passed with default leak detection, **no LD_PRELOAD or suppressions** after
adopting direct X11 linkage. Owned libraries, app and tests are instrumented;
SDL and system libraries are not comprehensively sanitizer-instrumented.
- Investigated M9.1's X11 unload issue: dynamically loaded X11 configuration left
1,700,544 bytes / 31,008 allocations reported at process exit in the repeated
fixture. Preloading libX11 avoided the report. SDL's supported
`SDL_X11_SHARED=OFF` fixes the observed repeated-lifecycle case by retaining the
linked system libraries for process lifetime. `ldd` confirms libX11 and libXext
dependencies; normal X11 sanitizer run passes. This is an adopted Linux build
constraint, not a claim to have repaired upstream Xlib or all native drivers.
- `clang-tidy-19 -p build/client-asan` on all five new C++ translation units:
passed with warnings-as-errors. Initial enum-size and multi-level pointer
conversion diagnostics were corrected. Repository `format-check` passed.
- Headless Release configured with **both client options OFF**, built `evo-sim`
and replayed seed 123456, eight entities, 1000 ticks: unchanged hash
`d61ad5bf0c928c76`, `replay_verified=true`. No SDL acquisition is part of this
configure/build path.
CI's isolated client stage now builds the actual app, runs static analysis and
labeled checks, exercises X11 lifecycles, then instruments owned code and repeats
software/X11 checks. No Docker build or remote Gitea execution was verified here.
No full headless suite rerun or TSan run: headless source did not change, and this
boundary requires single-main-thread use. Tests do not establish physical
multi-monitor DPI, real keyboard/focus behavior, Wayland, hardware GPU loss or
human visual acceptance. Those remain M9.3/M9.8 qualification work. No cluster
operation was performed.
## M9.3 camera and input intentions — 2026-10-01
Implemented ADR-0079 and CAMERA_INPUT.md: SDL-free local-mm camera and held input
state, explicit logical/pixel conversion, cursor-anchored zoom, bounded
sequence/tick intentions and polling-before-projection. The app demonstrates
arrows/wheel navigation, a WASD intention indicator and Tab UI capture. Its tick
is a local fixture counter; no server/session/authoritative movement is present.
Executed:
- Windows MSVC 19.44 Release via `tools/windows-check.ps1 -Preset client -Action
check`: 3/3 labeled tests passed. Native Windows lifecycle/input executable
and the final `evo-client --frames 3` smoke passed.
- Linux GCC 14 Release, existing combined probe/client build: 5/5 labeled tests
passed; native Xvfb/X11 lifecycle and input events passed. The final app-only
startup/error-reporting adjustment was additionally built/tested with MSVC and
Clang ASan below (the GCC suite preceded that adjustment).
- Clang 19 Debug ASan/UBSan: all owned camera/input/platform/app/test code built;
3/3 labeled tests and native Xvfb/X11 tests passed with leak detection enabled,
without preload or suppressions. Final hidden/software three-frame app smoke
also passed. SDL/system libraries remain outside comprehensive instrumentation.
- Pure camera tests cover 220 world/logical/pixel round trips over four viewport/
DPI/aspect combinations and five zoom scales (1e-6 mm tolerance for fixtures),
anchor preservation, pan direction, resize stability, NaN/infinity, numeric
limits and unchanged state after rejected operations.
- Input tests cover key release/opposites/diagonals, focus re-arm, ignored repeats,
UI capture, bounded/consumed wheel accumulation, invalid input, monotonically
increasing sequence/tick and stale/overflow rejection. Native SDL tests cover
WASD/arrows, Tab repeat, focus loss/regain, wheel direction/pointer, minimize,
saturated event batches and queued keys crossing external capture release.
- clang-tidy 19 with warnings-as-errors passed on all six changed/new translation
units; corrected a duplicated wheel-error branch. Final app-only adjustment
also passed analysis. Repository format-check passed.
- Headless Release configured with both client options OFF, rebuilt `evo-sim`,
replayed seed 123456 / eight entities / 1000 ticks: unchanged
`d61ad5bf0c928c76`, `replay_verified=true`.
The existing client CI target now picks up the controls library and expanded
lifecycle suite automatically. Docker/Gitea execution was not observed locally.
No full headless suite or TSan rerun; owned client state remains main-thread-only.
No physical keyboard/IME, human visual, real multi-monitor DPI, Wayland or real
GPU-failure qualification is claimed from injected events and Xvfb. No network
traffic, database, world mutation or deployment operation was introduced.
## M9.4 microscopic environment rendering — 2026-10-01
Implemented ADR-0080 and ENVIRONMENT_RENDERING.md: owned revision-tagged
`EnvironmentPresentation` copied from M6.3 views (no tokens/store pointers),
single-slot `EnvironmentLayer` ordering/freshness, and bounded camera-projected
rendering with explicit missing, not-requested, present-zero, undefined-moisture
and stale encodings. The app shows a local in-memory fixture authority (finite
same-channel water moves via `FieldStore::transfer`); it is not a server view.
Executed:
- Windows MSVC 19.44 Release via `tools/windows-check.ps1 -Preset client -Action
check` (run as a child `powershell.exe`): 4/4 labeled tests, including the new
`client-environment` suite. App runs with `--frames 240/300`, `--refresh-every
2`, `--layer organic-carbon` exited 0; `--layer bogus` exited 2;
`SDL_VIDEO_DRIVER=evo-invalid` still fails at SDL initialization (error 2),
after fixture creation, so `client-init-failure` keeps testing the SDL path.
- Linux GCC 14 Release (`build/client-probe-linux`, probe + client): 6/6 labeled
tests; native Xvfb/X11 lifecycle suite; X11 app runs (moisture, refresh 3).
- Clang 19 Debug ASan/UBSan (`build/client-asan`): 4/4 labeled tests, native X11
lifecycle, hidden X11 app (refresh 2), visible X11 moisture app and dummy-driver
nitrogen app (400 frames, refresh 3) with leak detection, no suppressions.
- clang-tidy 19 with warnings-as-errors on the four new/changed translation units:
passed after reserving a test vector, widening a test product, and annotating
two intentional out-of-range enum inputs (the rejection under test).
Repository-wide clang-format 19 check passed.
- Tests cover: synthetic known fields (exact ramp colour for log2 density -3,
present zero, missing checker geometry, hollow unrequested outline, undefined
zero-capacity moisture, collapsed small habitats); seams across chunk
boundaries -16/0/16 at three zooms, three centres and two DPI ratios, compared
with `Camera::world_to_screen`; voxels on the exclusive world bound; stale
updates (older tick, chunk-revision regression, foreign world, regressed and
pre-presentation watermarks, late-but-stale presentations, reset); failed
local validation after a field transfer; unchanged field/terrain snapshots
after presenting/rendering; presentations outliving their stores; 20 invalid
presentation shapes; 330 camera/mode/freshness combinations over 256 cells,
all within viewport, the 2313 bound, constant buffer capacity and Frame
limits including host UI.
- Profile (`environment_profile`, 256 visible cells, stale, nitrogen layer,
checker detail): 1701 rectangles; 53,248 presentation-cell bytes; 46,260-byte
reused output buffer; mean build 13.4 us (MSVC Release, this workstation),
6.6 us (GCC 14 Release, WSL2), 175 us (Clang ASan Debug). Indicative only; no
frame-time threshold is claimed.
- Visual sanity check: the fixture's output was rasterized offline for water,
stale, nitrogen (missing), moisture and organic-carbon (not requested). This
raised the not-requested outline from grey 140 to 210 for contrast with rock.
This is not human/colour-vision acceptance.
- Headless: a fresh MSVC Release directory with both client options OFF built
`evo-sim`; seed 123456 / eight entities / 1000 ticks replayed with unchanged
`d61ad5bf0c928c76`, `replay_verified=true`.
Limits: pre-existing `build/release` failed to link untouched headless objects
(unresolved MSVC STL `__std_rotate`/`__std_find_first_not_of_trivial_pos_1`),
consistent with stale objects from a different STL build; the fresh directory
above passed, and that directory was left as found. Docker/Gitea execution of the
extended `ci-client` stage was not observed locally. No full headless suite or
TSan rerun (headless sources unchanged; client state is main-thread-only). No
physical DPI, real GPU, Wayland, accessibility or human visual acceptance is
claimed. No network traffic, database, deployment or persistent format changed.
## M9.5 organism and debug views — 2026-10-01
Implemented ADR-0081 and ORGANISM_RENDERING.md: biology-derived
`OrganismPresentation` (no controller/owner/level fields), `OrganismLayer`
ordering/freshness, budgeted organism projection with explicit truncation, and an
opt-in rectangle-only hover inspector. ImGui was evaluated as optional and not
adopted. The app adds a static 48-organism fixture built from real genomes.
Executed:
- Windows MSVC 19.44 Release via `tools/windows-check.ps1 -Preset client -Action
check` (child `powershell.exe`): 5/5 labeled tests including `client-organisms`.
App runs `--frames 300 --refresh-every 2 --inspect` and `--frames 120 --layer
moisture` exited 0; `--frames 0` exited 2 (usage).
- Linux GCC 14 Release (`build/client-probe-linux`): 7/7 labeled tests, native
Xvfb/X11 lifecycle suite, X11 app with `--inspect --refresh-every 3`.
- Clang 19 Debug ASan/UBSan (`build/client-asan`): 5/5 labeled tests, X11
lifecycle, X11 app (`--inspect --layer moisture --refresh-every 2`, 300 frames)
and hidden dummy-driver app (`--inspect`, 200 frames); leak detection on, no
suppressions.
- Organism tests: phenotype equals `derive_phenotype` (20 um cube, hand-checked
bounds 390..410 at 1000 units/mm); a structured binding pins the cell's seven
fields; membrane thickness cap, motility ring and ingestion mark only with the
capabilities; conversion/growth/damage change body colour; visibility floor and
halo; byte-identical rectangles and inspector panels for player- and
NPC-controlled hosts with identical biology (ids 5 vs 900), translation-only
differences when moved; death (remnant colour, no ring/mark, same size) and
removal (absence, then empty presentation); 1024 visible organisms collapsing to
bodies within the 1755 app budget, explicit truncation at budget 500 (498 drawn
plus marker), invalid budget; a worst-case frame (environment bound + organisms
+ inspector + UI) within `Frame` limits; regression/foreign/watermark ordering
and stale dimming; 15 invalid presentation shapes and over-capacity; topmost picking on
floored bounds; gauge values and widths.
- Visual sanity check (offline rasterization of the app fixture at habitat zoom
and 2000 units/mm with the inspector): 3-unit floored organisms were hard to see
against habitats, so the floor became 5 units with a dark halo for collapsed
bodies. Not human, colour-vision or accessibility acceptance.
- clang-tidy 19 with warnings-as-errors on the five new/changed translation units:
passed after reserving a test vector. Repository-wide clang-format 19 passed.
Limits: Docker/Gitea execution of the extended `ci-client` stage was not observed
locally. No headless source changed; no full headless suite or TSan rerun. The
fixture population is static presentation data, not ecology. No network, database,
deployment, persistent format or simulation rule changed.
## M9.6 narrow secure dev-session protocol — 2026-10-01
Implemented ADR-0082, DEV_SESSION_PROTOCOL.md and schemas/view/dev-session-v1.md:
protocol v1 codec and FrameReader, SessionAdmission, dev-only HMAC grants and a
TLS 1.3-only channel over caller byte transport, using Mbed TLS 4.1.1 LTS.
Dependency evidence (2026-10-01): upstream BRANCHES.md lists 4.1 as LTS supported
until March 2029 (3.6 until March 2027). `mbedtls-4.1.1.tar.bz2` SHA-256
`3359a349e23db3d5536fcee032ae7b2ecbfc08972fab643089b5cbf2a375c98c` matched the
release's published sha256sum file (same origin). License: Apache-2.0 OR
GPL-2.0-or-later; Apache-2.0 taken, copied to third_party/mbedtls/LICENSE. Release
tarballs need no Python/Perl with programs/tests off. Windows extraction skipped
one symlink inside an unused upstream PQC example directory. MSVC Release static
archives: mbedtls 647 KB, mbedx509 376 KB, tfpsacrypto 1.65 MB; the network test
executable is 584 KB. A compile-time check proves the owned TLS 1.3-only
configuration (TLS 1.2, DTLS, renegotiation, tickets, PSK modes removed) applies.
Executed:
- Windows MSVC 19.44 Release: `tools/windows-check.ps1 -Preset network -Action
check` 1/1; `-Preset client` 5/5 after moving the SDL-free client value targets
to the top-level CMake file.
- Linux GCC 14 Release (`build/network-linux`): network suite passed with no
owned-code warnings; existing client build 7/7.
- Clang 19 Debug ASan/UBSan (`build/network-asan`): network suite passed.
- libFuzzer (`evo-dev-session-fuzz`, ASan/UBSan, empty seed corpus, max_len 8192):
1,272,671 runs in 61 s (cov 685, 158-entry corpus), then 269,854 runs in 31 s
after the analysis fixes; no crash, leak or canonical re-encoding abort.
- Seeded mutation fuzz inside the suite: 30,000 rounds, 8,208 accepted frames, each
re-encoded byte-identically.
- Tests also cover: explicit little-endian Intent golden; truncated, bad magic,
version 0/2, unknown type, flags, oversized, trailing and wrong-type frames;
unsupported required features; grant length bounds; invalid enum bytes; views
failing presentation rules; 257-cell counts; byte-by-byte reassembly and early
oversized-header rejection with latching; dev keys refused in stage/world0,
short/zero keys, every single-bit grant change rejected, foreign key, moved-from
key; admission world/region/lifetime/skew/expiry/reuse/duplicate/capacity/
grant-table rules, every IntentStatus, rate windows and epoch advance; TLS 1.3
loopback (version, suite, frame and 100 kB transfers with 700-byte partial
writes, close_notify), untrusted CA, wrong server name, mismatched private key,
bad PEM, tampered record, transport failure and plaintext rejection.
- clang-tidy 19 (warnings as errors) on the seven network translation units:
passed after restructuring enum decoding to cast only validated bytes, reading
the latched TLS error once, checking a test return value, and annotating
intentional invalid-enum/use-after-move test inputs. Repository clang-format
passed. CI path-selector fixtures passed including the new `network` suite.
- Headless: a fresh GCC 14 Release configure with default options contained no
Mbed TLS or SDL references; `evo-sim` replay of seed 123456 / 8 entities / 1000
ticks gave unchanged `d61ad5bf0c928c76`, `replay_verified=true`.
Limits: the `ci-network` Docker stage and Gitea matrix entry were not executed
locally. No sockets, server integration, secret provisioning or deployment exist
yet (M9.7). TLS interoperability was tested only between Mbed TLS endpoints. No
performance or latency measurement is claimed. Certificates in tests are
throwaway; no dev CA or HMAC key was created or committed.
## M9.7 connection to one authoritative dev region — 2026-10-01
Implemented ADR-0083 and DEV_SESSION_OPERATIONS.md: non-blocking TCP (Winsock/
POSIX), `SessionServer` in evo-server's loop, `SessionClient`, evo-client remote mode,
`evo-dev-session credentials|grants|probe`, the drill-time dev deployment override
and `tools/ops/session-drill.sh`. Maintainer decisions (2026-10-01): proceed although
M8-dev/stage is not passed; durable no-effect intentions; explicitly empty organism
presentation; local real-process evidence plus a dev deployment path.
Executed:
- Windows MSVC 19.44 Release: `-Preset network` 1/1 and `-Preset client-network`
6/6 (network, client environment/organisms/lifecycle/smoke/init-failure). The
network suite now includes four loopback session scenarios over real Winsock + TLS:
connect/view/intent/durable/not-committed/replay/invalid vector; duplicate
subject, reused grant, foreign key and region, capacity; stale epoch with unknown
outcome, stale layers, fresh-grant reconnect and fresh baseline, and a server
"restart" on the same port at a new epoch with an earlier tick; not-ready refusal
without spending the grant, byte-rate flood closure (server-side decision),
plaintext and silent connections dropped (the latter at the 200 ms deadline).
- `evo-dev-session` on Windows: credentials written, overwrite refused (exit 1),
`--environment stage` refused (exit 2), 4-grant bundle = 432 bytes.
- Linux GCC 14 Release with PostgreSQL 18.6 + network (`build/m97-postgres`): network
suite passed; the **full** `evo-postgres-tests` suite passed (adapter, lease,
checkpoint, host, server process, the new region dev session test, recovery, ops,
clone, store, load, upgrade, migration). The session process test runs real
evo-server processes: dev-only and synthetic-carrier refusals (exit 2), TLS
session, server-store view and empty organism layer, admitted intention whose
`INT1...` request appears in the host's `ack` line at the reported tick/commit,
replay rejected, duplicate session refused, SIGKILL with the client reporting
loss and stale layers, successor on the same port ready after lease expiry,
reconnect at epoch 2 and a second durable outcome with a higher commit.
- End-to-end rehearsal on the development-v1 region (world 100, region 1, frame 9;
`evo-world-clone provision`) with tool-generated credentials: probe durable ack
(tick 11, commit 1); SIGKILL; successor Ready in 3472 ms at epoch 2 from committed
tick 11; two `server_shutdown` refusals during recovery that did not spend grants;
reconnect and second durable ack (tick 22, commit 3); `probe passed`. The
**Windows** evo-client (`--connect` through WSL2 localhost forwarding, software
renderer, 300 frames) then received `session welcome session=2 epoch=2 tick=45
commit=42` and exited 0. (The rehearsal script's `server2 exit=127` is a `wait`
on a non-child PID in the script, not a server exit status.)
- Clang 19 ASan/UBSan: network suite passed. clang-tidy 19 (warnings as errors):
network library, session tests, dev tool, evo-client and evo-server/main plus the
session process test all clean, after consuming best-effort results explicitly,
a `must()` test helper for required optionals, and a GCC-reported genuine
dangling reference in the process test fixed. Format check, CI selector fixtures
and deploy policy (67 objects, 0 failures) passed; a temporary violation
(`REGION_SESSION=dev`, `0.0.0.0:7400`) produced exactly the two new failures.
Defects found and fixed during validation: grants were spent on `server_shutdown`/
`capacity` refusals (exhausting reconnection while a successor recovered); the
session process test reused region ids 41/42 owned by the recovery suite; Mbed TLS's
own `-Werror` failed on a GCC 13 `-O3` false positive (upstream warnings no longer
fatal; context serialization compiled out).
Limits: **no in-cluster evidence**. This workstation has no kubeconfig and Docker
was stopped, so no image was built or pushed, nothing was applied to evolution-dev,
and `tools/ops/session-drill.sh` has only been syntax-checked (no shellcheck
available). The CI `network`, `client` and `postgres` stages with the new options
were not executed remotely. Port-forward reachability of the pod-loopback listener
is the documented design, not yet observed. M8-dev/stage remains not passed. The
connected desktop run was hidden/software-rendered; no human visual check of remote
mode. Intentions have no biological effect by design.
## M9.4–M9.7 review corrections — 2026-10-02
Responds to the review of `bb0412c..8e31932` (`.cache/reviews/M9.4-M9.7-review.md`,
not in Git). All five findings were confirmed against the code and fixed:
1. **[P1] TLS retry lost queued frames.** `TlsChannel::send` is replaced by a bounded
(1 MiB) `queue` + `flush`: an interrupted record is retried with identical
arguments, later frames wait behind it, `close()` flushes before close_notify, a
full queue returns `backpressure` without queuing. Server and client route every
frame through it. Regression (`interrupted_writes`): zero progress and partial
progress then zero, frames queued during the stall, both directions, exact ordered
sequences; close while stalled; backpressure.
2. **[P1] Carrier ids aliased.** Carriers are server-issued `'I','N' | 48-bit epoch |
64-bit ordinal`; the client sequence is kept separately. Epochs beyond 48 bits are
refused at Hello; ordinal exhaustion stops carrying. ADR-0083's low-32-bit packing
is withdrawn. Regression (`carriers_never_alias`): sequences 1 and 2^32+1 get
distinct ids; `not_committed` and `committed` reach their own sequences.
3. **[P2] Grants reusable after restart.** New durable `GrantLedger` (fsync before
Welcome, atomic prune on open, fail closed on corruption/I-O/65536 entries), loaded
into admission by a replacement process; `--session-grant-ledger` (dev:
`/objects/dev-session-grants.v1`). Regressions: restart with the same ledger
refuses the consumed grant and admits a fresh one; corrupt ledger refuses server
creation; expired entries pruned; and in the end-to-end rehearsal the successor of
a SIGKILLed evo-server refused the first, consumed grant (`closed reason=3`).
4. **[P2] No liveness after Welcome.** Protocol v1 (still undeployed) gains Heartbeat
(type 9) and Close `timeout` (10). Both sides heartbeat after 1 s without output and
treat 5 s of silence as dead (server closes `timeout`, freeing the subject; client
marks stale, reports unknown outcomes, reconnects). Regressions (`liveness`): idle
but healthy session survives 1.5 s with no ticks; a silent server is detected and
reconnected to a fresh baseline; a silent client is closed and its subject reused.
5. **[P1, inherited] ci-client C compiler.** `-DCMAKE_C_COMPILER=clang-19` added; all
other C-enabling fresh configures already name it.
Follow-ups taken: `SessionClient::send` is bounded (256 unresolved, channel queue
capacity) and refuses instead of buffering; unresolved tracking now starts at send,
so a loss before IntentResult is also reported as unknown. The abuse test now floods
through a raw TLS peer (the client can no longer flood) and checks the server's
`rate_limited` decision. Client event-vector bounds remain M9.8 work.
Executed after the fixes: Windows MSVC `network` 1/1 and `client-network` 6/6; Linux
GCC 14 network suite and the **full** PostgreSQL integration suite (all suites,
including the session process test with the ledger and new carrier ids); Clang 19
ASan/UBSan network suite; clang-tidy 19 (warnings as errors) on all changed network,
client and server units (one `string_view::data` finding fixed in the ledger parser);
libFuzzer on the amended codec, 523,479 runs in 61 s, no crash or canonicality abort;
repository format, CI selector fixtures and deploy policy (67 objects, 0 failures).
End-to-end rehearsal: durable ack, SIGKILL, successor Ready in 3472 ms at epoch 2,
grants kept across two `server_shutdown` refusals, second durable ack, consumed grant
refused by the successor, Windows evo-client welcomed (`session=2 epoch=2`), successor
drained with `exit 0`.
Not done: the reviewer's own reproduction binaries target the removed `send` API and
were not rebuilt; equivalent regressions above replace them. The `ci-client` Docker
image and Gitea pipeline were not rerun here, and no cluster operation was performed.
## 2026-10-02 — consumed-grant ledger follow-up corrections
Resolved the two c588be2 review findings. Ledger open now acquires a nonblocking
exclusive lifetime lock on a stable sibling file before reading or compacting
(POSIX flock; Windows exclusive-sharing handle). Duplicate starts on either the
same port or a different port cannot replace the active ledger. The parser caps
records at 37 bytes before LF and rejects every nonempty incomplete final record
before compaction, preserving the original bytes. No ledger format migration or
simulation/protocol version change; valid existing files remain compatible.
Regression evidence: competing SessionServer starts fail with a ledger error;
the original server subsequently admits a grant, restarts, and rejects its replay.
Direct competing opens return busy. Every nonempty truncation of a valid final
record (including shortened numeric expiries) and an oversized line is rejected
without changing the file or creating a compaction temporary. Complete/empty files
and existing expired-entry compaction remain accepted.
Executed: Ubuntu GCC14 network CTest 1/1; Clang19 ASan/UBSan network CTest 1/1;
Windows MSVC network preset configure/build/CTest 1/1; clang-tidy19 on the changed
implementation and session tests with warnings as errors, no owned diagnostics;
clang-format19 on changed C++ files; git diff --check. The first restricted Windows
Ninja invocation stalled without compiling; its task-owned build processes were
stopped, then the same preset passed outside the restricted process environment.
Not rerun: Docker/Gitea, full PostgreSQL process suite, fuzzing, TSan, or live cluster
operations. No deployment performed. The text ledger detects malformed framing,
not syntactically valid corruption or deletion of entire records; ADR-0083 now states
that integrity limit explicitly. Client event-list bounds remain M9.8 work.
## 2026-10-02 — M9.8 decision capture and delivery segmentation
Recorded confirmed platform/distribution/UI choices and deferred VM/cluster evidence
in M9_8_DELIVERY_PLAN.md. ADR-0084 proposes the UI boundary and comparison; no library
version or dependency was adopted. Reviewed upstream toolkit documentation, checked
repository-relative links and git diff --check. Documentation only: no code tests,
package build, runner execution, endpoint verification or deployment performed.
## 2026-10-02 — M9.8a pinned engineering UI probe
Adopted Dear ImGui 1.92.9b for an isolated SDL3/SDL_Renderer3 compatibility probe;
normal client/headless builds do not require it. Exact upstream commit, locally
measured archive checksum, maintenance and notice evidence are in ADR-0084 and
third_party/imgui/README.md. No upstream code modified. Embedded bitmap font used;
no extra font dependency. Source preparation is explicit, with hash verification.
Executed: Windows MSVC Release probe CTest 2/2; Linux GCC14 Release 2/2;
Clang19 ASan/UBSan 2/2 (including instrumented toolkit sources); real X11 software
renderer probes under Xvfb in Release and sanitizer builds. Three full SDL/UI
initialize/render/shutdown cycles per successful invocation. A synthetic press and
release activates a real widget exactly once; geometry bounds and pixel readback
are checked under NDEBUG. Invalid explicit renderer fails. This is synthetic input
and render evidence, not physical keyboard/DPI/accessibility acceptance.
clang-tidy19 on owned probe: no owned diagnostics; clang-format19 and CI selector
fixtures passed, including a new toolkit-pin selection case. Fresh headless configure
with client/network/probes off passed without evaluating SDL/ImGui dependencies.
An initial const mismatch with upstream's mutable draw API was corrected. Initial
static analysis used a GCC database with unsupported module flags; rerun against the
Clang database passed. Edited shell scripts were normalized to LF before selector
fixtures ran successfully.
Artifact observations: Windows Release probe EXE 398336 bytes, SDL3.dll 1998848;
Linux Release executable 1098464 bytes. These are probe totals, not a final client
package size or an incremental apples-to-apples benchmark. No minimum hardware,
frame-rate or accessibility claim yet. Low-spec VM pending; live cluster drill is
explicitly deferred; control-plane IP is not assumed to be a MetalLB endpoint.
Not run: Docker/Gitea image execution, physical GPU/DPI matrix, VM soak, client UI
integration/launcher or packaging acceptance. Existing network and headless simulation
behavior is unchanged; no simulation version or golden was modified.
## 2026-10-02 — M9.8b owned UI state and bounded session events
Implemented toolkit-independent UiState: eight expiring/dismissible notices,
240-byte messages, an observable eviction count, one id-matched modal and explicit
monotonic time. Tests cover 10000-notice bursts, expiry boundary, malformed/oversized
messages, stale time, deadline overflow, modal contention and stale close IDs.
External native UI capture can no longer be escaped through the legacy Tab toggle.
SessionClient queues at most 1024 events, reserves loss/unknown capacity, preserves
queued results, and disconnects on backlog until take_events drains it. The burst
regression accounts for every sent intention as a rejection or unknown, proves no
grant burning while paused and confirms reconnect after drain. Backpressure is a
local event, not a new wire message or simulation rule. Presets and the selective
Docker client targets include the new UI state test executable.
Executed: Windows MSVC client-network configure/build/CTest 7/7; Linux GCC targeted
UI-state/network/native-lifecycle tests 3/3; Clang ASan/UBSan UI-state/network 2/2.
Static analysis initially flagged the new test main's escaping exception; added the
same process error boundary used by other executables. Windows exposed a timing
assumption in the initial one-message-per-sleep burst test; bounded batches replaced
that dependency. The explicit-target Windows preset initially omitted the new test
binary; fixed its target list and reran all seven tests successfully.
Visible dialogs/toasts, launcher and toolkit focus routing remain M9.8c. These are
owned state and adapter contracts, not a claim of finished UI. No Docker/Gitea run,
VM qualification, live cluster test, packaging or deployment was performed.
Final additional checks: Linux GCC state/network/lifecycle 3/3; Clang ASan/UBSan
native lifecycle 1/1; clang-tidy19 on owned state/session implementation,
state/network tests and native adapter/lifecycle tests passed with no owned
diagnostics. Final changed-file formatting passed.
## 2026-10-02 — M9.8c launcher, legend, labelled inspector and click selection
Implemented toolkit-independent launcher, selection, label and outcome-log models
(src/client/ui) and the toolkit overlay in the native adapter (events routed to the
toolkit first; its keyboard/pointer capture suppresses world keys, wheel and clicks).
The app panels show a labelled local fixture vs dev region, legend, inspector,
session status with the last ten intention outcomes, notifications and a disconnect
dialog. `--inspect` is replaced by click selection; `--pick`/`--screenshot` support
unattended review. Added `SessionClient::unreachable_attempts()` (additive).
Tests: `client-ui-model` (endpoint/field bounds, file size/shape limits, fixed error
text without paths, launcher phases incl. cancel/retry, three-loss bound, refused
grant, exhausted bundle; selection topmost/cell/none, disappearance clears, missing
view keeps; value/zero/missing/not-requested/undefined distinct; outcome log bounds).
Lifecycle `toolkit_capture`: widget click is not a world click, drag is not a click,
focused text field captures keys, release needs fresh presses, screenshot hook.
Network `unreachable_server`: refused connects emit no events, spend no grant, count.
Executed: Windows MSVC `client` 7/7, `network` 1/1, `client-network` 8/8 (final run
after all fixes). Linux: clang-19 client-probe with clang-tidy enabled (keep-going
build of all client targets, no owned diagnostics after fixes) and CTest 11/11;
Xvfb/X11 lifecycle, fixture (`--pick`) and launcher runs; Clang ASan/UBSan client
CTest 11/11, Xvfb lifecycle and fixture; GCC 14 client-network build and client|network
CTest 8/8; Clang ASan/UBSan network tests; clang-tidy on session client, network tests,
app main/panels (client-network); repository format check.
Visual review (dummy driver, software renderer, BMP screenshots): launcher; fixture
with organism selected; fixture cell on the nitrogen layer (missing, present zero and
not-requested rows distinct); connect to a closed port; live dev region (WSL
development-v1 server via the M9.7 rehearsal script) with the cell inspector and
Session panel. Review found and fixed: inspector values clipped (now wrapped),
legend hint clipped, launcher error line overflow, Session table column mismatch,
and a closed port staying "Connecting" indefinitely because pre-session failures
produced no event (now bounded via the new counter; observed failing with Retry/
Cancel). Windows reports a refused loopback connect after about two seconds.
Static tidy findings fixed: implicit widening in a lifecycle test, by-value lambda
parameter in the UI-model test.
Limits: no manual interactive session with physical mouse/keyboard, DPI matrix,
screen reader or accessibility review; the remote review used one connect without
WASD intentions, so the outcome list was empty in the screenshot (outcome states are
covered by unit tests). No Docker/Gitea run, packaging, deployment or cluster test.
No simulation version, wire format or golden changed.
## 2026-10-02 — M9.8c review corrections (.cache/reviews/m98c-review.md)
1. Same-batch panel input: pointer ownership is now resolved per SDL event from the
toolkit's retained window rectangles/popup stack (pinned internal API, ADR-0084);
a press on a panel owns the button until release and drops world key-downs for
the rest of that batch. Regressions without settling frames: world->panel move +
press/release (no world click, button still activates), move + wheel (no zoom;
world wheel still zooms), click-to-focus + W in one batch (no movement; field
captures; fresh press after unfocus moves).
2. Fixture/live distinction: `SessionHost` owns launcher phase and remote session;
fixture, cancel, disconnect and failure drop the session; it is serviced only
while connecting/connected. `client-ui-model` uses a fake session/server:
connecting->fixture and failed->fixture followed by an available server produce
no steps, Welcome (grant use) or intentions; ordinary connect still works.
3. Tab capture: an external release no longer cancels it. Regression drives the
host order (poll, set_ui_capture(false), sample, present) over several frames,
requires capture until the second toggle and a fresh, non-repeat press after.
4. Notices: wrapped text beside a reserved 22-pixel close column, width min(320,
viewport-16), stack bounded to 60 % of the height with a not-shown/discarded
count. New `client-panels` renders the production panels through the adapter:
the three real session notices and a 240-byte message fit (content width and no
horizontal scroll), the close button clicked in one batch dismisses exactly that
notice with no world click; a full stack stays on screen.
Negative checks: the new lifecycle test fails against the previous adapter (same-
batch click), the panel test against the previous notice layout (content width),
and the host-order Tab test with the former clearing restored (Windows MSVC).
A GCC-only failure in the new fake-session test was a test bug (by-value argument
alive until the full expression ends); split the assertion.
Executed: Windows MSVC client 8/8 and client-network 9/9 (final). Linux: clang-19
client-probe build with clang-tidy enabled (keep-going, no owned diagnostics) and
CTest 12/12; Xvfb/X11 lifecycle and panel tests; Clang ASan/UBSan client CTest 12/12
and Xvfb lifecycle/panels; GCC 14 client-network build, client|network CTest 9/9;
clang-tidy on app main/panels (client-network); repository format check. Visual:
live WSL dev region showing the wrapped connection notice with its close button.
Limits unchanged: no physical mouse/keyboard session, DPI or accessibility review;
the live screenshot sent no intentions. No Docker/Gitea, packaging or deployment.
## 2026-10-02 — M9.8d rebindable controls (ADR-0085)
Implemented SDL-free bindings (nine actions, 92 named HID key positions, Escape
reserved, modifiers/locks/system keys excluded), a staging editor with conflict
refusal, and a versioned all-or-nothing controls file (4096 bytes, printable ASCII,
temp-file replace). The adapter maps keys through the bindings, releases and drains
on `set_bindings`, and offers binding capture ahead of the toolkit and world. The
app adds a Controls panel (launcher and banner), Legend key hints, startup loading
with a fixed-text notice on rejection, and `--controls FILE` / `--show-controls`.
Tests: `client-bindings` (names, defaults, conflicts incl. own secondary, missing
primary, reserved/modifier keys, editor refusal leaves draft unchanged, reset,
serialize/parse round trip, comments/CRLF/tabs/order, unknown/duplicate/missing
actions, bad names, extra tokens, non-ASCII/NUL, size bound, unsupported version,
file load/save/replace, unreadable/too large/invalid/missing directory). Lifecycle
`rebinding`: rebinding while held releases and drains (new key queued before the
change does not act), unbound old key ignored, two keys per action never stick,
rebound toggle, capture consumes the key (repeat ignored, toolkit never sees it,
rest of batch drained), Escape returned, focus loss and cancel give no key.
`client-panels` controls panel: fits viewport, a slot click requests capture for
Move left without a world click, chosen key consumed before the toolkit.
Executed: Windows MSVC client 9/9 and client-network 10/10 (final). Linux: clang-19
client-probe with clang-tidy enabled (keep-going; one test finding fixed) CTest
13/13; Xvfb/X11 lifecycle and panels; Clang ASan/UBSan CTest 13/13 plus Xvfb
lifecycle/panels; GCC 14 client-network CTest 10/10; clang-tidy app main/panels
(client-network); repository format check.
Visual review (dummy driver, software renderer): Controls panel with a custom file
(J/L/I/K, F1 + Keypad0) and matching Legend hint; invalid file (missing actions)
showing "Saved controls rejected: Move left needs a primary key..." with defaults in
use and the file byte-identical afterwards. Review found the startup notice expired
immediately (UI clock not yet advanced); fixed. A CRLF parsing defect was found by
the new test and fixed. CTest smoke runs create the empty per-user settings
directory (SDL creates it on lookup); no controls file is written by tests.
Not run: interactive rebinding with a physical keyboard, non-US layouts, IME, the
Apply/save path end-to-end through the GUI (covered by file and editor tests), DPI,
accessibility, Docker/Gitea, packaging or deployment.
## 2026-10-02 — M9.8d review corrections (.cache/reviews/m98d-review.md)
1. Toolkit keys held before capture: `begin_key_capture` drops the toolkit's key and
modifier state plus queued keyboard/text events. 2. Chosen key ownership: the key
stays owned across polls until release or focus loss; repeats, text and release
are consumed. Also fixed (found while writing the regression): without an overlay
or while paused the toolkit queue was never consumed; idle toolkit frames run and
the queue is bounded at 512 events (lossy for the toolkit, never a stuck press).
Tests now use realistic key events (scancode, keycode from the US fixture, down,
modifiers) in lifecycle and panel suites. New lifecycle `capture_balances_toolkit`:
positive controls (Q and Left Shift press and release in the toolkit), Shift+Q+W held
then released during capture under external capture for choose, cancel and focus
loss (all up in toolkit and world), chosen J held over three polls with repeats and
text (no queued toolkit input, J never down), fresh J afterwards reaches the toolkit,
owned key freed by focus loss, a 2000-event flood while hidden stays within the bound
and ends released. New panel `apply_save_restart`: real adapter, production panels
and `ControlsFlow` host logic; slot click arms capture, Up is refused as Pan up's key
(draft keeps A), J chosen, Apply via a click writes the file and shows "Controls
saved."; a restarted application loads it and J moves left.
Negative checks (Windows MSVC): removing the toolkit reset fails the held-key check;
removing key ownership fails the repeat/text check.
Executed: Windows MSVC client 9/9, client-network 10/10 (final, after formatting).
Linux: clang-19 client-probe with clang-tidy (keep-going, no owned diagnostics)
CTest 13/13; Xvfb/X11 lifecycle and panels; Clang ASan/UBSan CTest 13/13 plus Xvfb
lifecycle/panels; GCC 14 client-network CTest 10/10; clang-tidy app main/panels;
repository format check.
Limits: physical keyboards, non-US layouts and IME remain unqualified; the Apply
rehearsal is automated with synthetic events, not a manual session.
## 2026-10-02 — M9.8d follow-up (.cache/reviews/m98d-followup-review.md)
Owned capture keys are now a bounded set (bitset over SDL scancodes) instead of one
optional key, so a second capture cannot overwrite an unreleased earlier choice.
Lifecycle regressions in `capture_balances_toolkit`: W then K chosen with K released
first (W repeat/text reach neither toolkit nor world until W's release, then a
fresh W moves and reaches the toolkit); reverse release order (K stays owned after
J's release; text flows again only when nothing is owned); cancel during a second
capture keeps the first key owned; focus loss frees both; positive controls after
each. Negative check: restoring single-key ownership fails the W repeat check.
Executed: Windows MSVC client 9/9 and client-network 10/10 (final, after
formatting). Linux: clang-19 client-probe with clang-tidy (keep-going, no owned
diagnostics) CTest 13/13; Xvfb lifecycle and panels; Clang ASan/UBSan CTest 13/13
plus Xvfb lifecycle/panels; GCC 14 client-network CTest 10/10; clang-tidy app
main/panels; format check. Synthetic event sequences only: physical OS repeat with
several held keys, non-US layouts and IME remain unqualified.
## 2026-10-02 — M9.8e Windows client package (ADR-0086)
Implemented the `client-package` preset (static MSVC runtime, `/Brepro`), the
`EVO_BUILD_IDENTITY` cache value (validated, printed by `evo-client --version`),
`tools/client/package-windows.ps1`, `verify-windows-package.ps1`,
`package-selftest.ps1` and shared guards. Decisions confirmed by the user: static
runtime; local packaging only (no workflow until a Windows runner exists); the
README states an internal test build with no licence granted.
Import audit: the default `/MD` client imported MSVCP140/VCRUNTIME140(_1) and
api-ms-win-crt DLLs; the package build imports only KERNEL32, USER32, SHELL32,
IMM32, bcrypt, WS2_32 and SDL3.dll, and SDL3.dll only Windows system DLLs. Notice
audit of the compiled SDL subset found HIDAPI, yuv2rgb, stb_image and Khronos
headers in addition to SDL's zlib licence; all are reproduced from the verified
sources together with Mbed TLS (Apache-2.0 option) and Dear ImGui (MIT, embedded
ProggyClean/ProggyForever notices, stb headers).
Executed on the Windows development host (MSVC 19.44.35217, CMake 3.31.6):
- Package at committed `ec1dced`: client-package CTest 10/10 inside the script;
identity `gec1dcedf2e66`, ZIP 1,716,303 bytes.
- Reproducibility: two `-Clean` builds into separate output directories (612 s
each) gave the identical ZIP SHA-256
c0c27b948e693ac199f693f1c3ae98a7c38adefa593053273c00b41c884100a8. Same machine,
path and toolchain only; other machines/toolchains are not claimed.
- Verification of that ZIP: digest, exact allowlist and SHA256SUMS, PE imports,
credential/profile-path scan, then the extracted client with only system
directories on PATH: --version, usage error (exit 2), dummy-driver fixture (60
frames) and launcher (30 frames), missing CA file (exit 1, fixed text, no path),
and `-Interactive` (default driver/renderer, real window, 180 frames); no files
were written into the package folder and no controls file without Apply.
- Self-test: extra/missing file, checksum mismatch, .pem and .bin files, private key
and certificate text, user-profile path and a dynamic-runtime image were each
refused.
- Earlier dirty-tree runs found and fixed: native stderr from the child build being
fatal under Windows PowerShell 5.1, the font notice extraction matching the file's
table of contents (619 KB copied), and checkout-dependent line endings of the
notice index.
- Regular presets after the change: Windows client 9/9, client-network 10/10;
Linux clang-19 tidy build 13/13, ASan/UBSan 13/13 with Xvfb, GCC 14 10/10, tidy
of app main/panels, format check; `client-package` configures on Linux; an
invalid identity is rejected at configure.
Not done: Gitea Actions publication (no Windows runner), a run on a clean Windows
VM (M9.8g), code signing, hardware/DPI qualification, legal review of notices.
## 2026-10-02 — M9.8f dev session route design (ADR-0087)
Maintainer decisions: dedicated internal MetalLB VIP, TCP 7400,
void.aeternalisgames.com, TLS ending in the region pod, development/test CIDR only
(firewall + Service source ranges + NetworkPolicy), no Ingress/nginx TLS/PROXY
protocol; operator-opened windows with server-enforced absolute expiry; scope limited
to design, manifests and policy tests. Nothing was applied to the cluster.
Implemented: `REGION_SESSION_LISTEN` (base stays 127.0.0.1:7400); the disabled
component `deploy/kubernetes/region/dev-session-route` (LoadBalancer Service with
Local policy, no node ports, pinned VIP annotation and source ranges; CIDR-only
ingress policy on TCP 7400; listener patch to the pod interface) carrying REPLACE
markers; policy rules for the route's exact shape, LoadBalancer/NodePort and
session-port exclusivity, no session Ingress, evolution-dev-only placement, quota
limits, and enabling preconditions (no markers, dev quota of exactly one
LoadBalancer, server `--session-window`); patch files are now recognized generally.
`tools/ci/deploy-policy-selftest.py` copies the manifests and applies 20 unsafe changes.
Executed (WSL Ubuntu, Python 3 with PyYAML 6.0.1): deploy policy check on the
repository, 70 objects in 68 files, 0 failures; self-test 20/20 refused, baseline
copy passes; ops-lib and ci-path fixtures pass; session-drill usage unchanged.
kustomize v5.4.3 (official release, SHA-256 checked against its checksums.txt, in a
scratch directory, not a project dependency) rendered region/dev as committed (listener
127.0.0.1:7400, no session Service) and a temporary enabled copy with example private
values (LoadBalancer evo-region-session and the CIDR policy in evolution-dev, listener
0.0.0.0:7400, session container port merged).
Not run: the ci-deploy Docker stage (no Docker in this WSL distribution; its commands
were run directly), any cluster/server dry-run, MetalLB allocation, DNS, firewall or live
connectivity. Server-enforced session windows and the Gitea window workflow are not
implemented; Argo delivery for windows and the existing void HTTP Ingress are open
maintainer decisions recorded in ADR-0087.
## 2026-10-02 — M9.8f review corrections (.cache/reviews/M9.8f-9f622cf-review.md)
P1 (rendered configuration): the policy check now renders all 17 non-component
kustomizations with pinned kustomize v5.8.1 (ADR-0088; official archive, SHA-256
029a7f0f…a819d checked against the release checksums) and validates the effective
objects: overlay namespaces, evo-region session/input/listener environment, a single
listener argument, container ports, Service types and session port, route shape, session
NetworkPolicies, quotas, Ingresses and no Secrets. Unsupported kustomize features
(plugin generators/transformers, Helm, Secret generators) are refused. Route enabling is
gated by `ROUTE_ENABLING_PERMITTED = False`, changed only with an ADR-0087 amendment
recording the implemented window contract and runtime expiry tests; the earlier
`--session-window` string test is gone. Route addresses must be RFC 1918 or listed in
`APPROVED_INTERNAL` (empty; 172.10.1.x is not RFC 1918).
P2 (health-check port): manifest comment and ADR-0087 now state that the Local policy
allocates `healthCheckNodePort`; allocation, quota accounting and firewall handling are
enabling preconditions. Review recommendations (manual Argo sync, preserving the void
endpoint, address confirmation, window limits and tests) are recorded in ADR-0087 as
pending maintainer decisions. The M9.8f gate stays open.
Executed (WSL Ubuntu, Python 3, PyYAML 6.0.1, kustomize v5.8.1 via the new prepare
script, idempotent on rerun): policy check 70 source objects / 126 rendered objects in
17 overlays, 0 failures; self-test 33/33 refused (20 earlier cases plus non-RFC 1918 VIP,
the reviewer's combined inline patches, inline listener/session/input patches, JSON
patches adding a session port, a LoadBalancer, a container port and a second listener
argument, namespace transformation, inline stage quota, Secret generator, Helm inflation);
the reviewer's m98f-policy-repro.py now exits 1 with the listener, session/input and
ops-port failures in both region/dev and dev; the check fails without kustomize and with
kustomize v5.4.3; ops-lib and ci-path fixtures pass.
Not run: the ci-deploy Docker stage (no Docker here; its commands were run directly,
including prepare-kustomize.sh), any cluster dry-run or live action.
## 2026-10-02 — M9.8f review of 8a1e5ef (.cache/reviews/M9.8f-8a1e5ef-review.md)
P2: rendered NetworkPolicies are now checked by effect: the default-deny contract in
every environment namespace, and every ingress rule able to select a region pod
(empty/expression selectors included) is refused if it admits the session port (no or
empty ports, port-less TCP entries, the `session` name, 7400, ranges spanning it), except
the fully validated route policy. Activation follow-up done now: the route Service and
policy validators (VIP, source ranges, annotations, selector, shape) are shared by the
source and rendered checks.
Executed (WSL, kustomize v5.8.1): policy check 0 failures (126 rendered objects, existing
ops/metrics scrape rule accepted); self-test 44/44 refused, adding the reviewer's
all-ingress default-deny patch, region policies without ports, with empty ports, by
the session name, by a 7000-8000 range, an every-pod policy on 7400, an expression
selector, a port-less entry, and rendered route patches of the selector, VIP and source
range; the reviewer's m98f-8a1e5ef-repro.py now exits 1 with default-deny and
session-port failures in platform/dev and platform/cluster. No cluster action. M9.8f
remains open.
## 2026-10-03 — M10.1 protocol timing and workload budgets (ADR-0089)
Maintainer ratified: 50 ms tick, 1 s flush with final outcomes p95 <= 1.5 s,
responsiveness LAN p95 <= 150 ms / 100 ms RTT p95 <= 250 ms, server budgets on one RKE2
worker and client budgets on the low-spec VM. Derived budgets and the exact/tolerant
determinism contract are in ADR-0089 and schemas/network/m10-budgets-v1.md.
Implemented: `SessionClient::traffic()` received-frame counters (additive API, checked
in network_session); `evo-dev-session trace` (continuous polling, nearest-rank
percentiles, CSV); `tools/network/session-trace.sh` (throwaway PostgreSQL, development-v1
region, evo-server with sessions and ops metrics, three loopback runs, cross-host pause).
Traces (development reference: Ryzen 7 9800X3D, 61.6 GB, Windows 11; WSL2 server):
committed-script run with Linux loopback at 250/100/1000 ms and Windows -> WSL2 at
250 ms: admission p50 0.9-1.4 ms, p99 <= 5.2 ms; durable p50 0.92-1.06 s, p95
1.37-1.41 s, max 1.46 s; outcome lead 9 ticks; view interval p50 50.1 ms, p99 <= 60 ms;
267-byte views, about 7.5 KB/s; tick debt max 0, server about 1 % CPU, 10.5 MB RSS. Two
earlier runs matched within about 1 ms. A 1 ms-sleep Windows client measured about
31 ms admission (Windows timer granularity); the trace now polls continuously.
Finding recorded in ADR-0089: genome v1 and metabolism rates are per tick, so the
quantum fixes biological rates; changing it needs a versioned rules migration (M11.1).
Executed checks: Windows MSVC network preset 1/1 (twice, after each change); Linux GCC 14
network tests and evo-dev-session build; Clang ASan/UBSan network tests; clang-tidy on
the session client, evo-dev-session and network_session tests; format check.
Not covered: emulated RTT/loss, multiple clients, the reference worker and low-spec VM,
movement/prediction; the targets are therefore not yet qualified (10.9).
## 2026-10-03 — M10.2a QUIC transport selection and build (ADR-0090)
Maintainer decisions: msquic; OpenSSL from source on Windows and Linux; M9 TLS/TCP kept as
dev/engineering fallback; owned UDP impairment relay for loss testing. Survey (GitHub
API, 2026-10-03): msquic v2.6.2 MIT, ngtcp2 v1.25.0 MIT, picoquic MIT, lsquic v4.10.0
MIT, quiche 0.30.0 BSD-2, GameNetworkingSockets v1.6.1 BSD-3, wolfSSL GPL-3.0 (excluded).
Pins (`tools/msquic-pin.cmake`, explicit `tools/prepare-msquic.cmake`): msquic 2.6.2
tag archive 206e4604... (two downloads identical); OpenSSL 3.5.9 603f5602... equal to the
release's published .sha256; xdp-for-windows d372b525 headers 9a52e1b5... (two downloads
identical). Strawberry Perl 5.42.3.1 and NASM 3.02 installed on the development host
with winget (installer hashes verified by winget, per the maintainer's go-ahead).
Executed: Linux standalone msquic build (GCC 14, 280 s, libmsquic.so 6.8 MB); preset
`quic` on Linux GCC 14 and Windows MSVC 19.44: `evo-quic-smoke-tests` and network tests
2/2 on both; smoke test repeated 5 times on each without failure; clang-tidy (clang 19)
on the smoke test clean; format check; Windows `network` 1/1 and `client-network` 10/10
after the windows-check PATH/compiler changes.
Found and fixed on Windows: a build-directory clash with the Linux run (separate
directories now), msquic's own `/WX` with MSVC `/analyze` C28020 in crypto_tls.c
(upstream-only `/WX-`), missing XDP headers (pinned), `ntdll` for msquic.h helpers, and
the Windows platform verifier ignoring `CaCertificateFile` (OpenSSL verifier selected
explicitly). Security finding recorded in ADR-0090: OpenSSL's built-in verification also
trusts its default verify paths; exact pinning is 10.2b work.
Not done: bounded channels, impairment relay, loss/HOL measurements, CI integration of
the quic preset, sanitizer builds of msquic, packaging of msquic.dll.
## 2026-10-03 — M10.2b bounded QUIC channels, exact pinning, impairment relay (ADR-0090)
Implemented `evo_network_quic` (Endpoint, MessageReassembler, framing),
`evo_network_impair` (UdpRelay) and `evo-quic-transport-tests` (label `quic`).
Tests: framing with 2,000 random-length messages under random chunking, oversized
frame refusal, empty messages; 100 ordered messages each way, datagrams each way, local
size limits, unknown connection, outbound reliable budget (busy), peer protocol
violation closing the connection; exact pinning (attacker CA via `SSL_CERT_FILE`
refused, genuine server still accepted with that environment, wrong name refused);
impairment through the relay (10 % loss, 10 % duplication, 20 % reordering, 5 %
corruption, 5 ms + 0-10 ms jitter both ways): 200/200 messages exactly once in order,
datagrams never duplicated and partly lost (262-290 of 300 across runs); 1,252-byte UDP
path: 1,200-byte datagrams and 1,100-byte messages cross it (6 larger PMTUD probes
dropped); total outage: both sides idle-closed after about 1.3 s (1 s idle timeout);
bounds: newest-wins datagram queue (at most 4 kept), connection cap refusal, reliable
inbound overflow closing as overloaded; 500 random UDP datagrams at the listener produce
no events and a genuine client still connects; clean local/peer close.
Found and fixed during development: a possible deadlock (msquic calls under our lock;
callbacks can run inline), OpenSSL's issuer lookup failing a genuine server when another
same-named CA was in `SSL_CERT_FILE` (owned pin made authoritative), msquic's 1,288-byte
default MTU exceeding a 1,280-byte path (MinimumMtu 1280), a flaky coverage assertion
(2 % corruption could produce none), two test bugs found by UBSan/assertions, Windows
`NOMINMAX`/`<array>` portability, and clang-tidy findings.
Executed: Linux GCC 14 preset `quic` CTest 3/3 and transport tests 5/5 repeated runs;
Clang 19 ASan/UBSan (Debug, leak detection) transport tests; clang-tidy on endpoint,
relay and tests; format check; Windows MSVC 19.44 preset `quic` 3/3 and transport tests
5/5 repeated runs.
Not done: loss/head-of-line measurements and TCP comparison (10.2c), msquic itself under
sanitizers, CI integration of the quic preset, integration into the region host/client.
## 2026-10-03 — M10.2c loss and head-of-line measurements (ADR-0090)
Added `evo-quic-hol-measure` (`tools/network/quic_hol.cpp`, preset `quic`) and ran it for
15 s per configuration (RTT {2, 50, 100} ms x loss {0, 1, 5, 10} %, 2 % reordering, 1 %
duplication, 16 KiB reliable messages) on Windows 11 (MSVC 19.44) and WSL2 Ubuntu (GCC 14)
on a Ryzen 7 9800X3D. Results: docs/networking/M10_TRANSPORT_MEASUREMENTS.md.
Found and fixed: Windows numbers first inflated to about 31 ms at 2 ms RTT by the ~15.6 ms
default timer tick (the relay and tool now call `timeBeginPeriod(1)`, relay balanced in
its destructor; Windows baseline 6 ms p95); datagrams refused before the DATAGRAM
extension was negotiated (measurement now waits for it); clang-tidy widening findings.
Executed: Linux format check, GCC 14 build, CTest `-L quic` 2/2, clang-tidy 19 on the tool
and relay; Windows MSVC preset `quic` build and CTest 3/3; both measurement runs.
Correction (later the same day): the Windows build preceded the final formatting pass,
which sorted `<timeapi.h>` before `<windows.h>`; the committed tool did not compile with
MSVC. Fixed in the next commit (separate include block); MSVC preset `quic` rebuilt, CTest
3/3 and a 2 s measurement run. The `-Action test` step of windows-check does not build:
always run `-Action build` first.
Limits: no kernel TCP/netem comparison (user-space relay by decision), single host and
loopback, development hardware, 15 samples of 16 KiB messages per configuration, run
length cuts off in-flight stream messages; msquic not under sanitizers; quic preset not
in CI.
## 2026-10-03 — M10.3a session grants, identity adapter and session control (ADR-0091)
Implemented `SessionControl` and `AccountId` (evo_network_protocol), ECDSA P-256 session
grants `GrantSigningKey`/`GrantVerifier` (evo_network_tls, Mbed TLS PSA) and the identity
adapter `evo_identity` (DevTestProvider, MemoryAccountDirectory, GrantIssuer); tests in
`tests/network_control.cpp` (part of `evo-network-tests`, label `network`).
Tests: sign/verify round trip; every one of the 160 grant bytes flipped is refused (key
id bytes as unknown key, others as bad signature); truncated and M9 dev grants refused;
dev-signed grants refused by stage and World0 verifiers with and without the dev public
key configured; key rotation, duplicate and excess keys, relabelled key id, invalid
point, key import and moved-from keys; DevTestProvider refused outside dev, credential
syntax; directory stability, capacity and serial/issue-time monotonicity; issuer end to
end into session control and environment mismatch; control scope (world, region,
environment), lifetime, skew and expiry; replay and older unexpired grants refused
(`stale_grant`), newest grant superseding the previous session, account and region
attempt limits with window reset, refused open not consuming the grant, account capacity
failing closed until older grants can no longer verify, refresh (extension, stale,
wrong account, unknown session), grant expiry and epoch change closing sessions; log label.
Executed: Linux GCC 14 preset `network` build and CTest `-L network` 1/1; Clang 19
ASan/UBSan (Debug, leak detection) `evo-network-tests`; clang-tidy 19 on the new sources
and tests; format check; Windows MSVC 19.44 preset `network` build and CTest 1/1.
Not done: QUIC session handshake (10.3b), durable control lease and serial persistence
across restarts and regions (10.3c), revocation, private-key provisioning tooling.
## 2026-10-03 — M10.3b session login over QUIC (ADR-0091)
Implemented the session-v1 codec (evo_network_protocol), `RegionSessions`
(evo_session_server) and `RegionSessionClient` (evo_session_client); tests
`evo-quic-session-tests` (label `quic`) and codec tests in `evo-network-tests`.
Codec: round trips of every message; every truncation, an extra byte, bad magic, other
version, flags, unknown types and invalid reasons/statuses/zero ids refused.
End to end on loopback QUIC with an injected wall clock: Welcome fields and holder;
replay (`stale_grant`); newer grant superseding the old connection (Close `superseded`);
older unexpired grant refused; refresh accepted, stale and wrong-account refusals with the
session kept; wrong region and stage-signed grant (`wrong_scope`), unknown issuer and a
modified grant (`unauthenticated`); expired grant at login and session expiry closing
only the expired session; client close releasing control; silent peer (`timeout`),
garbage, wrong first message and unknown required feature (`protocol`), other version
(`unsupported_version`); epoch advance (`stale_epoch`) and Welcome with the new epoch;
shutdown; capacity refusal not consuming the grant; region attempt limit.
Found and fixed during development: a connection cap of twice the session limit refused
new logins while refused connections were still flushing their Close (now the transport
limit, which must be at least the session limit).
Executed: Linux GCC 14 preset `quic` build and CTest `-L "quic|network"` 4/4; session
tests 10 consecutive runs 10/10; Clang 19 ASan/UBSan (Debug, leak detection) session and
network tests (msquic itself uninstrumented); clang-tidy 19 on the new sources and tests;
format check; Windows MSVC 19.44 preset `quic` build and CTest 4/4 (binary rebuilt and
timestamp checked).
Not done: wiring into evo-server and the engineering client, the durable control lease
(10.3c), an impairment-relay run of the login path.
## 2026-10-03 — M10.1 trace safety corrections
The trace runner now requires a new output directory, creates it with owner-only
permissions, and leaves existing paths untouched. A rejected intention now makes
`evo-dev-session trace` return failure; the runner records each local run's status and
fails the experiment if any run fails.
Executed: GCC 13 Linux Release build of `evo-dev-session` in an isolated
`build/network-wsl` directory; network CTest 1/1; clang-format-19 check for the changed
C++ file; Bash syntax check; temporary-directory check confirmed an existing output
directory is refused and its marker file preserved. `git diff --check` passed.
Not done: no end-to-end trace against a server deliberately configured to reject an
intention; no Docker/Gitea run. The return path is compile-checked but rejection is not
covered by a dedicated application integration test.
## 2026-10-03 — M10.3c durable account control (ADR-0091, schema v10)
Implemented migration v10 (`account_controls`, three fenced functions, apply chain,
privilege audit), the store adapter (`account_control_store.hpp` in evo_postgres), the
`ControlLedger` interface and `SessionControl::check`, `PostgresControlLedger`
(evo_control_ledger_postgres) and the ledger path of `RegionSessions`.
Database tests (`tests/account_control_store_tests.cpp`, world 9100, real runtime login):
runtime cannot read the table; first claim; replay and older serial refused; same-region
takeover without wait and fencing of the old generation; cross-region claim waiting
15-20 s of the previous 20 s control, fencing the old region's renewal and release; renewal
never shortening the waited expiry; release and renewal after release; durable serials
after a region restart under a new epoch with no wait for the older epoch and the old
lease fenced; database-time expiry ending control and not being waited for; invalid
durations, serials, sessions and the zero account refused; two regions presenting the
same serial concurrently with exactly one winner; the PostgreSQL ledger's claim, stale,
renew, release and missing-lease mapping. `postgres-check.sh`: history probes moved to
v11/v10 checksum drift; audit refuses SELECT on `account_controls` and a missing EXECUTE
on `claim_account_control`; migration tests expect history 1..10.
Session layer with a scripted ledger (`evo-quic-session-tests`): durable stale, fenced and
undecided claims refused without admission; claim names account, serial and the Welcome
session; periodic renewals; refresh records the newer serial; fenced renewal closes as
`superseded` without release; Welcome delayed by the claim's wait; undecided renewals
close as `shutdown` before the control lapses and release it; a refresh finding a newer
grant elsewhere closes as `superseded` with no refresh answer; client close releases.
Found and fixed during development: a refresh that found a newer grant elsewhere first
sent an "accepted" RefreshResult with expiry 0; a `postgres-check.sh` probe that used
version 10 as an unknown future version; a commit-id collision between test files.
Executed: Linux GCC 14 `tools/ci/postgres-check.sh` complete (all integration suites,
backup, host-loss and upgrade rehearsals, migration and audit probes); clang-tidy 19 on
the store, ledger and database tests (Clang build with network and PostgreSQL) and on the
session sources and tests; format check; GCC preset `quic` CTest 4/4, session tests 10/10
repeated; Clang 19 ASan/UBSan session and network tests; Windows MSVC 19.44 preset `quic`
build and CTest 4/4 (binary timestamp checked).
Not done: evo-server and client integration (10.4), ASan of the PostgreSQL suite,
a real two-process cross-region takeover with QUIC clients, cluster migration (no deploy).
## 2026-10-03 — M10.4a intentions, actions and admission (ADR-0092)
Implemented session-v1 types 6-12 and Welcome server_tick (codec), `IntentAdmission`
(evo_network_protocol), RegionSessions host facts, datagram intentions, actions, queries,
carry/committed and outcome delivery, and the client API (intentions, actions, queries,
receipts, watermark, outcomes).
Unit tests (`evo-network-tests`): round trips of every new message, every truncation and
extension refused, an intention with appended state claims refused, invalid statuses and
results refused; admission refusals in order (other session, unknown session, stale epoch,
vector, replayed sequence, tick in the past, beyond 8 ticks, same tick, rate after 8
attempts, not ready); cumulative watermark and its 20-tick repeat; action receipt,
pending query, zero id, unknown kind, wire kind 257 (no truncation), duplicate retry keeping
the original tick, action rate, carry once, commit to `committed_no_effect`, other
account's query, outcome visible to a reconnected session, never admitted twice, unresolved
bound across intentions and actions, abandon finalizing only uncarried actions, bounded
final outcomes.
Loopback QUIC (`evo-quic-session-tests`) with a simulated host: Welcome server tick;
intention receipt, outside window, watermark after the tick runs, not ready; action receipt,
pending query, carry, durable outcome, duplicate receipt then the same outcome, the outcome
from a reconnected session, an uncarried action finalized `not_committed` by an epoch
change, intentions in the new epoch; a raw peer appending state claims to an intention
datagram is closed with `protocol`.
Found and fixed during development: an 8-bit `ActionKind` decoded from the u16 wire field
would have truncated unknown kinds to known ones (now raw u16 validated by admission);
unchecked optional accesses in the client (tidy).
Executed: Linux GCC 14 preset `quic` CTest 4/4; session tests 10/10 repeated; Clang 19
ASan/UBSan session and network tests; clang-tidy 19 on the new and changed sources and
tests; format check; Windows MSVC 19.44 preset `quic` build and CTest 4/4 (binaries rebuilt,
timestamps checked).
Not done: durable action records and crash resolution (10.4b), evo-server wiring and
process kill tests (10.4c).
## 2026-10-03 — M10.4b durable action records (ADR-0092, schema v11)
Implemented migration v11 (`region_actions`, `region_commit_actions`, `check_action_block`,
`commit_actions`, three-argument publication, six-argument resolution,
`find_region_action`; apply chain and audit), `encode_actions`, the store overloads,
actions in `PreparedCommit`/publication/resolution, host `submit_action` with
`ActionAck`/`DroppedAction`, `ActionLedger` in admission and `PostgresActionLedger`.
Database tests (`tests/region_action_tests.cpp`, regions 7/131-135, commit ids tagged 0xA7):
publication with actions and lookups (including another account's and an absent id);
exact retry and resolution with the block; a changed block conflicting (retry and
resolution); nine invalid blocks refused (tick outside, tick 0, kind, zero account, zero id,
unsorted, duplicate, one id at two ticks, wrong length) and the encoder refusing the same;
a later batch carrying an already recorded id committing with the first record kept; a
batch without actions; lookups fenced with an old lease; the PostgreSQL action ledger.
Host: acknowledgement with the tick each action rode, late action riding the next tick,
duplicate, invalid and full submissions, not ready; crash before the transaction (successor
finds no record), crash after the commit before any acknowledgement (record durable,
successor at commit 1); unknown publication outcome resolved by exact bytes including the
actions. Unit tests: duplicate from the record, unavailable record refusing admission,
refused attempts costing no lookup, exact commit tick, final and unknown drops, unknown
decided once ready (committed or not committed), evicted outcome found on record.
`postgres-check.sh`: history 1..11, future-version probe 12, v11 checksum drift, audit
refusing SELECT on both new tables and EXECUTE on `check_action_block`.
Found and fixed during development: updating `region_commits` with the block was refused by
the immutable-history trigger (blocks now live in their own append-only table); a repeated
action id would have failed the later commit and quarantined a healthy region (now the first
record is kept and outcomes come from it); the encoder and SQL accepted one id at two ticks
in one block; test fixtures colliding with other tests' world, commit ids and accounts.
Executed: Linux GCC 14 `tools/ci/postgres-check.sh` complete (18 integration and rehearsal
suites, migration and audit probes); clang-tidy 19 on the changed persistence, host,
ledger, session and test sources (Clang builds with network+PostgreSQL and with quic);
format check; GCC preset `quic` CTest 4/4, session tests 10/10 repeated; Clang 19 ASan/UBSan
session and network tests; Windows MSVC 19.44 preset `quic` build and CTest 4/4 (binaries
rebuilt, timestamps checked).
Not done: evo-server wiring, headless client and process kill tests with real sessions
(10.4c); ASan of the PostgreSQL suite; a real two-region move with a pending action.
## 2026-10-04 — M10.4c evo-server QUIC listener and headless client (ADR-0092)
Implemented the opt-in `--m10-*` listener in evo-server (`apps/evo-server/m10_bridge.hpp`:
RegionSessions with PostgreSQL control and action ledgers on their own `PgSlot`, under
`RegionHost::authority()`), `evo-session-probe` (dev issuer keys, signed dev grants, login,
intentions, actions, queries) and `tests/region_m10_session_tests.cpp` (evo-postgres-tests
with EVO_BUILD_QUIC; region 7/151).
Process evidence (real evo-server, synthetic ticks, fast lease, in-process QUIC clients):
intention receipt and watermark; action receipt then a committed outcome matching its
durable record; SIGKILL after the record of a second action exists, restart, a newer grant:
the query answers the record, and a retry of the first action is a duplicate with the same
outcome; SIGKILL while a third action's tick had not run: no record, the next process
answers `unknown`, the retry is admitted and commits once; the probe logs in, gets
admitted intentions and two committed actions and exits 0; SIGTERM drains with exit 0.
Found and fixed with real processes: clients that sent only actions learned the server tick
too rarely (watermarks repeated every 20 ticks and only after intentions), so a later
action named a past tick (`outside_window`); the watermark now goes to every session every
simulated tick. The first intention after Welcome can be refused while the QUIC datagram
extension is not yet negotiated: tests retry (the probe retries each tick) and the client
API documents it. Also a shadowed variable in the probe (MSVC C4456, GCC -Wshadow) and
unchecked optional accesses (tidy).
Executed: Linux GCC 14 `tools/ci/postgres-check.sh` with EVO_BUILD_QUIC (20 integration and
rehearsal suites including the M10 process test, migration and audit probes); clang-tidy 19
on evo-server, the probe, the ledgers, the host and the M10/action/control tests (Clang
build with PostgreSQL, network and QUIC) and on the session sources and tests; format
check; GCC preset `quic` CTest 4/4, session tests 10/10 repeated; Clang 19 ASan/UBSan
session and network tests; Windows MSVC 19.44 preset `quic` build (including the probe) and
CTest 4/4 (binaries rebuilt, timestamps checked).
Not done: the engineering GUI client on M10 sessions; packaging (the server image does not
enable QUIC); a two-process cross-region move; ASan of the PostgreSQL suite.
## 2026-10-05 — Dev session route moved to UDP 7400 for QUIC (ADR-0087 amendment)
Finding: the deployed M10 QUIC listener (UDP) sat behind a route that admitted only TCP
7400 (container port, Service, NetworkPolicy, policy check). Pod-side checks (ready,
listener bound on 7400) could not detect it. Also found: `void.aeternalisgames.com`
resolves to 172.10.1.160 (control plane) from the workstation, not the VIP 172.10.1.202.
Changed: route container port, Service and NetworkPolicy to UDP 7400 only; policy check
(route exactly UDP 7400, UDP container port, any-protocol session exposure elsewhere) and
six new self-test cases; ADR-0087 amendment; session-v1 frozen note.
Executed: `tools/ci/deploy-policy-check.py` (71 objects in 69 files, 130 rendered objects
in 17 overlays, 0 failures) and `tools/ci/deploy-policy-selftest.py` (44 cases, all
refused) with pinned kustomize v5.8.1, before (38 cases) and after; rendered region/dev
inspected: Service `evo-region-session` UDP 7400 only on 172.10.1.202 with source range
172.10.1.0/24, `Local`, no node ports; NetworkPolicy UDP 7400 from 172.10.1.0/24; pod
port `session` UDP 7400; no other Service exposes 7400.
Not done (needs the maintainer): Argo sync of this change; the edge firewall rule;
internal DNS to 172.10.1.202; retrieving the dev CA and issuer key (no cluster credentials
on the workstation); the end-to-end probe through the VIP. The route is not verified.
## 2026-10-05 — Dev session route reachability over UDP (not yet acceptance)
After the maintainer synced the UDP route and set internal DNS, from workstation
172.10.1.169 (inside 172.10.1.0/24): `void.aeternalisgames.com` resolves to the VIP
172.10.1.202 (earlier answers of 172.10.1.160 were cached). `evo-session-probe` with a
throwaway CA and issuer key (generated locally; cannot authenticate) against
`172.10.1.202:7400` and `void.aeternalisgames.com:7400` ended with transport cause
`untrusted` after 62-67 ms: a QUIC server answered with its certificate, which the pinned
throwaway CA refused. Control: `172.10.1.202:7401` ended after the 5 s Welcome deadline with
no transport event (nothing answered). This shows UDP 7400 reaches the M10 listener through
DNS, the MetalLB VIP, the Service and the NetworkPolicy from the same subnet. It is not the
acceptance run: that needs the dev CA and issuer key from the cluster Secret (no cluster
credentials on the workstation yet), and a client on the same subnet does not cross the
edge firewall.
Probe changes: host names are resolved (IPv4) and printed (`resolved HOST ADDRESS`), and
`closed` lines name the QUIC transport cause (`RegionSessionClient::transport_cause`).
Executed: Windows MSVC preset `quic` build; Linux GCC preset `quic` CTest 4/4; clang-tidy 19
on the client, probe and evo-server sources.
## 2026-10-06 — M10.5a session protocol v2 and replication core (ADR-0093)
Implemented session protocol v2 (`Interest`, `SnapshotChunk`, `SnapshotAck`; version 1
refused), `ReplicationServer` and `SnapshotAssembler` (`include/evo/network/replication.hpp`)
and `tests/network_replication.cpp` (part of `evo-network-tests`).
Tests: v2 codec round trips, version 1 refused, chunk header rules (baseline >= tick, index
>= count, count 0 or > 16, zero epoch or tick), exactly one datagram of records accepted and
one byte more refused, every header truncation refused; radius clamped to 7; a fresh
225-cell baseline arriving nearest first within 16 KiB per tick over two ticks; nothing
outside the interest square ever received; converged ticks sending one empty chunk; a near
change sent at once and a far change deferred exactly 4 ticks; removal; a dropped chunk
leaving the snapshot unacknowledged and the next delta against the older baseline; the
acknowledged baseline kept 40 ticks and a fresh baseline after that; acks for evicted or
another epoch's snapshots ignored; a new interest restarting from a fresh baseline with only
the new square; concealment (an organism one session may not perceive and organisms outside
interest absent from the encoded bytes); client rejection of truncated, unknown-kind and
other-slice records and disagreeing chunk headers, stale and foreign-epoch chunks ignored,
deltas against unknown baselines dropped, at most 4 assemblies; 32 dense sessions for 40
ticks with a changing row (report: max 16,320 bytes per session per tick, mean 1,404,
28,800 retained entity states).
Found and fixed during development: eviction of the acknowledged baseline when acks lag
four or more ticks (normal at 100 ms RTT) would have forced fresh baselines (it is now always
kept); a dangling reference to an erased assembly entry; enum values cast before validation
in the record reader; MSVC narrowing in tuple keys and an ambiguous map erase.
Executed: Linux GCC 14 preset `quic` CTest 4/4; session tests 10/10 repeated; Clang 19
ASan/UBSan session and network tests; clang-tidy 19 on the new and changed sources and
tests; format check (only unrelated uncommitted files of the maintainer fail it); Windows
MSVC 19.44 preset `quic` build and CTest 4/4 (binaries rebuilt, timestamps checked).
Not done: wiring into RegionSessions, evo-server and the clients (10.5b); the GUI client
(10.5c).
## 2026-10-06 — M10.5b snapshot wiring (ADR-0093)
Implemented replication in `RegionSessions` (open/close with the session, `Interest`,
`SnapshotAck`, default interest, `replicate` with per-connection datagram limits),
snapshot assembly and acknowledgement in `RegionSessionClient`, the evo-server source (the
region's habitat sites in the interest square, `query_environment` and
`present_environment`), and probe options `--interest` and `--snapshots`.
Loopback QUIC (`evo-quic-session-tests`): a 21 x 21 synthetic slice; the default interest
(clamped to radius 7) arriving as 225 cells, none outside the square; a change reaching the
client; acknowledgements keeping deltas (fresh baselines fewer than snapshots); a new
interest replacing the view with the 25-cell square; the session staying open.
Real processes (`tools/ci/postgres-check.sh`, evo-server with QUIC): snapshots of the test
region's default interest (one habitat site), every cell in one slice and one 15 x 15
square, a snapshot every tick (40 chunks, 8,888 bytes in 40 ticks); the probe printing
snapshots and `snapshot-bytes` with three snapshots.
Found and fixed during development: test cells requesting water without moisture (the
client correctly refused the snapshot and closed); a local name hiding the fixture's
`env()`; clang-tidy optional and return-value findings.
Executed: Linux GCC 14 `postgres-check.sh` with EVO_BUILD_QUIC (20 suites); GCC preset
`quic` CTest 4/4; session tests 10/10 repeated; Clang 19 ASan/UBSan session and network
tests; clang-tidy 19 on sessions, client, replication, bridge, probe and tests; Windows MSVC
19.44 preset `quic` build (including the probe) and CTest 4/4 (binaries rebuilt, timestamps
checked). The repository format check fails only on `include/evo/server/session_window.hpp`
(already on main, not part of this change).
Not done: the GUI client (10.5c); deployment of v2 (the dev server still runs v1).
## 2026-10-06 — M10.5c engineering client on M10 sessions (ADR-0093, ADR-0091, ADR-0086)
Implemented the client's remote mode over `RegionSessionClient`
(`apps/evo-client/remote_session.hpp`):
- snapshot-fed views;
- the camera follows the default view, then drives the interest;
- intention record states, including `simulated`;
- reconnect with the bundle's next grant, with a growing pause.
Supporting changes:
- the launcher reads 160-byte grants, takes `authority_path`, and defaults to the dev route;
- `quic::resolve_ipv4`;
- `evo-session-probe grants`;
- evo-server `--m10-max-grant-lifetime` (dev only);
- presets `client-network` and `client-package` with QUIC;
- msquic with a fully static CRT;
- `msquic.dll` and the msquic and OpenSSL licences in the Windows package;
- `CL=/Brepro` and `SOURCE_DATE_EPOCH` for the package's OpenSSL build.
New test `client-remote-session` (labels client and quic-session) runs against a loopback
region:
- launcher flow and the 225-cell default view;
- following, then the camera interest (144 cells clipped);
- admitted then simulated intentions;
- an epoch change: unknown outcomes, stale views, reconnect with the next grant and the
views restored;
- client-side numbering across sessions;
- superseded, wrong region, an exhausted bundle, and an unresolvable endpoint.
The process test now runs evo-server with `--m10-max-grant-lifetime 900` and the probe with
a 900 s grant.
Found during development:
- test cells requesting water without moisture (the client refused the snapshot and closed);
- a budget-cut first snapshot skewed a one-time recentre, so the camera now follows until
the user moves it;
- session sequences collided in the outcome record, so the record now numbers entries
itself;
- msquic's default partial-static CRT imported the UCRT forwarders;
- a clang-tidy optional-access finding.
Executed:
- **Linux GCC 14**
- preset `client-network` (with QUIC): CTest 14/14; `client-remote-session` passed 5 times
in a row;
- preset `quic`: 4/4;
- `postgres-check.sh` with EVO_BUILD_QUIC: 20 suites, including the region m10 session
integration.
- **Linux Clang 19:** clang-tidy 19 on the client app, remote test, host lookup, probe,
launcher and UI-model test; clang-format 19 on the changed files.
- **Windows MSVC 19.44** (binaries rebuilt, timestamps checked):
- preset `client-network`: 12/12;
- preset `quic`: 4/4.
- **Windows package:**
- `package-windows.ps1`: client-package CTest 12/12;
- `verify-windows-package.ps1` passed;
- `package-selftest.ps1`: 10 refusals, including the missing `msquic.dll` and the
dynamic-runtime image;
- `msquic.dll` imports only advapi32, bcrypt, crypt32, iphlpapi, kernel32, ncrypt, ntdll,
user32, winmm and ws2_32;
- two `-Clean` builds gave identical ZIPs (sha256 `aaabc228…312cb2`, identity
`g33e91458e7ee-dirty`), the same as the earlier incremental build.
- `evo-session-probe grants`, checked by hand:
- 8 grants, 1280 bytes, mode 0600;
- refuses an existing file, a lifetime under 60 s, a count over 64 and connect flags;
- `--serial` respected.
Not done:
- a GUI run against evolution-dev (the dev server still runs session v1 and the flag is
not in the dev manifest);
- ASan of the client remote test (msquic under sanitizers is still open);
- a clean-VM package run.
## 2026-10-06 — M10.6a probe-motion experiment core (ADR-0094)
Implemented (EXPERIMENT, not a game rule):
- probe-motion-v1, shared and integer-exact (`include/evo/network/probe_motion.hpp`);
- the server's `ProbeMarkers` (spawn, area, bounded schedule, catch-up bound);
- session protocol v3 marker records, kinds 5 and 6:
- the own marker is always sent, with its applied sequence and area;
- other markers pass the spatial filter;
- v1 and v2 are refused;
- the client `MotionPredictor` (bounded inputs, rebase and replay, corrections, error
samples), `CorrectionSmoother` and `InputLead` (`include/evo/client/motion_prediction.hpp`).
Tests (`tests/network_motion.cpp`):
- the rule's arithmetic and clamping;
- the marker rules;
- client/server loops with whole-tick delays:
- with enough lead, every prediction is exact (zero corrections);
- with too little lead, refusals until the lead adapts, then exact;
- lost inputs give corrections;
- predictor bounds, refusal, drop and reset;
- smoothing and snap, and lead adaptation;
- marker records through the real codec: own versus other, the own marker outside the
interest, unchanged markers not resent, removal, and four malformed records.
The tests that used "3" as a future protocol version now use 4.
Executed:
- **Linux GCC 14**
- `evo-network-tests` and `evo-quic-session-tests`;
- preset `quic`: CTest 4/4;
- `client-network`: 14/14;
- `postgres-check.sh` with QUIC: 20 suites, including region m10 session.
- **Linux Clang 19:** build, ASan/UBSan `evo-network-tests`, clang-tidy 19 on the new and
changed sources and tests.
- **Windows MSVC 19.44** (fresh binaries):
- preset `quic`: 4/4;
- preset `client-network`: 12/12.
The repository format check fails only on `include/evo/server/session_window.hpp`, as
before.
Not done:
- wiring into sessions, evo-server, the probe and the GUI (10.6b);
- impaired traces (10.6c).
Deployed v2 servers refuse v3 clients until redeployed.
## 2026-10-06 — M10.6b probe-marker wiring (ADR-0094)
Implemented:
- `RegionSessions` markers (`set_motion_area`, schedule on admission, advance per poll,
close and loss, clear on a new epoch, markers in every snapshot source);
- the evo-server motion area (the habitat sites' bounding cells, `m10 motion area`);
- the probe's `marker` lines;
- the engineering client's prediction, adaptive lead, resend after loss or refusal,
smoothing, marker drawing, the follow camera and interest, and Session panel rows.
Evidence:
- `client-remote-session`: the marker at its spawn, followed by camera and interest; an
intention moves the predicted marker +x, and the authority confirms it with zero maximum
error and no pending inputs. Passed 3 times in a row.
- The real-process test (`region_m10_session_tests`): the probe's own marker moved +x with
its intention applied, and evo-server printed its motion area. The snapshot report is
now 8,966 bytes in 40 ticks, including the marker records.
Executed:
- **Linux GCC 14:**
- preset `client-network`: CTest 14/14;
- preset `quic`: 4/4;
- `postgres-check.sh` with QUIC: 20 suites.
- **Linux Clang 19:** build, plus clang-tidy 19 on the client app, the remote test,
RegionSessions, the probe and probe_motion.
- **Windows MSVC 19.44** (fresh binaries; built while the formatting pass ran, which
changed whitespace only):
- preset `quic`: 4/4;
- preset `client-network`: 12/12.
clang-format 19 was applied to the changed files.
Not done:
- impaired traces (10.6c);
- a GUI run against evolution-dev, which needs a v3 server.
## 2026-10-06 — M10.6c prediction traces (ADR-0094)
Implemented:
- `evo-motion-trace` (the client's own session code through the impairment relay against
a real-time 50 ms loopback region);
- the CTest `motion-trace-smoke`;
- the relay takeover by a new sender after the current client has been silent for 1 s;
- a refusal count in the client status, replacing the unused predictor field.
Figures are in docs/networking/M10_PREDICTION_TRACES.md. Linux, one run of each profile:
- clean and lan: 100 % exact;
- rtt60-loss2: 97.9 %;
- rtt120-loss2: 99.3 %; restart resync 2,946 ms;
- rtt200-loss5: 94.1 %; p99 500 mm, max 989.9 mm.
Every profile converged after stopping. Windows was similar, with a maximum of 1,399.8 mm
at rtt200-loss5. The smoke profile passed 3 times in a row on Linux.
Found and fixed during development:
- the relay ignored a reconnecting client (new source port), so restart resync timed out;
- refusals were undercounted when an input had already been dropped;
- a clang-tidy widening-cast finding.
Executed:
- **Linux GCC 14**
- preset `client-network`: CTest 15/15, including motion-trace-smoke and
client-remote-session;
- preset `quic`: 4/4, including quic-transport with the changed relay.
- **Linux Clang 19:** clang-tidy 19 on the trace tool, client app, remote test, predictor
and relay.
- **Windows MSVC 19.44** (fresh binaries):
- preset `quic`: 4/4;
- preset `client-network`: 13/13, including motion-trace-smoke;
- a full trace run.
The Postgres suite was not rerun: no server code changed since 10.6b.
Not done:
- traces over the dev route (a v3 server must be deployed first);
- many-client load (10.8).
## 2026-10-06 — M10.7a tick clock and remote interpolation (ADR-0095)
Implemented:
- `TickClock` (slew 10 %, forward snap beyond 4 ticks, monotonic, at most 1 tick beyond the
newest observation);
- `MarkerInterpolator`:
- interpolation, then at most 2 ticks of extrapolation, then stale;
- appearance and removal at their snapshot ticks;
- reordering ignored; no interpolation across gaps over 8 ticks;
- adaptive delay of 2-6 ticks from the 90th-percentile gap;
- `MotionPredictor::predict_between`.
Tests (`tests/network_interpolation.cpp`):
- clock basics: freeze while silent, no backward step, forward snap, reset;
- drift with jitter at 0 % and ±4 % (4,000 ticks): within 1.5 ticks, monotonic, no snaps;
- interpolation, removal, appearance, extrapolation and staleness, reordering, long gaps,
reset;
- adaptive delay up and down;
- fractional prediction.
Executed:
- **Linux GCC 14:**
- preset `quic`: 4/4;
- `client-network`: 15/15.
- **Linux Clang 19:** ASan/UBSan `evo-network-tests`, and clang-tidy 19 on the new sources
and tests (after fixing one optional-access finding).
- **Windows MSVC 19.44:**
- preset `quic`: 4/4 (rebuilt after the fix);
- `client-network`: 13/13.
Not done: client wiring, the loss view and the two-client trace (10.7b).
## 2026-10-06 — M10.7b client interpolation, loss view and traces (ADR-0095)
Implemented:
- the engineering client's tick clock, interpolated remote markers (grey when stale), the
own marker between ticks, and the frozen stale view on loss (banner, all markers stale,
own marker at its last authoritative position), reset on Welcome;
- `TickClock::silence` and silence-based staleness; long-gap holds marked stale;
- error recording limited to scheduled ticks (`predict_between(..., record)`);
- `evo-motion-trace --remote` and the CTest `motion-trace-remote-smoke`;
- a shared region bench in the trace tool.
Found and fixed during development:
- an outage never showed stale (the frozen clock kept the render tick inside the buffer);
- post-outage holds were shown as current;
- false prediction errors from presenting one tick past the scheduled ticks (a Windows-only
`client-remote-session` failure);
- MSVC rejecting `std::erase_if` with a member-pointer predicate;
- two clang-tidy optional-access findings.
Figures are in docs/networking/M10_PREDICTION_TRACES.md:
- remote markers shown as current: p99 at most 60.7 mm, max 100 mm;
- outages: 84-90 % of samples shown stale, recovery 150-950 ms;
- smoke variance: one of three runs had 32 % extrapolated and 36 % stale samples, with
current errors still 0.
Executed:
- **Linux GCC 14:**
- preset `client-network`: CTest 16/16 (including both trace smokes and
client-remote-session);
- preset `quic`: 4/4;
- full remote and prediction trace runs.
- **Linux Clang 19:** ASan/UBSan `evo-network-tests`, and clang-tidy 19 on the trace tool,
client app, panels, remote test, interpolation sources and tests.
- **Windows MSVC 19.44** (fresh binaries):
- preset `client-network`: 14/14;
- `client-remote-session` passed 5 times in a row after the fix;
- preset `quic`: 4/4.
The Postgres suite was not rerun: no server code changed in M10.7.
Not done:
- interpolating organisms (none move);
- a GUI run against evolution-dev (needs a v3 server).
## 2026-10-06 — M10.8a probe handoff schema v12 (ADR-0096)
Implemented:
- `migrations/0012_probe_handoff.sql`:
- `probe_records` and `probe_transfers`;
- `open_probe`, `save_probe`, `prepare`, `stage`, `commit`, `abort`, `find`;
- `handoff_account_control`;
- SQLSTATEs EVL13-15;
- the apply chain, migration runner, privilege audit allowlist, CI version checks and
tamper tests;
- the C++ store (`include/evo/persistence/probe_store.hpp`, and `handoff_account_control`
in `account_control_store.hpp`);
- `schemas/probe-handoff-v1.md` and the POSTGRESQL.md section.
`tests/probe_handoff_store_tests.cpp` (world 9200, real PostgreSQL) covers:
- ownership and fencing; argument refusals;
- the happy path with retries of every call before and after commit;
- one open transfer per probe; transfer ids bound to one account;
- crossing back;
- source restart after prepare;
- destination restart after staging (re-stage required; the old epoch fenced) and after
commit (the retry answers);
- a failed destination (source abort); a destination abort;
- a partition (the source lease lapses while the destination commits; the recovered source
is fenced);
- 24 commit/abort races on separate connections (7 commits, 17 aborts; both sides always
observed the same outcome);
- the account-control handoff (refused before commit, for another region or account; a wait
while the source holds control; none after release).
After every scenario: one row per probe, the test inventory conserved (n x 1000), and frozen
exactly when an open transfer names the owner as source.
Executed:
- **Linux GCC 14:** `postgres-check.sh` with QUIC: 21 suites, including probe handoff,
migrations (v1-v11 upgrade paths to v12), tamper and audit checks;
- preset `quic`: 4/4.
- **Linux Clang 19:** clang-tidy 19 on `region_lease_store.cpp` and the new test (two
findings fixed); clang-format 19.
The suite was rerun on the final code after the tidy fixes: 21 suites passed again; that
run's races ended 11 commits, 13 aborts.
Not done:
- tickets (10.8b), region wiring (10.8c) and two-process kills (10.8d);
- deploying v12 (needs a rebuilt migrations image; a maintainer step).
## 2026-10-06 — M10.8b handoff tickets and admission (ADR-0096)
Implemented:
- `HandoffSigningKey`, `HandoffVerifier`, `VerifiedHandoffTicket` and `HandoffLedger`
(`include/evo/network/handoff_ticket.hpp`; schema `network/handoff-ticket-v1`);
- `SessionControl::check_handoff` and `open_handoff`;
- `RegionSessions` ticket admission (marker routing, neighbour keys, durable step first,
shared session-opening tail);
- `PostgresHandoffLedger` (stage, commit, `handoff_account_control`).
Tests:
- `network_control`:
- tickets round-trip; a flipped byte every 7 bytes is refused;
- wrong source key, scope (world and destination), expiry, lifetime over 120 s, a session
already ended, wrong environment, unknown key and invalid claims are all refused;
- grants and tickets never verify as each other;
- control: a ticket supersedes a grant session, replay is refused, the session expiry is
capped, and later grants still open.
- `quic_sessions` (loopback QUIC):
- no keys: refused;
- a handoff ledger without a control ledger: configuration error;
- an admitted ticket: the durable step for the session Welcome names, the ticket's session
expiry, renewal under the handoff's generation;
- replay, another destination, and a refused or undecided durable step: refused
(unauthenticated, wrong_scope, capacity).
- `probe_handoff_store_tests` (PostgreSQL): the ledger admits and moves ownership and
control; a retry converges; another account, an unknown or aborted transfer, and no lease
are refused.
Executed:
- **Linux GCC 14:** `postgres-check.sh` with QUIC: 21 suites (race run: 16 commits, 8
aborts); preset `quic`: 4/4; `evo-network-tests` and `evo-quic-session-tests`.
- **Linux Clang 19:** ASan/UBSan `evo-network-tests`; clang-tidy 19 on the ticket,
session-control, region-sessions, ledger and test sources; clang-format 19.
- **Windows MSVC 19.44:** preset `quic`: 4/4; `client-network`: 14/14 (built before the
formatting pass, which changed whitespace only).
Not done:
- the source side and the client switch (10.8c);
- two-process kill tests (10.8d).
## 2026-10-06 — M10.8c-1 region handoff wiring (ADR-0096)
Implemented:
- session protocol v4 (HandoffOffer, SessionRedirect, `handed_off` and `moved`, ghost marker
kind; v1-v3 refused);
- the client library keeps offers and redirects;
- `ProbeLedger` (durable open, save, prepare, abort, incoming) and the PostgreSQL version;
- `HandoffAdmission` (state after the switch);
- `RegionSessions`:
- neighbours and the movement area;
- crossing detection, freeze, prepare, control release, a signed ticket, the offer and the
close;
- timed aborts and ghost polling;
- logins placing the durable probe (clamped into the area), with a redirect or `moved`
when it is owned elsewhere;
- saves at session end;
- `ProbeMarkers` `open_at`, `place`, `freeze` and ghosts;
- schema v13 `incoming_probe_transfers` (apply chain, audit, CI checks);
- client ghost drawing.
Found and fixed during development:
- a dangling marker pointer after clamping (two `place` calls);
- the spawn computed from the area instead of the configured one;
- a probe left beyond the edge by an aborted handoff crossing again at login (now clamped).
Tests:
- codec tests for the offer and redirect (round trip, text and padding rules, zero region or
transfer, lengths), with the v4 version, type and reason boundaries;
- ghosts and freezing through replication;
- `two_region_handoff` (two regions in one process over a fake world database):
- a crossing; a ghost at the destination before the switch;
- a ticket login at the crossing position (owner and generation moved); the ghost gone;
- a redirect; a crossing back;
- an abandoned handoff aborted after its deadline, the probe resumed at the edge, and the
late ticket refused;
- PostgreSQL: incoming transfers appear for the destination only while open.
Executed:
- **Linux GCC 14:**
- `postgres-check.sh` with QUIC: 21 suites (migrations to v13; races 8 commits, 16
aborts);
- presets `quic` 4/4 and `client-network` 16/16;
- `evo-quic-session-tests` 3 times in a row.
- **Linux Clang 19:** ASan/UBSan `evo-network-tests`; clang-tidy 19 on the protocol, motion,
replication, interpolation, client, region-sessions, handoff-ledger, store and test sources
(findings fixed); clang-format 19.
- **Windows MSVC 19.44:** presets `quic` 4/4 and `client-network` 14/14.
Not done:
- the engineering client's handling of offers, redirects and multi-region grants, and
evo-server and probe configuration (10.8c-2);
- real processes (10.8d).
Deployed v3 servers refuse v4 clients until redeployed; v12 and v13 need a rebuilt
migrations image.
## 2026-10-06 — M10.8c-2 client switch and handoff configuration (ADR-0096)
Implemented:
- the engineering client follows handoff offers (ticket, no grant, no pause, view kept) and
redirects;
- grants are picked by region claim in rising serial order, and outdated grants are
discarded;
- views reset on a new region;
- evo-server `--m10-area`, `--m10-handoff-key` and `--m10-neighbour` (dev only), wiring the
PostgreSQL probe and handoff ledgers;
- the probe tool's multi-region `grants` and its `handoff`/`redirect` lines;
- the shared fixture `tests/fixtures/fake_handoff_world.hpp`.
Tests:
- `client-remote-session` `crossing_regions`:
- the GUI session walks into region 2 and continues there with the ticket (no grant used,
connected, marker past the border, probe owned by region 2);
- a second client's login at region 1 is redirected to region 2 and takes over.
It passed 3 times in a row on Linux, plus once more after a tidy fix.
Executed:
- **Linux GCC 14:**
- `postgres-check.sh` with QUIC: 21 suites;
- presets `quic` 4/4 and `client-network` 16/16.
- **Linux Clang 19:** clang-tidy 19 on the client app, panels, remote test, probe, QUIC
session test and evo-server `main.cpp` with the bridge (one finding fixed); clang-format
19.
- **Windows MSVC 19.44:** presets `quic` 4/4 and `client-network` 14/14.
Not done:
- real two-process handoffs on PostgreSQL with kills (10.8d);
- a GUI run across regions on evolution-dev (one dev region only; needs v4 and v13
deployed).
## 2026-10-06 — M10.8d two-process handoff fault matrix (ADR-0096)
Implemented:
- `tests/region_handoff_process_tests.cpp`: two evo-server processes (regions 7/161 and
7/162, adjacent `--m10-area`s, handoff keys and each other as `--m10-neighbour`, fixed
loopback ports) with in-process session clients that follow offers and tickets. It covers:
- a crossing and a crossing back, plus a redirected stale login;
- destination SIGKILL after stage, then a re-stage and commit under the new epoch;
- destination SIGKILL after commit, then a converging ticket retry;
- a refused ticket replay after the probe moved on;
- a refused same-process duplicate;
- source SIGKILL after prepare: one client still arrives, the other never does and is
aborted by the restarted source's sweep;
- a destination that is down, with the live source aborting and the late ticket expired.
After every scenario, SQL checks one owner per probe, the conserved inventory, and
frozen-iff-open-transfer.
- Schema v14 `overdue_probe_transfers` (apply chain, audit allowlist, upgrade and drift
checks), `ProbeLedger::overdue`, and the session layer's recovery sweep
(`overdue_poll`, ticket lifetime plus 10 s).
- The replay guard in `PostgresHandoffLedger::admit`: a committed transfer admits only
while the record is still owned here at generation plus 1 and not frozen.
- Library tests:
- the v14 store call (age bounds, source only, aborted transfers excluded);
- the refused retry in the ledger test;
- the sweep over the fake world in `two_region_handoff`.
Findings: the two recovery gaps above were found by this matrix, before any deployment of
v4/v12/v13.
Executed:
- **Linux GCC 14:** `postgres-check.sh` with QUIC, all suites including the new one (restart
waits 0.03-4 s; the replay ran at 8 s of the 30 s ticket lifetime). It passed 3 times in a
row, plus a final run after a tidy-only change. Presets `quic` 4/4 and `client-network`
16/16.
- **Linux Clang 19:** clang-tidy 19 on `region_sessions.cpp`, `postgres_handoff_ledger.cpp`,
`region_lease_store.cpp`, the QUIC session, remote client, probe store, process,
migration and Postgres runner tests (one finding fixed); clang-format 19.
- **Windows MSVC 19.44:** presets `quic` 4/4 and `client-network` 14/14. The PostgreSQL
process tests are Linux-only.
Not done:
- deployment of protocol v4 and schemas v12-v14 (needs a rebuilt migrations image; not done
from routine development);
- a GUI run across two regions on evolution-dev (only one dev region exists).
## 2026-10-07 — M10.9a load measurement and governor (ADR-0097)
Implemented:
- evo-server measures each simulated tick's work: the host step plus the M10 bridge
(session poll, replication, action carry). Waits are excluded, and the session layer's
share is kept separately.
- `LoadGovernor` (`include/evo/server/load_governor.hpp`, portable): a window of 40 ticks,
refuse at p95 ≥ 15 ms or any tick debt, shed one level per second at p95 ≥ 20 ms or
debt ≥ 2, and relax one step per 5 s of calm (p95 < 10 ms, no debt).
- `RegionSessions::set_load`: while refusing, grant logins are closed with `capacity`
before signature work; handoff tickets are still admitted.
- `ReplicationServer::set_shed`: levels 0-3 cap the radius (7/7/5/3) and the near radius
(3/3/3/2) and stretch far updates (x1/x2/x2/x4). Stored interests are kept, and own
markers are unaffected.
- `/metrics`: `evo_m10_*` gauges (sessions, refusing, shed level, work and session-work
percentiles, maximum) and a refused-logins counter, rendered only with the listener.
evo-server prints `m10 load ...` on every change.
- The ADR index gains the missing entries for ADR-0094 to ADR-0096.
Tests:
- `ops-contracts`: the governor state machine (thresholds, step interval, hysteresis, calm
restart, debt, policy validation) and the M10 metric families, including worst-case size.
- `network`: replication shedding (caps, no fresh baseline, 16-tick far cadence at level
3, recovery at level 0).
- `quic-sessions`: grant logins refused under load (the grant is not consumed), a ticket
login admitted, and logins accepted after relaxing.
- Postgres `region_m10_session_tests`: a real evo-server's `/metrics` shows one session,
measured work, not refusing, shed 0.
Executed:
- **Linux GCC 14:** `postgres-check.sh` with QUIC, all suites; presets `quic` (ops,
network, QUIC sessions) 3/3. The local reference (Release, one session) measured work
p95 7.1 ms and p99 8.8 ms, with the session layer at p99 4.5 ms. This is a development
reference only, flagged for 10.9b.
- **Linux Clang 19:** clang-tidy 19 on the governor, ops, replication, sessions, evo-server
`main.cpp` with the bridge, and the tests (one finding fixed); clang-format 19.
- **Windows MSVC 19.44:** presets `debug` 60/60 (ops-contracts included), `quic` 4/4 and
`client-network` 14/14.
Not done:
- qualification on the evolution-dev reference worker (10.9d; needs v4 and v12-v14
deployed);
- load generation (10.9b) and the abuse suite (10.9c).
## 2026-10-07 — M10.9b evo-load (ADR-0097)
Implemented:
- `evo-load` (`tools/network/load.cpp`, quic preset): bots on worker threads (intentions
every tick at the input lead with prediction, actions every 2 s, interest moves every
10 s, re-login after any close), stepped ramps.
- Per step, it reports pooled and per-client distributions (response, receipt, outcome,
login, prediction error), snapshot rate, gaps and bandwidth, drops, failed logins and
refusals, the generator's CPU, a starvation check and a budget verdict. With
`--metrics`, it adds the region's gauges. The run ends with `sustainable` and
`saturated_at`.
- A Postgres process smoke (`region_m10_session_tests` `load_smoke`): 4 then 8 bots against
a real evo-server.
- Fixes found while ramping:
- The process fixture's `Child::pump` rounded short waits down to 0 ms and never read,
so a chatty child blocked on its stdout pipe.
- The load governor refused on a single tick of debt. It now refuses at 2 and sheds at
4; one tick only interrupts calm (ADR-0097 amended).
- The M10 end-to-end test waits until the server is not refusing before the headless
probe logs in, and dumps the probe and server lines on failure. One run had failed
there intermittently before the debt change.
Executed:
- **Linux GCC 14:**
- `postgres-check.sh` with QUIC, all suites, 3 runs in a row (2 after the final change);
- the smoke passes with every bot logged in once and both steps within budget;
- presets `quic` (ops, network, QUIC sessions) 3/3.
- **Local development reference ramp** (temporary steps 8-64, not committed, not
evidence): within budget up to 24 bots; the governor refused from 32 (work p95 about
14 ms, maximum 17 ms); session work stayed near 0.25 ms per session; responses p95 about
103 ms, outcomes about 665 ms, 20 Hz snapshots, no starvation. Beyond the 32-session cap,
extra logins failed at the transport limit and established sessions were unaffected.
- **Linux Clang 19:** clang-tidy 19 on `load.cpp`, the governor, ops and the M10 process
tests (clean); clang-format 19.
- **Windows MSVC 19.44:** presets `quic` 4/4 (`evo-load` builds) and `debug` 60/60.
Not done:
- the reference-worker run (10.9d);
- the abuse suite (10.9c).
## 2026-10-07 — CI correction for M10.9a/b (ADR-0097)
With agent access to Gitea Actions it became visible that `checks (postgres)` had failed
on both M10.9 pushes (`b8cb9db`, `1ba6748`). The `server-image` and `image` jobs were
therefore skipped, and the last published server image is still from `97a3756`. The local
runs recorded above passed, but CI is the authority and was red.
Cause: `region_m10_session_tests` asserted that a lightly loaded region is not refusing.
On the shared CI runner (Release build) one session measured work p95 22 ms and p99 37 ms
(session layer p99 22 ms), so the load governor correctly refused, and the test's
assumption about host speed failed.
Fix: evo-server gains `--m10-load-governor on|off` (dev only; default on). With `off`,
work is still measured and exposed, but nothing is refused or shed, and the gauges report
refusing 0 and shed 0. The M10 session, load-smoke and two-process handoff tests use it;
the governor itself remains covered by its unit tests.
Executed: local `postgres-check.sh` with QUIC, all suites. The CI result for this commit
is checked via the Gitea API after the push.
## 2026-10-07 — X11 lifecycle check scope
The client CTest lifecycle suite runs 32 full open/event/close cycles with SDL's dummy
video driver. The additional Xvfb invocation runs one full cycle under X11, alongside
the X11 rendering and UI-probe runs. This retains real X11 creation, software rendering,
resize, input/window events, renderer reset/loss and teardown coverage while avoiding a
second 32-cycle stress loop on the Xvfb backend. If `Application::open()` fails in this
test, its output includes the cycle and typed application error before SDL teardown.
Evidence: in the 2026-10-07 Gitea client run, all 13 CTest tests and the X11 probe/UI
smokes passed; only the extra 32-cycle X11 invocation failed at `Application::open()`.
The same 32-cycle lifecycle test passed under SDL's dummy driver. The X11 failure did
not report its typed error, so the failing X11 backend stage was not established.
## 2026-10-07 — M10.9c-1 fuzzing and in-process abuse (ADR-0097)
Implemented:
- `tests/fuzz/session_fuzz.cpp` as a libFuzzer target (`evo-session-fuzz`, with the code
under test compiled in) and the seeded `session-fuzz-smoke` CTest (label `network`).
- Four `quic-sessions` abuse tests, each measuring an honest session tick by tick:
Hello flood, datagram flood, slow peer, and a connect/disconnect storm.
- `evo-session-probe run --grants FILE [--grant N]`, covered by the M10 process test.
- `tests/client_lifecycle.cpp` uses `SDL_getenv`: MSVC rejected `std::getenv`
(deprecation warning as error) from `1204750`, which CI cannot see (no Windows runner).
Results:
- **libFuzzer** (Clang 19, ASan and UBSan): 35.6 million inputs, then 23.3 million after
the snapshot-application assertion was added, 5 minutes each. Coverage 1541, then 1650;
no crash or sanitizer report.
- **Smoke:** 68,000 inputs per run.
- **Hello flood:** 160 peers. Every connected one was refused with the right reason
(52 `protocol`, 51 `unauthenticated`, 50 `timeout`), connections peaked at 29 of 32,
and the honest session kept 60 of 60 snapshots and receipts.
- **Datagram flood:** 1920 intentions accepted by the sender, of which 1914 were refused
by the throttle. Garbage closed the peer with `protocol`, and the honest session kept
40 of 40.
- **Slow peer:** retained entities peaked at 392 against a bound of 490, and the honest
session kept 80 of 80.
- **Storm:** 120 logins (69 seen established). The region returned to one session and one
connection, and the honest session kept 60 of 60.
- **evolution-dev, with the renewal's bundle:** grant 0 was welcomed (epoch 51), and its
first intention and an action were admitted. Nothing is simulated there outside a drill
(`REGION_INPUTS` is empty by policy), so there were no outcomes or snapshots.
Executed:
- **Linux GCC 14:**
- presets `quic` and `network` 5/5;
- `postgres-check.sh` with QUIC, all 18 suites, with the probe's bundle login.
- **Linux Clang 19:** clang-tidy 19 on the fuzz target and smoke, `quic_sessions.cpp`, the
probe and the M10 process tests (8 findings fixed); clang-format 19.
- **Windows MSVC 19.44:** presets `quic` 5/5 and `client-network` 15/15.
Not done (10.9c-2): lease loss with 20 or more bots, and a live raw-UDP sender of mutated
datagrams that loads the server's per-connection datagram cap.
## 2026-10-08 — M10.9c-2 lease loss under load (ADR-0097)
Implemented:
- `region_m10_session_tests` `lease_loss_under_load`: 24 `evo-load` bots against a real
evo-server on a fixed loopback port. It runs a SIGKILL and then a SIGTERM drain, each
followed by a successor under a new epoch.
- The test fixture's `Setup` gains a `listen` setting (default `127.0.0.1:0`).
- Scope: no standalone raw-UDP sender was built. The per-connection datagram cap's eviction
cost is analysed in ADR-0097, and making eviction O(1) is a follow-up.
Results (development reference, WSL, Release):
- **SIGKILL:** the successor was ready after 3.5 s, and all 24 bots were back 5.0 s
later.
- **Drain:** the successor was ready after 0.18 s, and all 24 bots were back 2.4 s later.
- **Bots over the 100 s hold:** 72 logins, 48 drops, no failed or refused logins; response
p95 106 ms; login p99 62 ms.
Executed:
- **Linux GCC 14:** `postgres-check.sh` with QUIC, all suites.
- **Linux Clang 19:** clang-tidy 19 on the M10 process tests (clean); clang-format 19.
- **Windows:** not applicable (the PostgreSQL process tests are Linux only).
## 2026-10-08 — evolution-dev crashloop and recovery
Incident:
- `b2f224d` set `REGION_SESSION_WINDOW` to end about 5.2 h after the pod started. evo-server
refuses windows longer than 4 h (`SessionWindow::max_duration_seconds`), so
`evo-region` exited with `error m10-window-too-long` in a loop.
- `4a72897` set a window of 3 h 50 min. The region then reached
`phase quarantined epoch=0 … exit 4`: the bot promotion `c3b40f4` (image `56647b49…`)
had never had its build retained. The provision hook still ran the previous image
(`2ccf6074…`, created 01:58Z).
- The two syncs requested afterwards with the agent token were both recorded by Argo
as limited to `Deployment/evo-region`, even though the second request carried no
resource filter. So no hooks ran. The cause is not visible with the agent's access
(possibly a guard on that account).
Recovery, within the agent contract (no reset, no SQL):
- Rendered `Job/aevolution-region-provision` from `deploy/kubernetes/dev` at
`4a72897` (image and `REGION_BUILD` `56647b49…`), deleted the stale Job and created it.
It completed with `already-provisioned … sequence=578 tick=12084`, the build retained
first.
- `rollout restart deployment/evo-region`. The region came up `phase running epoch=54
committed=12084`, `ready 1`, with history intact. Argo reports Synced and Healthy, and a
public-endpoint login attempt with an untrusted key was answered `unauthenticated`
(TLS and protocol fine).
Prevention:
- The deploy policy now refuses a `REGION_SESSION_WINDOW` that does not name its UTC
activation or that ends more than 4 h after it (two new selftest cases, one with the
`b2f224d` value).
- DEPLOYMENT.md documents the Deployment-only sync symptom and the hook re-run.
Executed: deploy policy check (0 failures) and selftest (46 cases, 0 failures).
## 2026-10-08 — M10.9d load drill harness (ADR-0097)
Implemented:
- `tools/ops/load-drill.sh` and its runbook `docs/networking/M10_LOAD_DRILL.md`:
synthetic inputs on for the run (restored on exit), `evo-load` through the public route,
and server reports matched to each hold.
- evo-server `m10 load-report` every 10 s. The agent cannot reach the ops port: no
port-forward, and the API-server proxy is refused.
- `evo-load --grants FILE --grants-per-bot C` with `exhausted=`, `hold_utc=` and
`close_reasons=`; `evo-session-probe grants --accounts N`.
- `include/evo/network/crypto_lock.hpp`: one recursive lock at every Mbed TLS/PSA entry
point of the network library (`MBEDTLS_THREADING_C` is off).
Findings:
- Concurrent first verifications on different threads were refused as `untrusted`, two
per smoke run in bundle mode (2 of 2 runs). With the lock, 3 of 3 runs had no failures.
- The drill summariser, tested on synthetic logs, counted reports from after the hold
(the next step's ramp). Fixed: only reports wholly inside the hold count.
Tests:
- M10 load smoke in bundle mode (8 accounts × 2 grants) with `exhausted=0`.
- The M10 process test checks `m10 load-report`.
- `quic-transport` `concurrent_clients`: 8 threads × 5 rounds all connect. It does not
reliably reproduce the race; the smoke does.
- The lease-loss test's close reasons read 24 × `shutdown` (drain) and 24 × idle
(SIGKILL).
Executed:
- **Linux GCC 14:** presets `quic` and `network` 5/5; `postgres-check.sh` with QUIC, all 18
suites.
- **Linux Clang 19:** clang-tidy 19 on `load.cpp`, the probe, `session_grant.cpp`,
`handoff_ticket.cpp`, `quic_endpoint.cpp`, `quic_transport.cpp`, evo-server `main.cpp`
with the bridge, and the M10 process tests (3 findings fixed); clang-format 19.
- **Windows MSVC 19.44:** presets `quic` 5/5 and `client-network` 15/15.
- **Drill script:** `bash -n`, plus the summariser on synthetic logs; shellcheck is not
installed here.
Not done: the reference run on evolution-dev. It needs this image promoted, a load bundle
and the maintainer's go-ahead.
## 2026-10-08 — M10.9d first evolution-dev load drill (ADR-0097)
Ran `tools/ops/load-drill.sh` on evolution-dev, with the image `6c96629e…` promoted and the
provision Job run with it. Steps 20/24/28/32, 60 s holds, 4 grants per bot, from the
maintainer workstation. Restore verified: `REGION_INPUTS` empty, epoch 60, `ready 1`, Argo
Synced and Healthy.
Result: not a capacity qualification.
- The governor refused from six seconds after the restart, with no session (p95 24.4 ms
over a partial window of startup ticks).
- 19 of 20 bots spent all 4 grants on refusals within about 10 s.
- The live log follow ended early, so only 4 server reports were captured.
Reference numbers with 1 session (`m10 load-report` at 14:33:58):
- tick work p95 4.0 ms, p99 and maximum 33 ms;
- session layer p95 0.68 ms, p99 3.0 ms;
- client response p95 111 ms, outcome p95 1.13 s, snapshots at 20 Hz.
Fixes (in the following commit):
- the governor acts on work only over a full window;
- evo-load backs off 10 s after a load refusal and reuses the unused bundle grant;
- the drill waits for a calm report, takes the pod log before restoring, and has a
stricter restore check.
The next run needs a new image and a new bundle: some of this bundle's grants are spent.
## 2026-10-08 — M10.9d second evolution-dev load drill and the shared replication truth
Second drill (image `7952185d…`), 18:21–18:26 UTC, restore verified:
- **20 sessions, client side:** all established; response p95 111 ms; outcome p95 about
1.2 s; snapshots at about 20 Hz; no starvation.
- **20 sessions, server side:**
- tick work p95 8–16 ms and p99 48–70 ms (budget 25 ms);
- session layer p99 1.2–2.0 ms per session (budget 0.5 ms);
- no tick debt;
- the governor refused from 18:22:15 and shed to level 3.
- **Startup:** the governor refused the first logins until 18:21:36, after a single calm
report. The drill now waits for two.
Fix:
- `ReplicationServer::tick` shares one encoding of the environment and organisms per
distinct interest; per session only markers, or everything with a perception
predicate.
- New `network` test: 21 sessions on two interests.
- Development reference, Linux Release:
- 21 sessions: about 356–366 µs per tick shared, against 947–995 µs per session
(old code, same test);
- existing replication tests unchanged.
Executed: see the commit's check list (Linux suites, clang-tidy, Windows). Not done: the
third reference run (needs the image, a new bundle and the go-ahead).
## 2026-10-08 — M10.9d third evolution-dev drill, budget v2, watermark coalescing
Third drill (image `f378a43d…`, 21:00–21:05 UTC). The pod was restarted first to load the
renewed certificate (checked with a throwaway-key login).
- **24 sessions, client:** all within budget (response p95 111 ms, 20 Hz, no starvation).
- **24 sessions, server:**
- tick work p95 9–16 ms, p99 25–34 ms;
- with no session the p99 is 23–25 ms (the commit flush);
- session layer 0.62–0.69 ms per session at p99;
- governor cap at 24.
- **Outcome latency invalid:** duplicate action ids, the reused accounts again.
Maintainer decisions:
- budgets v2 (p95 ≤ 25 ms over all ticks, commit ticks ≤ 50 ms; follow-up to move the
commit off the tick);
- optimise before declaring.
Changes:
- `evo-load`: action ids from the current time, `duplicate_actions`, receipts read before
outcomes; the drill ramps at 500 ms.
- The `load-report` splits poll and replicate and gives commit-tick and other-tick maxima.
- `IntentLimits::watermark_coalesce`, 4 in evo-server, plus an ADR-0092 amendment.
- `schemas/network/m10-budgets-v2.md`.
- The drill summary judges by v2.
Local profile, 24 bots (WSL, Release):
- `poll` about 1.3 ms against `replicate` about 0.45 ms per tick;
- temporary probes inside `poll` (reverted): event handling about 0.76 ms per tick
(9,600 datagrams per 10 s), watermark sends about 0.36 ms, control renewals about
0.8 ms each;
- with coalescing: watermark sends 4,800 → 1,200 per 10 s, `poll` about 1.05 ms per tick.
Executed:
- WSL Release `quic` build plus the quic and network CTest labels (including ops-contracts and
`session-fuzz-smoke`), passed.
- The 18 Postgres suites (including the `evo-load` smoke), passed.
- clang-tidy 19 on the changed sources, clean.
- `bash -n tools/ops/load-drill.sh`, passed.
- Windows `debug` 60/60, `quic` 5/5, `client-network` 15/15.
Not done:
- shellcheck: not installed in this WSL.
- The fourth drill (needs the new image, a fresh bundle and the go-ahead).
## 2026-10-09 — M10.9d fourth evolution-dev drill, batched durable lookups
Fourth drill (image `754446fe…`, 01:00–01:06 UTC, restored Ready):
- 24 sessions within the client budgets;
- the server was not within budgets v2 (commit ticks 58–70 ms, session p99 30–39 ms).
The cause was one synchronous store round trip per action, at admission and again at
commit (ADR-0097, fourth run). Fix:
- `ActionLedger::find_all` and `find_region_actions` (one statement over the schema v11
function, no migration);
- `IntentAdmission::admit`/`verify`, with one verify per poll in `RegionSessions`;
- one request per commit.
Executed:
- WSL Release `quic`: build, plus the quic, network, ops-contracts and session-fuzz CTest
(6/6), passed.
- The 18 Postgres suites, passed. They include the new `find_region_actions` checks:
order, absent pairs, empty, fenced, and `PostgresActionLedger::find_all`.
- `deferred_actions` in `network_control`.
- clang-tidy 19 on the changed sources, clean after two fixes.
- Windows `debug` 60/60, `quic` 5/5, `client-network` 15/15.
Not done: the fifth drill (needs the new image, a fresh bundle and the go-ahead).
## 2026-10-09 — M10.9d fifth evolution-dev drill
Image `c597c0ba…`, 11:28–11:33 UTC, restored Ready. Steps 20/24/28/32, 60 s holds:
- clients within budget at every step;
- server within budgets v2 at 24, 28 and 32, the configured session cap. The worst values
were work p95 12.9 ms, commit ticks 44.3 ms and session p99 13.3 ms, with no tick debt.
- At 20: 4 login refusals during the ramp, and per-session p99 0.554 ms (budget 0.5).
- Details are in ADR-0097, fifth run. Drill files were kept locally (load and server logs).
Not done:
- saturation beyond the 32-session cap;
- native-client observation during a drill.
## 2026-10-09 — M10.9 session cap 2000 (maintainer decision)
evo-server's session cap is 2000 (plus 64 connections of headroom). The library bounds
are 4096 in `SessionControl`, `IntentAdmission` and `ReplicationServer`. New tests check
that each accepts 2000 and refuses 4097. No drill was run, by decision: saturation and
native-client observation are deferred to a later 2k drill. Its prerequisites are in
ADR-0097.
Executed:
- WSL Release `quic`: the network test passed (after a test fix: the helper config's
`max_accounts` of 8 was below the 2000 cap), and quic/ops-contracts/session-fuzz
passed.
- The 18 Postgres suites passed; every evo-server process starts with the 2000 cap, and
the `evo-load` smoke passed.
- clang-tidy 19, clean.
- Windows `debug` 60/60 and `quic` 5/5.
Not run: Windows `client-network`. The maintainer's `evo-client.exe` was running from
that build directory, so the link could not replace it (LNK1104). The change touches no
client code.
## 2026-10-09 — engineering client: default server name for the dev route
The maintainer's Windows client could not connect to the dev region ("unreachable or its
certificate does not match"). The connect form defaulted the server name to
`dev-region.evolution.test`, the local test PKI's name, while the dev certificate is for
`void.aeternalisgames.com`. The default is now `void.aeternalisgames.com`, matching the
default endpoint. `client_ui_model` checks it, and `M10_CLIENT_SESSIONS.md` names it.
Executed:
- Windows `client` 9/9.
- Windows `client-network` 15/15. This also covers the session-cap change above, whose
`client-network` run had been blocked by the running client.
## 2026-10-09 — engineering client: used grants and UI scale
The maintainer found that their login failures came from a grant that had already been
used. The client had given up on the first `stale_grant`. Now a used grant spends only
itself: the session tries the next grant in the bundle, and a bundle of used grants ends
as exhausted. Other refusals still stop at once. The maintainer also reported that the
UI stayed at 13 px in a maximized window.
ADR-0084 amendment:
- the embedded scalable font;
- a UI scale from window height (720-point reference, 1.0–2.5) times a Ctrl+= / Ctrl+-
nudge;
- panel sizes and table columns in logical units;
- movable panels placed again at each new scale.
`--size W,H` sets the initial window size, for engineering review.
Executed:
- `client_remote_session`: a stale grant moves to the next grant; an all-used bundle
ends as exhausted.
- `client_ui_state`: scale clamps, nudge grid, nonsense inputs.
- Windows `client` 9/9 and `client-network` 15/15.
- Screenshots of the launcher and the fixture at 960×640 and 1920×1080, reviewed: the
scale is 1.5 at 1080, text is crisp, and the inspector labels fit.
Not run: an RDP display-scale check. Under RDP the logical height drives the scale, the
same as on a local display.
## 2026-10-09 — M11.1 time-based rates, segment 1 (ADR-0099)
`accounting::per_tick`, `exact_per_tick` and `per_second_from_per_tick` were added,
with a new `rate-contracts` test. The test covers:
- exact ×20 migration and round trip at 50 ms, under both roundings;
- directional rounding at an inexact tick, bracketing the exact amount one unit apart;
- sub-unit rates;
- tick bounds;
- the largest rate;
- overflow;
- refusal of an inexact migration.
Executed:
- WSL `linux-debug` and `linux-asan`: `rate-contracts` and `ledger-contracts` passed.
- clang-tidy 19 (`linux-analysis`): clean.
- Windows `debug`: 61/61.
No biology uses the conversion yet, so no golden is affected.
## 2026-10-09 — M11.1 time-based rates, segment 2: genome schema 2 (ADR-0099)
Genome schema 2 stores movement and ingestion per second (ranges ×20;
`schemas/genome-v2.md`). Schema 1 records stay valid and read as exactly ×20, and drafts
default to schema 1. Development version 1 converts genome rates at its 50 ms quantum.
Mutation keeps the parent's schema and step. The lineage archive header names exactly
the highest schema held, and the development opening names schema 1.
New tests:
- `schema_v2_rates` (genome);
- range boundaries per schema;
- `schemas_mix_without_rewriting` (archive);
- `schema_v2_rates_develop_identically`: phenotypes and plans equal across schemas, and
sub-tick rates round down;
- `successors_keep_their_parents_schema` (mutation).
Two existing tests used 2 as "an unknown future schema". They now use the next unknown
schema, and the archive decoder now refuses a header that misstates its contents. No
golden changed: the frozen schema 1 record and its digest are byte-identical.
Executed:
- WSL `linux-debug` and `linux-asan`: 61/61 each.
- clang-tidy 19 on the changed sources and tests: clean.
- The 18 Postgres suites passed, including region upgrade and clone comparison, so
checkpoints are unchanged.
- Windows `debug`: 61/61.
## 2026-10-09 — M11.1 time-based rates, segment 3: environmental rules 3 (ADR-0099)
Environmental community rules 3 has rules 2's layout with irradiance, growth, intake
and decay per second. It derives a per-tick form at creation and restore: exact
rationals, caps rounded down. Its format is 3 (`schemas/environmental-community.md`).
Development version 1 outputs stay tied to their 50 ms quantum, and the
tick-abstract M2 fixtures are not migrated. Both are recorded in ADR-0099.
`rules3_per_second` tests:
- a ×20 twin of a rules 2 configuration over 60 ticks, through feeding, a consumer
death and decay, with reports and every inventory equal each tick;
- snapshot and restore as format 3, then identical steps;
- at 25 ms, a 2000-per-second intake cap admits 50 and refuses 51;
- an overflowing rate and an over-long tick are refused.
`environmental-qualification` used rules 3 as "unknown". It now uses 4, with the same
intent.
Executed:
- WSL `linux-debug` and `linux-asan`: 61/61 after that change.
- clang-tidy 19 on the changed sources and tests: clean.
- The 18 Postgres suites: passed.
- Windows `debug`: 61/61.
## 2026-10-09 — M11.1 development version 2 and the primitive founder (ADR-0099)
Development version 2 is defined per second. The rate period follows from the
development version (`development_rate_microseconds`), and the budget carries it as
`rate_microseconds`. The M2 fixture adapters refuse anything but version 1.
`bud_funding_seconds` and `primitive_founder` were added.
`development_v2_and_the_first_bud` checks:
- the hand-calculated per-second phenotype and budget;
- first bud at 300, 600 and 901 s for feeding 100%, 50% and 33% of the time, against
the bounds of the decision's 5–15 minute range;
- never at 10%;
- version 1 rates stay per 50 ms;
- the fixtures refuse version 2.
Development and metabolism tests that used version 2 as unknown now use 3.
Executed:
- WSL `linux-debug` and `linux-asan`: 61/61 each.
- clang-tidy 19: clean, after a zero-surplus guard and value reads in the test.
- The 18 Postgres suites: passed.
- Windows `debug`: 61/61.
## 2026-10-09 — CI run 136 network failure: the phenotype stays wire-stable
CI run 136 (`51cf23d`) failed `network` at the organism-view canonical re-encoding
(`network_protocol.cpp:162`). The cause was `Phenotype::rate_microseconds`, added with
development version 2. Both session protocols encode phenotypes field by field, so the
period did not survive a round trip. My local checks had not built the network presets.
Fix: the phenotype is restored to its replicated shape. The rate period follows from
the development version (`development_rate_microseconds`), and `MetabolicBudget`,
which is not replicated, carries it. No protocol change; first-bud times unchanged.
Executed:
- WSL `linux-debug` 61/61 and `linux-asan` 61/61.
- WSL `network-clang` 65/65 and `quic-linux` 68/68.
- clang-tidy 19: clean.
- Windows `debug` 61/61, `quic` 5/5, `client-network` 15/15.
## 2026-10-09 — M11.2 contract review (documentation only)
Proposed ADR-0100 and M11_2_DELIVERY_PLAN document two incompatible reuse paths:
the two-actor checkpoint's exact living-set closure and the no-effect action
publication's first-result-wins semantics. They propose a new cellular profile and
atomic life/funding/result publication. Starting funding remains a maintainer
decision; no M11.2 implementation or gate completion is claimed.
Executed: `git diff --check`, plus whitespace and relative-link checks on both new
documents. Reviewed existing headers, migration 0011 and relevant tests as listed
in the delivery plan. No C++/SQL/wire code changed; builds, CTest, clang-format,
clang-tidy and sanitizers were not run for this documentation-only change.
No environment limitation was encountered during these documentation checks.
No deployment, world mutation or provisioning was performed.
## 2026-10-09 — M11.2a funded primitive creation
Implemented portable server lifecycle orchestration and the biology-derived
founder endowment (ADR-0100). Tests independently check composition, reserve and
heat, account equivalence, successful retries at exhausted capacity/funding,
per-channel shortage rollback, identity conflicts, invalid/overlapping sites,
empty funding and u64 channel bounds. No durable spawn or session integration
is claimed yet. No existing golden or wire layout changed.
Executed:
- Linux GCC 14 debug: 62/62 CTest checks; final test-literal revision rechecked.
- Linux Clang 19 ASan+UBSan: 62/62.
- Windows MSVC release (NDEBUG): 62/62.
- clang-tidy 19 on changed sources/tests: clean after correcting a widened test
multiplication; Clang also caught a signed initializer-list conversion, fixed.
- Repository format-check and `git diff --check`: passed. The formatter first
found a pre-existing mixed newline in session_window.hpp; normalization is
semantically empty and produces no Git content diff.
Environment: the sandboxed Windows Ninja build stalled before compiler execution;
the same release preset succeeded outside the sandbox. WSL required escalated
access. No reference-worker load qualification or deployment was performed.
## 2026-10-09 — M11.2b creation-state checkpoint codec
The primitive-lives-v1 envelope restores exact source, heat, bodies, reserves,
archive, allocation and retry state. Tests cover empty/populated recovery,
identical continuation, every truncated prefix, trailing bytes, oversized input,
unknown versions/quantum, mismatched scope, and resealed corruption of placement,
growth, ownership, funding and archived history. No durable-store claim yet.
Executed: Linux GCC 14 debug lifecycle test, Clang 19 ASan+UBSan lifecycle test,
Windows MSVC release 62/62 CTest checks, clang-tidy 19 on the two changed source/test
files, repository format-check and `git diff --check`. All passed. The Linux
full-suite results from 11.2a are not claimed as new codec runs. No wire type or
deployed profile changed; no deployment occurred.
## 2026-10-09 — M11.2c-1 durable life reservations
Schema v15 adds a world allocator, one pending reservation per account and
retained refusals. Reservation success allocates identity only; it neither spends
funding nor publishes a body. Dedicated real-PostgreSQL tests cover independent
connections/regions, concurrent retries, takeover routing, stale generations,
expiry after allocator lock waits, exhaustion, rollback, immutable refusal
results and runtime table-access rejection.
Executed successfully:
- Focused `tools/ci/postgres-check.sh --life-reservations-only` against a fresh,
disposable PostgreSQL 18.6 cluster, including the final exhaustion/rollback cases.
- Full GCC 14 PostgreSQL/QUIC build in `build/m97-postgres`, then the complete
`tools/ci/postgres-check.sh build/m97-postgres/evo-postgres-tests`: adapter,
lease/control, checkpoint, host/session, process-kill, recovery, clone, migration,
backup/restore, host-loss, upgrade, constraints and effective privilege checks.
Log: `/tmp/m112c-pg-correct.log` in WSL; final success marker verified.
- Shell syntax checks and CI path-selection regression checks (16 cases).
- Clang 19 clang-tidy on `tests/region_migration_tests.cpp` using the Clang
PostgreSQL compilation database, repository format-check and `git diff --check`.
Environment: WSL required escalated access. An earlier run selected an obsolete
build cache with PostgreSQL disabled and was stopped; it is not counted. An initial
clang-tidy invocation used a GCC compilation database and rejected GCC module flags;
the proper Clang database completed cleanly. No sanitizer rerun is claimed for this
SQL subsection; prior lifecycle sanitizer results remain recorded above. No wire
type changed. All database work used disposable local clusters; no world deployment,
reset or provisioning occurred.
## 2026-10-09 — M11.2c-1 reservation adapter
Added the Linux C++ reserve/refuse API and real-database tests for unsigned IDs
above INT64_MAX, UINT64_MAX request IDs, exact reconnect retries, stable refusals,
missing provisioning, invalid inputs, exhaustion, stale control and takeover
routing. A deliberate allocator lock wait produces an unknown mutation response;
reconnecting with the same request reconciles exactly one allocation.
Executed: GCC 14 Release build and focused PostgreSQL adapter/SQL/concurrency
tests; Clang 19 ASan+UBSan build and the same focused tests; clang-tidy on the
adapter and both affected test files; full format-check, shell syntax and diff
checks. All passed. The full PostgreSQL/process suite above belongs to the schema
subsection; it was not repeated for these adapter-only changes. No wire format,
deployed world or SQL migration changed in this subsection.
## 2026-10-10 — M11.2c-2a creation-batch replay
Added bounded primitive-batch-v1 trace encoding and detached funded replay.
Checks cover canonical round trips, every truncated prefix, length/version and
reserved fields, maximum counts, ordered unique births, scope and tick bounds,
duplicate old requests, exact replay equivalence and channel conservation. A
second birth with insufficient funding discards the first candidate birth too;
empty batches preserve the complete biological payload.
Executed: Linux GCC 14 debug lifecycle test; Clang 19 ASan+UBSan lifecycle test;
Windows MSVC Release/NDEBUG full 62/62 CTest checks; clang-tidy 19 on new source
and affected lifecycle tests; repository format-check and diff checks. All passed.
The static-analysis session was interrupted across the chat pause and rerun to
confirmed completion. No existing network type, golden, database profile or
deployment changed; atomic database publication remains unimplemented.
## 2026-10-10 — M11.2c-2b portable checkpoint profile
Primitive profile 3 declares history profile 2, its version bundle, exact 50 ms
quantum, funded payload closure and bounded creation traces. Tests cover genesis,
successors, idle cuts, future/zero births, scope, versions, lengths, trace bounds,
foreign identity allocations and rejection of unsupported clone genesis. Existing
environmental profile tests and goldens are unchanged. SQL v15 still rejects
profile 3; portable support does not yet enable durable life publication.
Executed successfully: Linux GCC 14 debug 62/62, Clang 19 ASan+UBSan 62/62,
Windows MSVC Release/NDEBUG 62/62, Windows network preset 2/2, and Windows QUIC
preset 5/5 on an isolated rerun. The first QUIC run failed the existing healthy-idle
heartbeat check at network_session.cpp:618 while network compilation and sanitizer
tests ran concurrently; every other QUIC check passed. The same binary passed all
five checks without competing build/test load. No network source was changed.
Clang-tidy 19 on the changed implementation/tests, full format-check and diff checks
passed. No database migration, provisioning or deployment was performed here.
The PostgreSQL server/clone caller build exposed GCC 14 Release's
`free-nonheap-object` diagnostic in the new trace encoder's vector growth path.
The encoder now allocates its exact bounded size once and writes through a cursor;
no warning suppression or format change. The final GCC Release server/clone build
and both lifecycle/checkpoint tests passed. The revised encoder was rechecked with
ASan+UBSan (both tests), clang-tidy/format, and Windows Release 62/62. WSL temporary
logs disappeared after process exit; the reproduced build log was retained under
`build/m112-profile-callers.log` instead.
## 2026-10-10 — M11.2c-2c schema v16 guarded publication
Executed the GCC 14 Release `evo-postgres-tests` build and complete
`tools/ci/postgres-check.sh` against PostgreSQL 18.6. This includes the new
primitive publication integration (reservation/control checks, changed and exact
retry behavior, rollback and concurrent retries), every existing process and
recovery integration, schema upgrade/checksum/privilege audits, backup/restore,
host-loss and failed-upgrade rehearsals. The full command passed. A prior run
correctly reported the historical v3 fixture had 16 migration rows while the
script still expected 15; the expectation was updated and the complete run then
passed. The focused publication/concurrency lane also passed after the fixture
used commit IDs unique across this shared test database.
No production world was provisioned or altered. This segment adds SQL profile
admission and a runtime binding read, but not the C++ object-store publication
adapter or primitive-capable server recovery/session path. Do not interpret its
green database gate as end-to-end life recovery. Full C++ unit/network/QUIC lanes
were not rerun because this segment changes no replicated network type.
## 2026-10-10 — M11.2c-2d persistence adapter
Added verified immutable-object publication for profile 3, exact prepared manifest
and trace retry, portable batch replay from the committed predecessor, and recovery
that checks payload closure, head-trace replay and durable life-binding agreement.
The focused fresh-cluster publication/recovery test passed. GCC 14 Release build,
clang-format 19, clang-tidy 19 on the new adapter and test, and the complete
`tools/ci/postgres-check.sh` passed, including all existing host/session/process,
backup/restore, host-loss, upgrade and migration checks. The final full run is in
`build/m112-c2d-full.log`.
No production world was provisioned or changed. The new adapter can persist and
recover profile 3, but `RegionHost` and session routing remain limited to the
existing environmental/lineage profiles; no primitive session can yet reach this
adapter.
## 2026-10-10 — M11.2d-1 session protocol v5 (in progress)
Added the v5 wire schema and codec changes for a separate reliable primitive-life
request, receipt, outcome and query family. v4 remains the default, and current
RegionSessions explicitly rejects v5 until per-connection negotiation and life
handling land. MSVC compilation of the changed network test units, protocol object,
client/server session bridge objects succeeded; affected libraries and test binaries
were relinked from CMake-generated commands. `ctest --preset network` passed (2/2),
and `ctest --preset quic` passed (5/5), including the new assertion that an
unnegotiated v5 Hello is refused.
The ordinary `tools/windows-check.ps1 -Preset network -Action check` invocation
still configures successfully, then leaves Ninja idle without output or child
compiler processes. WSL startup is denied by this host. Thus these are targeted
MSVC compile/link and CTest results, not a successful full preset build.
The operator's live-region smoke test did pass using the prebuilt v4
`evo-session-probe`, `C:\Dev\tokens\m10-grants.bin` grant 0 and the supplied dev CA:
`void.aeternalisgames.com:7400` returned Welcome, exit 0. This confirms existing v4
connectivity only; it does not execute or validate v5 or primitive-life handling.
## 2026-10-10 — M11.2d-2 durable primitive-life outcomes (unverified)
Implemented schema v17 durable `existing_life` outcomes, current-controller
world/account/request lookup, and atomic cancellation of unpublished reservations
when control changes region or region epoch. Updated v5 outcomes and typed adapter
tests. Direct clang-format 19.1.5 dry-run on all changed C++ files and `git diff
--check` passed.
The network preset configured successfully, but its check invocation and a direct
`evo-network-tests` Ninja target remained silent without completion for more than
35 seconds and were interrupted, matching the previously recorded idle Ninja
condition. Direct MSVC Release/NDEBUG compilation succeeded for the modified
network protocol/test units, persistence adapter and its changed PostgreSQL test
units. The protocol library and network test executable were rebuilt from those
objects; `ctest --preset network` passed (2/2) and `ctest --preset quic` passed
(5/5).
`psql` and PostgreSQL 18.6 are unavailable in this environment; Docker's Linux
engine is also not running. Schema v17 and its PostgreSQL integration, migration,
upgrade and privilege tests have not run locally. CI must pass these before the
host/session subsection advances; the maintainer will confirm that result.
## 2026-10-10 — M11.2d-3 primitive region host (CI pending)
Added a separate creation-only `PrimitiveRegionHost`: fenced lease acquisition
and renewal, profile-3 recovery before readiness, durable new-life reservation and
refusal handling, detached replay, atomic publication and exact retry/query
reconciliation. It does not advance biology and is not wired to `evo-server` or
QUIC. Added a real-PostgreSQL integration case for recovery, first birth, exact
retry and world/account exclusivity.
Executed: GCC 13.2 Release/NDEBUG compilation with warnings-as-errors for the new
host translation unit and changed PostgreSQL test translation unit; `git diff
--check`. Both compilations passed. The local Windows shell cannot configure the
Linux-only PostgreSQL target, and this turn did not run the PostgreSQL test binary.
CI found clang-format violations in the three new/changed host and test files plus
the `performance-move-const-arg` clang-tidy warning on the v17 outcome adapter.
Both were corrected. clang-format 19.1.5 dry-run and clang-tidy 19.1.5 on
`region_lease_store.cpp` pass locally. Awaiting CI rerun for the database,
migration, host test and repository checks; no local runtime pass is claimed.
The subsequent CI build found one more `performance-move-const-arg` and eight
`bugprone-unchecked-optional-access` errors in `primitive_region_host.cpp`.
Removed the ineffective move and made lease grants and nested optional values
explicitly checked before access. clang-format 19.1.5 dry-run, clang-tidy 19.1.5
with warnings-as-errors, GCC 13.2 Release/NDEBUG compilation with warnings-as-
errors, and `git diff --check` pass locally. PostgreSQL integration remains
unavailable in this Windows environment; awaiting CI rerun.
The next CI analysis build reached the PostgreSQL tests and reported four
`bugprone-unchecked-optional-access` errors in the life-reservation test's nested
`expected<optional<...>>` checks. Reworked those assertions to extract safe
fallback values while still checking both success/presence states. clang-format
19.1.5, clang-tidy 19.1.5 with warnings-as-errors, GCC 13.2 Release/NDEBUG
compilation with warnings-as-errors, and `git diff --check` pass locally. The
PostgreSQL test binary remains unrun locally; awaiting CI rerun.
Windows clang-tidy 19.1.5 additionally reports `bugprone-exception-escape` on
the pre-existing PostgreSQL connection lambda in this test; this warning was not
present in the reported Linux CI diagnostics. The local tidy run excluded only
that check, so Linux CI remains authoritative for it.
The next PostgreSQL CI run passed analysis compilation and the publication tests,
then failed the online migration integration check because its expected schema
history stopped at v16. Updated that expectation through v17. clang-format 19.1.5
and `git diff --check` pass. This Linux-only test cannot compile in the Windows
shell because its process fixture requires `poll.h`; awaiting CI compile and
runtime confirmation.
That CI run then passed the migration integration but failed the focused SQL
reservation check: it still expected v15's `EVL16` conflict, while v17 records a
durable `existing_life` refusal and allocates its own identity. Updated the check
to assert the returned identity and rejection value, and aligned the reservation
contract text with v17. PostgreSQL is unavailable in this Windows environment;
the focused and full database checks await CI.
The next run failed that assertion because the new request was first issued only
after the original reservation had been refused. Moved the first conflicting
request before that refusal, then kept the later call as an exact-retry check.
The maintainer confirmed the CI rerun passed after `4c4ab36`; the v17 migration,
reservation contract check and full PostgreSQL gate are green. PostgreSQL remains
unavailable locally. This closes validation for 11.2d-2 and 11.2d-3; the primitive
host is still not connected to `evo-server` or enabled for v5 sessions.
## 2026-10-10 — M11.2d-4 primitive owner provisioning (CI pending)
Extended `evo-world-clone provision` with an explicit `primitive-v1` opening.
The operator must supply the frame, source ID, energy/carbon/nitrogen/phosphorus/
other funding values and ordered habitat sites. The tool uses the profile-3
object limit, refuses a different existing profile/frame or a changed primitive
opening, and only reports an existing primitive opening after full recovery and
state comparison. Added a PostgreSQL process test for runtime-role refusal,
owner provisioning, exact retry, frame/opening mismatch and recovered values.
Executed locally: Visual Studio clang-format 19.1.5 dry-run on both changed C++
files and `git diff --check`; both passed. The Windows environment cannot run the
Linux-only PostgreSQL process test. WSL launch was denied with `E_ACCESSDENIED`,
and the configured Linux compiler is not available directly in PowerShell. The
Windows repository format target stalled in Ninja and was stopped; formatting
was verified directly with the same pinned clang-format version. The primitive
provisioning build and PostgreSQL process test await Linux CI. No deployed world
was provisioned or changed.
## 2026-10-10 — M11.2d-5 primitive evo-server selection (CI pending)
Added explicit `--profile primitive` startup through `PrimitiveRegionHost`, with
profile-3 payload bounds and readiness only after lease acquisition and complete
checkpoint recovery. The primitive mode refuses synthetic ticks and M9/M10
session options; it does not provision genesis or advance biology. The real-
process test checks successful recovery and ready state, unchanged genesis tick,
lease release on SIGTERM, refusal of synthetic tick mode, and quarantine without
genesis.
Executed locally: clang-format 19.1.5 dry-run on the changed C++ files and
`git diff --check`; both passed. The Linux-only server and PostgreSQL process
tests cannot run in this Windows environment: WSL launch was denied with
`E_ACCESSDENIED`, and the configured Linux compiler is not available directly in
PowerShell. CI compilation and runtime qualification are pending. No deployed
world was provisioned or changed.
## 2026-10-10 — M11.2d-6 authenticated v5 session execution (CI pending)
`RegionSessions` now negotiates a per-connection protocol version, keeps existing
regions v4-only, and enables primitive-profile v5 only with durable account control
and both life handlers. V5 welcomes and stream replies use v5; mismatched versions
close. Life requests require the authenticated session id, current control epoch,
ready host and nonzero durable generation; the callback receives the verified grant's
account identity, never a client-supplied account. Query is read-only. Primitive
`evo-server` can opt into the dedicated QUIC listener with an issuer key and
database-backed control ledger. A QUIC contract test exercises v5 Hello/Welcome and
life receipt/outcome with a fake durable ledger.
Ran clang-format 19.1.5 on changed C++ files and `git diff --check`; both passed.
MSVC 19.44 Release/NDEBUG compilation of `region_sessions.cpp` and
`quic_sessions.cpp`, manual relinking of the focused executable, and
`ctest --preset quic` all passed (5/5). The normal Ninja build stalled as recorded
above. PostgreSQL is unavailable locally, and the server process target is Linux-only;
full `evo-server` compilation and real PostgreSQL process results await CI.
Cross-region routing remains incomplete because the accepted v5 outcome does not
carry an owner region/address; see the delivery plan. No deployed world was provisioned
or changed.
## 2026-10-10 — M11.2d-6 client and probe callers (CI pending)
`RegionSessionClientConfig` now selects v4 or v5 explicitly (v4 default). The client
encodes Hello and stream calls in its selected version, collects primitive-life
receipts/outcomes, and exposes create/query methods only for v5. M10 movement, action,
and interest methods are unavailable in v5. `evo-session-probe` accepts `--protocol v5`,
`--new-life ID` and `--life-query ID`, prints replies, retries unresolved exact requests,
and rejects combinations with M10 actions or movement options. The QUIC session test
now exercises these client APIs end to end with a durable-control fake.
Targeted MSVC Release/NDEBUG compilation of the modified session client, QUIC test and
probe, manual relink of the test/probe, and `ctest --preset quic` passed (5/5). Formatting
and `git diff --check` passed. Full server/process/PostgreSQL qualification remains in CI.
The Linux/PostgreSQL process suite now includes an isolated profile-3 lifecycle:
owner-provision genesis, authenticate over v5, publish a founder, gracefully restart,
and query the same immutable identity/commit receipt after recovery. This regression
was added but cannot run locally on this Windows host; its CI evidence remains pending.
## 2026-10-10 — M11.3b-1 phenotype-derived motion proposal (CI pending)
Added the pure cellular motion proposal and square-footprint contact helper under
`evo::biology`. It uses the v2 per-second founder budget and 50 ms tick, clamps the
20 μm-class body inside a bounded trial area, reports movement energy without
spending it, and rejects invalid vectors, positions and rate periods. Focused tests
cover cardinal/diagonal movement, directional rounding, energy cost, boundaries,
unsupported rates and edge contact. No network, persistence, feeding or server
integration is included yet.
Executed locally: MSVC 19.44 warning-as-error Debug compile/link and focused CTest
(1/1 passed); direct MSVC Release/NDEBUG compile/run (passed); clang-format 19.1.5
dry-run; clang-tidy 19.1.5 against the generated compile database; and
`git diff --check`. The CMake configure completed with Visual Studio CMake 3.31.6
and module scanning disabled (the project does not use C++ modules), but the Ninja
build target idled at 0 CPU and was stopped; the test executable was compiled
directly and then run by CTest. Linux GCC, ASan+UBSan and server/PostgreSQL process
qualification remain pending CI. This small resolver does not make 11.3 or a
playtest-ready cellular stage complete.
## 2026-10-10 — M11.3c-1 channel-limited uptake proposal (CI pending)
Added a pure uptake proposal that computes each 50 ms intake capacity from the
development-v2 per-second budget, then independently bounds energy and C/N/P/other
by patch stock and reserve headroom. It returns both post-transfer inventories and
the exact amount acquired; it does not mutate either side, so the cellular host can
publish them as one candidate. ADR-0101 now states that absorbed material accumulates
in reserve without changing the fully formed body.
Executed locally: MSVC 19.44 warning-as-error Debug compile/link and focused CTest
(1/1 passed); direct Release/NDEBUG compile/run (passed); clang-format 19.1.5
dry-run; clang-tidy 19.1.5 against the generated compile database; and
`git diff --check`. Linux GCC and sanitizer runs are unavailable locally; the full
cellular-owner conservation, contention, durable publication and PostgreSQL process
checks remain pending CI. This helper does not yet close the server-side transfer.
## 2026-10-10 — M11.3a-1 finite cellular habitat map (CI pending)
Added the region/frame-scoped cellular habitat owner: explicit 1 mm patch
coordinates, per-patch provenance and finite C/N/P/other/energy openings; strict
area, ordering, count and total-inventory bounds; and a canonical versioned
snapshot/restore codec. It refuses absent patches rather than supplying stock.
Read-only uptake proposals combine the patch stock with the previous rate-limited
proposal, leaving publication to the later cellular host.
Executed locally: MSVC 19.44 warning-as-error Debug compile/link and focused CTest
(1/1 passed); direct Release/NDEBUG compile/run (passed); clang-format 19.1.5
dry-run; clang-tidy 19.1.5 against the generated compile database; and
`git diff --check`. Linux GCC, ASan+UBSan and cellular server/PostgreSQL process
qualification remain pending CI. This habitat codec does not replace or alter
profile 3 and is not yet included in a live region checkpoint.
## 2026-10-10 — M11.3a-2 cellular perception projection (CI pending)
ADR-0103 fixes the exploratory sensor radius at 5,000 μm from the organism's
square footprint to a patch square. The pure habitat query rejects invalid
footprints and returns only canonical, in-range provisioned patch addresses and
exact available inventories; source provenance and out-of-range stock are not
part of the projection. The range is about 25 seconds of starting-founder
full-speed travel and is a trial value, not a final sensory rule.
Executed locally: warning-as-error MSVC 19.44 Debug compile/run and Release
`NDEBUG` compile/run of the focused habitat tests (both passed); focused CTest
for cellular motion, uptake and habitat (3/3 passed); clang-format 19.1.5
dry-run; clang-tidy 19.1.5 using the MSVC `/std:c++latest` argument; and
`git diff --check`. Linux GCC, ASan+UBSan, authenticated server visibility and
the full CI suite remain pending CI. This projection is not yet wired into live
replication or a persistent cellular checkpoint.
## 2026-10-10 — M11.3c-3 single-life cellular tick (CI pending)
Added ADR-0104 and a pure one-life tick candidate. It debits per-tick
maintenance first, attempts movement only from the remaining starting reserve,
and applies one explicitly targeted intake only after movement and contact
validation. An unaffordable move leaves the cell still while maintenance and
valid intake remain in the candidate; unaffordable maintenance refuses the
candidate for same-tick starvation resolution by the later life owner. The
returned balances conserve energy and each matter channel. No owner, shared-stock
arbitration, death closure or durable publication has been integrated.
Executed locally: warning-as-error MSVC 19.44 Debug and Release `NDEBUG`
compile/run; focused CTest for motion, uptake, habitat and the composed tick
(4/4 passed); clang-format 19.1.5 dry-run; clang-tidy 19.1.5 using MSVC
`/std:c++latest`; and `git diff --check`. Linux sanitizers, shared-life
contention and server/checkpoint integration remain pending CI and later gates.
## 2026-10-10 — M11.3c-2 contact-gated patch uptake (CI pending)
The habitat uptake proposal now validates the cell footprint against the finite
arena and requires footprint contact with the requested stocked patch before it
calculates any transfer. Missing and non-contact patches are distinct refusals;
the helper still only proposes post-transfer balances and does not publish or
mutate either owner.
Executed locally: warning-as-error MSVC Debug and Release `NDEBUG` compile/run,
focused CTest for motion, uptake and habitat (3/3 passed), clang-format 19.1.5,
clang-tidy 19.1.5 with MSVC `/std:c++latest`, and `git diff --check`. CI
qualification, simultaneous claims and atomic patch/life publication remain
open.
## 2026-10-10 — M11.3b-2 funded movement candidate (CI pending)
Added a detached movement transition that combines bounded phenotype-derived
motion with reserve sufficiency and an energy-only reserve-to-heat transfer.
Insufficient reserve or heat overflow refuses the complete candidate; matter
and caller-owned inputs remain unchanged. Maintenance remains a separate
tick-ordered transition for the later cellular state owner.
Executed locally: warning-as-error MSVC Debug and Release `NDEBUG` compile/run,
focused cellular CTest, clang-format 19.1.5, clang-tidy 19.1.5 with MSVC
`/std:c++latest`, and `git diff --check`. CI and integration with durable
intent/outcome publication remain pending.
## 2026-10-10 — M11.3b-3 cellular intention v6 codec (CI pending)
ADR-0105 and `schemas/network/session-v6-cellular-intention.md` define a strict
fixed-size datagram carrying only session/epoch, sequence/tick, movement intent
and an optional intake patch address. V1-v5 encodings remain unchanged. Decode
rejects noncanonical absent targets and reserved bytes. The codec intentionally
does not admit this message in `RegionSessions`; that server path is not yet
implemented.
Executed locally: warning-as-error MSVC 19.44 Release compile/run, CTest
`cellular-intention-protocol-v6` (1/1 passed), compile-only MSVC check for the
existing `network_control.cpp` regression case, clang-format 19.1.5 on changed
ranges, clang-tidy 19.1.5 with MSVC `/std:c++latest`, and `git diff --check`.
The CMake/Ninja build of `evo-network-tests` remained idle for about 90 seconds
(cmake 0.02 CPU seconds, ninja 0.09, no `cl`/`link` child), so I interrupted it;
the full M10 network suite and server admission/QUIC integration remain pending.
Linux network CI has not yet qualified this version.
## 2026-10-10 — M11.3b-4 cellular intention admission (CI pending)
Cellular v6 intentions now pass through the established authenticated-session,
epoch, movement-bound, increasing-sequence, future-tick, rate and unresolved
input checks. The pending schedule is read-only until the simulated watermark
advances and orders by tick, account, session ID and sequence, including the
case where one account has multiple open sessions. The cellular region owner
and durable publication path still do not consume this schedule.
Executed locally: warning-as-error MSVC 19.44 Debug and Release/NDEBUG direct
compile/run of the focused admission regression; existing CTest
`cellular-intention-protocol-v6` (1/1 passed); clang-format 19.1.5 on the new
test, changed header and implementation ranges; clang-tidy 19.1.5 on the changed
test and implementation using `-std=c++23`; and `git diff --check`. The network
CMake configure stalled during MSVC compiler detection, so the new CTest
registration and complete network build await CI. Linux and full M10 network
checks remain pending.
## 2026-10-10 — M11.3b-5 opt-in v6 session integration (CI pending)
`RegionSessionsConfig::accept_cellular_intentions` now opts into v6 logins and
routes cellular datagrams through bounded intention admission. The host reads
the deterministic tick schedule before advancing the simulated watermark; the
client sends movement and an optional patch address. V6 cannot invoke M10 probe
actions, request M10 interest, or receive default M10 probe snapshots/markers.
The path still has no cellular life owner, durable publication or cellular view.
Executed locally: warning-as-error MSVC 19.44 Release/NDEBUG compile of the
changed server, client and QUIC test translation units; direct relink of the
prepared QUIC session-test binary; CTest `quic-sessions` (1/1 passed, including
the v6 acceptance/refusal and legacy-view isolation cases); clang-format 19.1.5
on changed ranges; clang-tidy 19.1.5 on all three changed C++ files using
`-std=c++23`; and `git diff --check`. CMake regeneration completed, but Ninja
idled without launching a compiler, so the changed objects and test executable
were built directly in the existing `build/quic` tree. Linux CI qualification
and the full application build remain pending.
## 2026-10-10 — M11.3c-4 deterministic shared-patch candidate (CI pending)
Added ADR-0106's trial arbitration: same-tick claims resolve by ascending stable
life identity and use the patch's remaining per-channel stock. The pure batch
candidate returns all life results and final patch stocks together; if a life
cannot produce its one-tick proposal, it refuses the whole candidate. It does
not close starvation into a corpse or publish state.
Executed locally: warning-as-error MSVC 19.44 Debug and Release/NDEBUG direct
compile/run of the focused region-step regression; Debug direct compile/run of
the existing habitat and single-life step regressions; clang-format 19.1.5 on
the new/changed C++ ranges; clang-tidy 19.1.5 with `-std=c++23` on the changed
implementation and test; and `git diff --check`. The tests confirm input-order
independent arbitration, per-channel conservation, no mutation of the source
habitat and no partial result for duplicate identities or starvation. CTest and
Linux sanitizer/CI qualification remain pending.
## 2026-10-10 — M11.3c-5 starvation and corpse transfer (CI pending)
ADR-0107 closes unaffordable maintenance in the same tick. The detached region
candidate removes the starved life, transfers its body and reserve matter to a
corpse, and reports accumulated heat separately. The helper still does not own
or durably publish the heat ledger, lineage outcome, habitat, or action result.
Executed locally: warning-as-error MSVC 19.44 Debug and Release/NDEBUG direct
compile/run of the focused region-step regression; the regression confirms
same-tick death, exact body and energy/matter reserve transfer, separate heat
output, no patch mutation, and unchanged all-or-nothing handling of duplicate
identities; and `git diff --check`. `clang-format` and `clang-tidy` were not installed or
available on PATH in this Windows environment. CTest, Linux sanitizer/CI
qualification, durable heat ownership, lineage and checkpoint publication
remain pending.
## 2026-10-10 — M11.3d profile-4 closure contract (implementation pending)
ADR-0108/0109/0110 and `schemas/cellular-lives-v1.md` specify an additive complete
cellular checkpoint with region-wide channel conservation while preserving
profile 3. This is documentation only;
no codec, SQL admission or recovery behavior is implemented or validated by
this entry. `git diff --check` is the applicable local check. Profile-4 code,
PostgreSQL CI and crash/restart tests remain required.
## 2026-10-10 — M11.3c-6 in-memory cellular owner (CI pending)
Added `server::CellularLives` as a portable lifecycle owner over funded
profile-3 opening state. It derives each life budget from its archived genome,
advances a complete stable-ID intention set against one shared habitat, and
atomically applies living results, death records, corpses, heat and patch stock
only when the region-wide totals still reconcile. The focused test covers
contested intake, a late invalid intention with no partial advance, and both
founders reaching terminal death while retaining complete lineage, inventory,
life/source provenance and final growth state in stable-order corpses.
Executed locally: direct warning-as-error MSVC 19.44 Debug and Release/NDEBUG
compile/run; Debug CTest `cellular-lives-owner` (1/1 passed); CMake configure and
generation registered the target; and `git diff --check`. The CMake/Ninja build
idled without starting a compiler, so the focused target was compiled directly.
`clang-format` and `clang-tidy` are unavailable on this Windows PATH. Linux
sanitizer/CI, profile-4 codec/restore, durable publication and session wiring
remain pending.
## 2026-10-10 — M11.3d-1 cellular snapshot and restore (CI pending)
Added the bounded profile-4 owner snapshot/restore codec and its wire schema.
Restore verifies checkpoint identity and tick, outer and nested checksums,
canonical bytes, archive/live/dead closure, genome-derived phenotype/body,
habitat bounds, corpse retention and region-wide channel conservation. Tests
round-trip a populated state, continue original and restored owners tick for
tick through starvation, then restore the terminal corpse/heat/lineage state;
they also refuse identity/tick mismatch, checksum damage, unsupported versions
and an out-of-bounds position with a valid refreshed checksum.
Executed locally: warning-as-error MSVC 19.44 Debug and Release/NDEBUG direct
compile/run; Debug CTest `cellular-lives-owner` (1/1 passed); and
`git diff --check`. `clang-format` and `clang-tidy` are unavailable on the
Windows PATH. PostgreSQL profile admission, publication/recovery tests, Linux
sanitizers and CI remain pending; the codec is not yet durably playable.
## 2026-10-10 — M11.3d-2 portable profile-4 manifest (CI pending)
Manifest v2 now recognizes cellular-lives-v1 as profile 4, applies its 16 MiB
payload cap, pins cellular rules/envelope/habitat/coordinate/genome/development/
founder/archive versions in dedicated in-memory fields, and restores payloads
through the profile-4 owner validator. Profile-4 clone genesis is refused;
profile-3 behavior and bytes remain unchanged. The focused test covers profile-4
genesis encode/decode, payload restore, version refusal, identity/length
rejection, clone refusal and a progressing successor manifest.
Executed locally: warning-as-error MSVC 19.44 Debug and Release/NDEBUG compile
and run, Debug CTest `region-checkpoint-contracts` (1/1 passed), and
`git diff --check`. CMake/Ninja again idled without starting compilation, so the
focused target was compiled and linked directly with the available build
artifacts. `clang-format` and `clang-tidy` are not available on PATH. PostgreSQL
admission, durable publication/recovery, sanitizer and CI checks remain pending.
## 2026-10-10 — M11.3d-3 PostgreSQL profile-4 parser (CI pending)
Added append-only schema migration 18. The database parser now recognizes the
profile-4 version bundle and 16 MiB cap, validates the fixed 50 ms quantum,
preserves profile-3 rules, and refuses profile-4 clone genesis. A database
trigger refuses progressing profile-4 commits until the dedicated cellular
publisher can atomically record all tick evidence. PostgreSQL integration tests
check acceptance of the canonical bundle and rejection of an altered archive
version. Migration history/apply paths, checksum plumbing and CI expectations
now include version 18.
Executed locally: `git diff --check`. The local Bash/WSL launcher was denied by
the environment, MSVC cannot compile the POSIX PostgreSQL integration test
because `poll.h` is unavailable, and `psql` is not installed. The migration and
parser tests were therefore not run locally; per the agreed workflow they must
run in CI. Durable cellular publication remains blocked intentionally.
## 2026-10-10 — M11.3d-4 durable cellular input contract (documentation)
While tracing the profile-4 host path, found that v6's per-session sequence and
provisional admission receipt cannot identify an exact retry after reconnect.
The existing durable action ledger needs stable account/action identity. ADR-0111
records the resolution: preserve v6 unchanged and use a reliable v7 cellular
action whose canonical input and final result commit with the cellular
checkpoint. No implementation or tests were run for this documentation-only
contract; durable host coding follows it.
## 2026-10-10 — M11.3d-5a reliable v7 cellular action codec (CI pending)
Added the v7 reliable-stream `CellularAction` wire value with stable nonzero
account/action identity, intended tick, signed movement fields and optional
patch target. Admission owns movement bounds. Added `ActionResult::committed`
for cellular actions and kept it
v7-only; v6 cellular intentions and their bytes remain unchanged. The schema
defines the exact 48-byte layout and identifies server-side checks still owned
by admission and durable publication.
Executed locally: warning-as-error MSVC 19.44 direct compile and run of
`tests/cellular_intent_protocol_tests.cpp` with the protocol codec (v6 and v7
round trips, version gates, zero identity, reserved bytes and absent-target
canonicality passed), plus `git diff --check`. The CMake/Ninja target regenerated
successfully but idled without starting compilation; direct compilation was
used. `clang-format` and `clang-tidy` are unavailable on this Windows PATH. CI,
server admission, action ledger input bytes, durable publication and restart
recovery remain pending.
## 2026-10-10 — M11.3d-5b cellular action ledger contract (documentation)
The server-routing audit confirmed that migration-11 action rows keep only the
stable id and committed tick/sequence, so they cannot compare retried cellular
input after reconnect or process restart. ADR-0112 specifies the canonical
24-byte biological input, immutable per-commit action block, per-action final
result, exact-byte duplicate rule and isolated profile-4 publisher. V7 login
and action admission remain disabled until these records and the recovery path
are implemented. This contract-only segment was checked with `git diff --check`;
the migration and PostgreSQL process checks are pending.
## 2026-10-10 — M11.3d-5c canonical cellular input bytes (CI pending)
Added the exact 24-byte biological input codec from ADR-0112. It deliberately
omits session and epoch so identical retries across connections have identical
stored bytes. The focused test verifies round-trip and cross-session canonical
equality and rejects zero tick, invalid target flags, nonzero reserved bytes
and absent-target coordinates.
Executed locally: warning-as-error MSVC 19.44 Debug and Release/NDEBUG direct
compile/run of the cellular protocol test and codec, plus `git diff --check`.
The CMake/Ninja target remains stalled before compilation. `clang-format` and
`clang-tidy` are unavailable on this Windows PATH. Durable ledger lookup,
PostgreSQL publication, crash resolution and CI remain pending.
## 2026-10-10 — M11.3d-5d input-aware action ledger API (CI pending)
Extended `ActionLedger` with single and batch durable-record queries that carry
an optional canonical cellular input beside the final outcome. Its default
adapter preserves existing implementations but returns no cellular input for
legacy rows; that cannot prove an exact v7 retry, so callers must refuse such
reuse. The network regression now asserts this fail-closed compatibility view.
Executed locally: warning-as-error MSVC 19.44 compile of the updated admission
implementation and network regression translation units, plus `git diff
--check`. The combined network executable was not rebuilt or run locally; CI
must run the network suite. The PostgreSQL implementation, exact-input matching
and cellular publication remain pending.
## 2026-10-10 — M11.3d-5e durable action record reads (CI pending)
Migration 19 adds the optional canonical input and result to the existing
account/action ledger, constrains cellular inputs to the 24-byte canonical form,
and adds a lease-fenced record lookup. The migration chain, checksum plumbing,
runtime function audit and upgrade history now include v19. C++ persistence
parses and validates legacy and cellular record shapes, and PostgresActionLedger
has a batched input-aware query; migration integration asserts that old probe
rows remain result 1 with no cellular bytes.
Executed locally: warning-as-error MSVC 19.44 compile of the persistence
adapter, PostgresActionLedger adapter and region-action regression translation
unit, plus `git diff --check`. PostgreSQL is unavailable on this Windows
environment, and the POSIX migration integration target cannot compile with
MSVC because `poll.h` is missing; per the agreed workflow CI must run the
migration and store tests. No publisher writes cellular rows yet.
## 2026-10-10 — M11.3d-5f cellular action admission scheduling (CI pending)
Extended `IntentAdmission` with v7 cellular action overloads. Scheduled
`CarriedAction` values retain the canonical bytes, exact retries resolve to the
same pending/final action, and reusing an id with changed bytes returns
`invalid_action`. Cellular commit outcomes use the distinct v7 committed result.
RegionSessions remains opt-in-off until the publisher/recovery gate is complete.
Executed locally: warning-as-error MSVC 19.44 direct Debug and Release/NDEBUG
compile/run of `cellular_intent_admission_tests` (authority, scheduling,
canonical bytes, pending/final exact retries and changed-input refusal passed),
plus `git diff --check`. The full network suite and CI are pending.
## 2026-10-10 — M11.3d-6a profile-4 publication SQL (CI pending)
Migration 20 adds a dedicated profile-4 publisher and ambiguous-commit resolver.
It validates canonical 48-byte cellular action records, retains the complete
action block as immutable commit identity, writes per-action canonical input and
the committed-cellular result in the same transaction as checkpoint publication,
and replaces the temporary profile-4 refusal gate with a private-evidence check.
The former checkpoint transaction is now a private helper behind the unchanged
profile-1/2/3 wrapper and the new cellular wrapper. The runtime grant audit
includes both public publisher entry points while excluding the private helper,
evidence table, and internal parser.
Added migration coverage for the canonical action-block parser, function grants,
and lack of runtime insert or private-helper access. `git diff --check` passes.
PostgreSQL migration and integration tests were not run locally because `psql` is
not installed in this Windows environment; Linux/PostgreSQL CI qualification
remains pending. The C++ writer, replay verification, crash/restart process tests,
and RegionSessions v7 enablement remain open.
## 2026-10-10 — M11.3d-6b cellular PostgreSQL adapter (CI pending)
Added `publish_cellular_checkpoint` and `resolve_cellular_region_commit` to the
region persistence interface. They call only the profile-4 SQL wrappers, preserve
their exact manifest/decision/action bytes, classify SQL failures through the
existing lease-store contract, and parse the normal durable receipt or resolution
status. The PostgreSQL adapter integration test now verifies malformed calls are
rejected as invalid requests.
`git diff --check` passes. The adapter and PostgreSQL integration tests have not
run locally; this Windows environment lacks the PostgreSQL build/runtime needed
for `EVO_BUILD_POSTGRES`, so CI is required. No cellular host calls these methods
yet.
## 2026-10-10 — M11.3d-6c cellular replay evidence (CI pending)
Added ADR-0113 and a portable profile-4 replay trace. It stores the stable life ID
and biological intention for every life at every tick, with canonical ordering,
strict bounds, and pre-allocation size checks. Replay restores an immutable
predecessor clone, applies each tick through `CellularLives`, and returns the
candidate for full canonical snapshot comparison. Migration 21 permits its
bounded variable trace size for profile 4 only and adds a lease-fenced read of the
immutable action block; the persistence adapter exposes that read.
Cellular lifecycle tests now cover encode/decode, deterministic replay, malformed
truncation/noncanonical bytes, total conservation, and wrong-region refusal. The
PostgreSQL store test covers lease-bound absence for an unknown commit. `git diff
--check` passes. C++ compilation, PostgreSQL tests, and Linux sanitizers are
pending CI because this Windows environment has no compiler or local PostgreSQL
runtime on `PATH`.
## 2026-10-10 — M11.3d-6d cellular action/trace binding (CI pending)
Added a portable canonical 48-byte action-block encoder/decoder matching the
PostgreSQL record layout. It enforces account/action identity, input shape,
tick-range membership, canonical order, uniqueness, and the 1,024-record cap.
The trace matcher requires each durable action to resolve to a living account in
the predecessor and to match the exact movement and optional patch target for its
tick. Lifecycle tests cover exact match and changed-input refusal.
`git diff --check` passes. C++ compilation and Linux sanitizer/CI tests are still
pending; this Windows environment has no C++ compiler in `PATH`.
## 2026-10-10 — M11.3d-6e cellular checkpoint publication adapter (CI pending)
Added profile-4 batch publication with verified predecessor replay, exact state
comparison, action-to-trace binding, verified immutable object uploads, and the
isolated SQL publisher. Prepared retry and ambiguous resolution retain and
validate the exact manifest, trace, and action block. The manifest codec now
enforces the variable trace-size range from ADR-0113 and migration 21; focused
tests cover its lower and upper rejection boundaries.
`git diff --check` and a GCC 13.2 C++23 `-fsyntax-only` pass for the new adapter,
manifest validator, and boundary tests pass. Full CMake build/CTest and PostgreSQL
integration are pending CI. The documented debug preset currently selects the
cached MSVC 19.44 compiler and fails CMake's C++23 feature check; this environment
also has no PostgreSQL runtime on `PATH`.
## 2026-10-10 — M11.3d-6f cellular action result recovery (CI pending)
Added lease-fenced cellular checkpoint recovery. It verifies the head payload,
trace/header and predecessor digest, replays the complete batch, and checks each
action row against the action block's account/action identity, cellular kind,
committed result, intended tick, head sequence and exact canonical input. The
PostgreSQL integration suite now exercises profile-4 genesis with a funded
founder, a two-tick batch carrying one canonical neutral action, atomic
publication, and recovery including its durable result row.
`git diff --check` and GCC 13.2 C++23 syntax checks with warnings as errors pass
for the changed adapter and integration test. Full build, PostgreSQL execution,
and sanitizer checks are pending CI; the local CMake preset remains unable to
configure with its cached MSVC compiler selection.
## 2026-10-10 — M11.3d-7a cellular region host (CI pending)
Added a profile-4 single-writer host that obtains and renews the region lease,
fully recovers before readiness, replays each supplied batch, and advances its
in-memory state only after a confirmed checkpoint receipt. An uncertain response
pauses admission and resolves or retries the exact prepared manifest, trace and
action block. Final action acknowledgements are exposed only after a confirmed
receipt. The PostgreSQL integration test exercises publication through the host,
its action result, and recovery through a fresh host instance.
`git diff --check` and GCC 13.2 C++23 syntax checks with warnings as errors pass
for the host and integration test. Full CMake build, PostgreSQL execution and
process fault tests remain pending CI; CMake's documented preset still fails its
cached MSVC C++23 feature check.
## 2026-10-10 — M11.3d-7b reliable cellular session actions (CI pending)
Added the v7 cellular action send API to the client session and opt-in server
admission. The server requires durable control and action ledgers, refuses v7
unless enabled, and keeps v7 away from the M10 view/intention datagrams. QUIC
session coverage checks admitted/carried canonical input, confirmed result,
exact retry after reconnect, and refusal when the same durable action ID is
reused with changed input. These changes do not yet connect session scheduling
to `CellularRegionHost` publication.
GCC 13.2 C++23 `-fsyntax-only` checks with warnings as errors pass for the
modified server/client sources and `tests/quic_sessions.cpp`, with and without
`NDEBUG`; `git diff --check` passes. `cmake --preset quic` cannot configure in
this checkout because its cached MSVC 19.44 compiler is reported as having no
known C++23 features. `clang-format` is unavailable on `PATH`. The QUIC test
binary, PostgreSQL suite, and CI have not run here.
## 2026-10-10 — M11.3d-7c action-to-life trace adapter (CI pending)
Added a QUIC-build-only server adapter that overlays carried v7 actions on the
scheduler's complete per-life tick inputs. It verifies tick and account
ownership, rejects duplicate actions for one life in a tick, and emits the
canonical action-block records while leaving other life behavior to the
scheduler. Focused tests cover the overlay, preservation of supplied inputs,
replay-trace matching and malformed/unowned/duplicate refusal.
GCC 13.2 C++23 strict-warning syntax checks pass for the adapter and new test in
debug and `NDEBUG`; `git diff --check` passes. The executable test, full QUIC
preset and CI have not run. The local QUIC preset remains blocked by cached MSVC
C++23 feature detection, as recorded above.
## 2026-10-10 — M11.3d-7d bounded cellular publication batches (CI pending)
Extended the action-to-life adapter with a batch builder. It clones only the
committed cellular checkpoint, binds the first trace header to the exact head
commit/digest, accumulates up to 64 ticks and 1,024 unique account/action
records, and advances a private working copy through `CellularLives`. Focused
coverage compares the built candidate against independent trace replay and
refuses stale input and action-ID reuse across ticks.
GCC 13.2 strict-warning syntax checks pass for the updated adapter and test in
debug and `NDEBUG`; `git diff --check` passes. Runtime test execution, the QUIC
preset and CI remain pending; the local cached-MSVC CMake limitation is
unchanged.
## 2026-10-10 — M11.3b-6 authoritative cellular patch perception (CI pending)
Added a life-scoped `CellularLives::perceive` API. It recomputes the cell's
authoritative phenotype footprint and returns only the exact in-range patch
address and stock, excluding provenance. The v7 action adapter now rejects
intake targets outside this view, including guessed distant patches.
GCC 13.2 strict-warning syntax checks pass for the lifecycle, session adapter
and focused tests in debug and `NDEBUG`; `git diff --check` passes. Runtime test
execution and full CI are pending; local CMake remains blocked by its cached
MSVC C++23 feature detection.
## 2026-10-10 — M11.3d-7f empty cellular genesis (CI pending)
Added an owner-only profile-4 genesis API and `evo-world-clone` configuration
for an explicitly supplied empty cellular opening. No life, identity or
founder is invented by provisioning. PostgreSQL integration coverage verifies
sequence-0/tick-0 publication, restored habitat/nursery closure, exact retry
and refusal of conflicting genesis.
GCC 13.2 strict-warning syntax checks pass for
`src/persistence/cellular_checkpoint_store.cpp` and
`tests/cellular_publication_tests.cpp`; `git diff --check` passes. The
Linux-only `evo-world-clone` translation unit cannot compile in this Windows
environment because `sys/random.h` is unavailable. The local CMake debug build
of `evo-cellular-lives-tests` remains running without output; it is not counted
as a pass. No PostgreSQL integration execution, CTest, clang-format or CI result
is claimed. The Linux/PostgreSQL qualification remains pending CI.
## 2026-10-11 — M11.3d-7g detached cellular birth candidate
Added `CellularLives::birth_candidate`, which derives primitive-founder
biology from a reserved identity, chooses the first unoccupied provisioned site,
debits body/reserve/construction funding from the nursery, records the founder
in a copied archive, and returns a validated detached candidate. The committed
owner is unchanged on both success and refusal. The focused lifecycle test
covers a funded birth, conservation, restore, duplicate account, invalid
identity/tick, insufficient funding, occupied-site refusal, and first action
on the following tick. `cellular_batch.cpp` also avoids a GCC 13
`maybe-uninitialized` false positive for an optional intake target under
optimized `NDEBUG` builds.
The focused test executable was manually compiled and run with GCC 13.2 in
debug (`-O0 -g`) and release (`-O2 -DNDEBUG`) modes; both print PASS. The
pre-existing missing-field initializer warning in
`cellular_region_step.cpp:29` is demoted for these builds; other warnings remain
errors. `git diff --check` passes. CMake/CTest, sanitizers, clang-format,
PostgreSQL integration and CI have not run for this segment; the local CMake
jobs remain stalled without output and do not count as validation. The current
Windows environment also lacks `clang-format`. Durable birth trace and server
integration remain unimplemented, so this does not make the cellular stage
playable.
## 2026-10-11 — M11.3d-7h cellular birth replay trace (database pending)
Profile-4 cellular trace format 2 now records a bounded birth set after each
tick's existing-life intentions. Recovery deterministically applies births at
the end tick, and the decoder retains format-1 support for old committed traces.
The C++ manifest bound was expanded; migration 22 updates the PostgreSQL
profile-4 trace ceiling and the migration runner/CI history checks.
The focused lifecycle test passes under GCC 13.2 in debug and `-DNDEBUG` builds,
including trace encode/decode, version-1 decoding, funded end-of-tick birth
replay, and malformed generation rejection. Strict-warning syntax compilation
of `src/persistence/region_checkpoint.cpp` and `git diff --check` pass. This
Windows environment has no `psql`, so migration execution and PostgreSQL
integration are pending CI. CTest, sanitizer checks, clang-format and CI have
not run for this segment. The Windows `bash.exe` shim is access-denied, and the
migration test translation unit cannot compile here because its process fixture
requires POSIX `poll.h`. The previously noted unrelated initializer warning in
`cellular_region_step.cpp:29` is demoted only for the manual focused test build.
## 2026-10-11 — M11.3d-7i cellular birth publication (database pending)
Migration 23 parses the format-2 birth records, checks their trace/manifest
binding, and routes publication through current account control and the exact
allocator reservation. It records the account-to-life binding in the same
transaction as the cellular checkpoint. The cellular publisher's runtime API
now requires the trace bytes, and recovery compares durable birth bindings with
the decoded tail trace before accepting the replay. A PostgreSQL integration
case covers a first birth from empty genesis, the v5 request outcome, and a
new-host recovery.
Strict GCC 13.2 syntax checks pass for the persistence adapter, cellular host,
and affected unit/integration test sources; the focused codec/lifecycle test
passes in debug and `-DNDEBUG`, and `git diff --check` passes. PostgreSQL
migration and integration execution remain pending CI because this Windows
environment lacks `psql`. The Windows `bash.exe` shim is access-denied, so the
migration scripts could not be syntax-checked here. CTest, sanitizer checks,
and CI have not run for this segment. The migration-test source itself cannot
compile locally because its fixture requires POSIX `poll.h`.
## 2026-10-11 — M11.3d-7j queued birth batch building
`CellularSessionBatchBuilder` accepts reserved birth records for a scheduled
tick. It canonicalizes their order, advances existing lives first, and applies
each birth only to a detached candidate. Failed funding, placement, or duplicate
identity validation does not partially advance the builder. The focused session
batch test covers successful empty-region birth replay, duplicate birth refusal,
and unchanged state after refusal.
The focused session-batch executable was manually compiled and run with GCC
13.2 in debug and `-O2 -DNDEBUG`; both pass. `git diff --check` passes. CTest,
sanitizers and CI are pending. The existing unrelated initializer warning in
`cellular_region_step.cpp:29` is demoted for these manual builds. This builder
is not yet wired into an `evo-server` cellular tick loop.
## 2026-10-11 — M11.3d-7k v5/v7 session admission
The region session host can now admit v5 life requests and v7 cellular actions
in the same configuration. A v5 connection remains v5; a v7 connection carries
reliable cellular actions. The QUIC session test covers both clients logged in
at one host and confirms a spawn response plus a carried cellular action.
GCC 13.2 strict-warning syntax checks pass for `region_sessions.cpp` and
`tests/quic_sessions.cpp`; `git diff --check` passes. QUIC runtime tests, the
full network preset, and CI remain pending. No wire layout changed in this
subsection.
## 2026-10-11 — M11.3d-7l host birth request queue
`CellularRegionHost` now exposes v5 create/query callbacks backed by the durable life reservation
store. Exact retries reuse and enqueue the same reservation, the scheduler can read the queued
births, and confirmed publication (including resolution of an uncertain commit) removes only the
matching queued records. The PostgreSQL integration test now covers reserve, retry deduplication,
publication, and born outcome through the host API.
GCC 13.2 strict-warning syntax checks pass for the host and PostgreSQL integration test, and
`git diff --check` passes. PostgreSQL runtime, full presets, and CI remain pending; local `psql`
and the POSIX-only migration test environment are unavailable.
## 2026-10-11 — M11.3d-7m v7 cellular snapshots
Reliable v7 cellular-action sessions now use the existing interest and snapshot
channel; v6 remains excluded until its separate perception contract is ready.
The QUIC regression test confirms a v7 client receives a cell snapshot and can
change interest. GCC 13.2 strict-warning syntax checks pass for the server
session, client session and QUIC test, and `git diff --check` passes. QUIC
runtime tests and CI remain pending. The configured Windows CMake QUIC target
produced no output for 25 seconds and was interrupted; it is not counted as a
build or test result.
## 2026-10-11 — M11.3d-7n cellular server process
The app now has an opt-in `--profile cellular` serve path with the M10 session
window, durable control/action ledgers, cellular ticks and cell snapshots. It
only advances while a living account has an established session or a first
birth is queued. The new bridge header passes GCC 13.2 strict-warning syntax
checking; the `CellularLives` accessor and tests pass strict-warning syntax
checking; `git diff --check` passes. The POSIX-only `evo-server/main.cpp` cannot
be compiled in this Windows environment (`poll.h` is unavailable). The
configured Windows CMake QUIC target previously stalled without output and was
interrupted. Linux/PostgreSQL CI and a real region run remain pending.
## 2026-10-11 — M11.3e v8 cellular playtest client
The client adds a cellular-trial connect mode, stable first-life request ID,
one reliable movement/feed action per server tick, owner perception display,
patch selection/continuous feeding, visible patch markers, and action/birth
status. Focused strict-warning GCC 13.2 syntax checks pass for the client app
with and without `EVO_CLIENT_NETWORK`, `panels.cpp`, and
`tests/client_remote_session.cpp`. The v8 protocol unit executable passes.
`git diff --check` passes.
The configured `client-network` CMake build emitted only its initial glob-check
line and was interrupted after roughly 50 seconds with no compiler or target
result. It is not counted as a build or test result. QUIC runtime, PostgreSQL
CI, and the real-region player probe remain pending. `clang-format` is not
installed in this Windows environment, so formatting validation is also
pending in CI.
The M11.3 cellular probe checklist was added for the first human session. It
does not count as live-region or usability evidence; those results must be
recorded after the operator confirms CI and runs the dev-region probe.
## 2026-10-11 — M11.3e deployment readiness report
The maintainer reports that the Git checks passed, the development `evo-region`
is up, and the client can connect. This is recorded as operator confirmation;
the CI run and checked revision were not independently available through the
current tools. A connection alone does not confirm the v8 cellular mode,
allocator-backed first life, movement, feeding, owner perception, or the human
playtest observations.
The local `build/client-network/evo-client.exe` predates the v8 client changes.
The Windows `client-network` build still stalls at Ninja's glob recheck. Running
the generated `VerifyGlobs.cmake` directly under both installed CMake versions
completed with no source-list mismatch, but did not resolve the Ninja hang. No
fresh client binary or local live cellular probe is claimed. The next gate is
the maintainer's short session using [the probe checklist](../playtests/M11_3_CELLULAR_PROBE.md);
11.4 remains gated on its feedback.
## 2026-10-11 — M11.2d-7a durable life-owner lookup (PostgreSQL pending)
Migration 24 adds a `SECURITY DEFINER` owner query that verifies the caller's
current region lease, returns only the unrejected world/account reservation's
region, and grants runtime only function execution. The C++ adapter validates
missing, malformed and multiple-row results. PostgreSQL integration coverage
checks absent accounts, cross-region lookup, a pending reservation, cancellation
after takeover, invalid account IDs, and migration/runtime privileges.
GCC 13.2 strict-warning syntax compilation passed for
`src/persistence/region_lease_store.cpp` and
`tests/life_reservation_store_tests.cpp`; `git diff --check` passed. The
`tests/region_migration_tests.cpp` translation unit cannot compile locally
because the Windows environment lacks POSIX `poll.h`. PostgreSQL/psql, CTest,
and CI have not run for this segment. The CI migration, privilege, and
PostgreSQL integration results remain required before closing the database
gate.
## 2026-10-11 — M11.2d-7b owner routing (CI and runtime pending)
The QUIC session path now resolves the durable life owner after grant
verification/scope checks and before calling the durable control claim. A
different owner is looked up only in the operator-provided `--life-route`
catalog; the listener sends the existing `SessionRedirect` for v4 and later,
then `Close(moved)`. An incomplete catalog or database lookup failure does not
claim control. Cellular server startup requires at least its own route entry.
GCC 13.2 strict-warning syntax compilation passed for
`src/server/session/region_sessions.cpp`,
`src/server/session/postgres_control_ledger.cpp`, and
`tests/quic_sessions.cpp`. A translation unit including the cellular bridge
also compiled, and the Visual Studio LLVM `clang-format` check and
`git diff --check` passed. The full `evo-server` source could not be compiled
locally because this Windows environment lacks POSIX `poll.h`; the QUIC test
binary, PostgreSQL integration suite, and CI have not run. The database owner
lookup and route behavior still require CI/runtime qualification.
## 2026-10-11 — M11.2d-7c two-region restart qualification (pending CI)
Extended `tests/region_m10_session_tests.cpp` to use the production primitive
session bridge and PostgreSQL. The scenario creates a life in region 35, tries
login in region 36 with both routes configured, checks `SessionRedirect` and
that `account_controls.region_id` stays 35, SIGKILLs and restarts region 35,
queries the same durable life, and checks the redirect once more.
The process-test source cannot be syntax-compiled locally because the Windows
toolchain lacks POSIX `poll.h`, which its existing process fixture requires.
The app bridge header compile, formatting check, and `git diff --check` passed;
the PostgreSQL process test has not run and requires CI confirmation.
## 2026-10-11 — M11.2d-7d client follows life-owner redirect (pending CI)
Extended `tests/client_remote_session.cpp` and its two-region fake world so the
second login is redirected by the new durable-life-owner hook before region 1
claims control. The remote client follows region 2, selects its grant from the
bundle, and establishes there. Claim counters assert that region 1 did not take
control. GCC 13.2 strict-warning syntax compilation and Visual Studio LLVM
`clang-format` passed; the test binary has not run. CI execution remains
required.
## 2026-10-11 — M11.3e-6a cellular diagnostics protocol v9 (CI pending)
ADR-0117 and `schemas/network/session-v9-cellular-diagnostics.md` define an
additive owner-only report of current reserve and latest committed tick effects.
The v8 message stays byte-identical. The codec rejects malformed enums,
reserved bytes, impossible status combinations and missing live-life diagnostics.
Visual Studio LLVM `clang-format` and `git diff --check` passed. MSVC 19.44
compiled the modified codec and protocol test with `/W4 /WX`, `/DNDEBUG`; the
standalone protocol test executable passed its v6, v7, v8 and v9 checks. The
configured CMake/Ninja build still stalls before compilation at its glob
recheck. The CI network suite remains pending; no server/client integration or
live cellular playtest is claimed for this contract-only slice.
## 2026-10-11 — M11.3e-5 readable cellular action feedback (CI pending)
The cellular client now explains each protocol admission status and durable
action result in plain language. A committed action tells the player to inspect
authoritative position and patch-stock changes. This clarifies session and
scheduling failures; it does not add the missing reserve or per-tick
maintenance/movement-cost display.
The modified `tests/client_remote_session.cpp` translation unit compiled with
MSVC `/W4 /WX`, the standalone status/outcome checks passed, Visual Studio's
LLVM `clang-format` check passed, and `git diff --check` passed. The normal
`client-network` Ninja build still stalls at the CMake glob recheck. A manual
test link against the existing build libraries was not valid evidence: those
libraries predate v8, first missing the new client methods and then failing
during old server-session setup. The full remote-session test, fresh client
link, and CI run remain pending; no cellular runtime or playtest success is
claimed.
## 2026-10-11 — M11.3e-6b/c v9 diagnostics delivery (CI green)
The committed cellular tick now exposes transient per-life feedback, the
cellular server bridge sends it only to the authenticated life owner over v9,
and the client displays reserve inventory, the last tick's maintenance and
movement costs, movement outcome and feeding outcome. v8 owner sessions still
receive the byte-compatible v8 perception; restart clears transient tick
effects. The QUIC test covers v9 owner filtering and v8 compatibility.
MSVC 19.44 compiled the changed biology, lifecycle, client-session and
server-session translation units, the cellular step/lifecycle tests, the QUIC
session test, the client panel and remote-session test with `/W4 /WX /DNDEBUG`.
Visual Studio LLVM `clang-format --dry-run --Werror` and `git diff --check`
passed. Freshly compiled source objects were manually linked with the existing
workspace libraries: the cellular step test passed, the cellular lives test
passed, and the QUIC session suite passed including v9 owner filtering and v8
compatibility. A freshly linked v9 client executable also passed its `--version`
startup check.
Direct CMake regeneration succeeded. The generated Ninja build still remained
at `Re-checking globbed directories...` before invoking target compilation, so
the CMake target/preset build itself has not completed. The Gitea public commit
status endpoint reports `state=success`, 10/10 contexts for exact head
`537a7e99b40d625a9db859a73c060edd6e397644` (run 238): checks, client, deploy,
ECS, network, PostgreSQL, TSAN, postgres-tools, image and server-image. The run
covers the pushed code; its automated results do not assess playability or fun.
The live-region human probe remains the only open 11.3 gate.
The local environment cannot query the operator's running region: `kubectl`
reports that no current context is set and its default `localhost:8080` API is
refused; no `evo-region` or `evo-server` process is running locally. The
maintainer reports that the region is up and the client connects. The detailed
11.3 cellular-loop probe notes have not yet been recorded. Mutation work has
started at the maintainer's direction while those notes are pending; eligibility
and balance remain explicitly open.
## 2026-10-11 — M11.4a funded mutation proposal
The new `fund_mutation` biology proposal revalidates the exact seeded offer,
derives the successor body at the same growth state, and returns channel-specific
reserve intake/release plus construction energy. It refuses underfunding without
returning a partial transition. The mutation contract tests passed after a fresh
MSVC 19.44 `/W4 /WX /DNDEBUG` compile of the modified mutation, genome,
development, metabolism and rates sources with `tests/mutation_tests.cpp`.
Visual Studio LLVM `clang-format --dry-run --Werror` and `git diff --check`
passed.
`cmake --build build/debug --target evo-mutation-tests` regenerated successfully
but stalled before target compilation, at the same Ninja glob recheck seen above.
The direct focused MSVC executable is local evidence only; full CI is pending.
The first CI run (240) also exposed a GCC `-Werror=missing-field-initializers`
diagnostic in the pre-existing `CellularRegionStepCandidate` aggregate setup.
The initializer now explicitly supplies empty starved and heat fields; CI for
that correction is part of the follow-up run below.
CI run 241 compiled the code but found `region-checkpoint-contracts` accepting
an impossible 119-byte one-tick cellular trace. Manifest shape validation now
checks both supported trace headers, rejecting that truncated shape while
retaining the 116-byte empty legacy trace. The focused region-checkpoint test
and full CI for this correction are pending.
The changed manifest implementation and checkpoint test translation unit both
compile under MSVC `/W4 /WX /DNDEBUG`. A standalone local link is unavailable:
the cached release libraries do not contain the latest cellular lifecycle
symbols and the cached debug libraries use a different CRT. The Linux CI run
provides the full linked test.
CI run 242 passed all 68 `checks` tests, including the new trace-shape regression,
but the final whole-tree format target found 25 recently edited cellular-stage
files that were not formatted to the CI-pinned clang-format 19.1.5 style. They
are now formatted with that exact version; the repository-wide check passes
locally and CI is pending on the formatting-only follow-up.
CI run 243 found `-Werror,-Wsign-conversion` in the existing cellular-motion
axis clamp, where mixed signed and unsigned conditional expressions selected
their common unsigned type. The clamp now applies signed deltas in explicit
branches and converts only their nonnegative magnitudes. Run 243 failed its
client build at this diagnostic before tests ran.
During run 244, clang-tidy 19 found an unchecked access to the nested optional
in `ActionLedger::find_record`. The function now moves the inner optional to a
local, checks its state, and unwraps only after that check. This is an existing
network-ledger warning exposed by the full network target; full CI is pending
for the combined repairs.
Run 245 confirms that optional-access fix removes its clang-tidy failure, then
reports two `-Wmissing-field-initializers` errors for legacy probe actions after
`CarriedAction` gained its optional cellular input. Those probe records now set
that input to `std::nullopt` explicitly; run 245 stopped before network tests.
The combined repairs are queued for a new full CI run.
Run 246 compiled the action-ledger changes and found that the legacy-action test
used `CHECK` macros to guard nested `optional::value()` calls. Clang-tidy cannot
prove that macro is a control-flow guard. The test now uses explicit throwing
precondition branches before unwrapping, preserving the same assertions. Run
246's network job stopped at those diagnostics before network tests ran.
Run 247 reached the PostgreSQL build and found two stale `evo-world-clone`
provision calls: a double dereference of the owned primitive-life state and an
array passed to an explicit retained-build span parameter. The primitive owner
is now dereferenced once. The first explicit span conversion did not satisfy
GCC 14's constructor constraints; the follow-up below uses the pointer/count
constructor. Run 247's PostgreSQL job stopped before its linked tests.
Run 248 reached the network build and found one more legacy `CarriedAction`
aggregate in the control tests with no explicit `cellular_input`. It now sets
`std::nullopt`, matching the existing probe-action contract. The network job
stopped before tests; the PostgreSQL job was canceled when run 248 was superseded.
Run 249 confirmed the owner dereference fix and rejected the const-array `span`
range constructor under GCC 14. The retained build digest is now provided with
an explicit data pointer and count. PostgreSQL again stopped at this compile
diagnostic before its tests.
Run 250 built the network source but exposed two test-gate issues: CTest had
registered the cellular protocol and admission suites while the network build
preset omitted their executable targets, and a version-rejection test still
treated protocol 7 as future after protocols 7-9 were added. The network,
QUIC, client-network and client-package build presets now build both
executables, and the negative test uses unsupported version 10.
Run 250's network test job failed on the stale expectation; the two omitted
tests were not run because their executables were absent. Follow-up CI is
required to confirm the corrected gate.
Run 252 exercised the repaired network preset, then found three compiler gates
outside the earlier test: clang-tidy reported an optional dereference in the
cellular-step test and a signed-width multiplication in the cellular-perception
protocol assertion; GCC 14 reported a missing diagnostics-field initializer in
the session server's cleared-perception record. The test accesses now use an
explicit optional value accessor after the checked guard, the size expression
widens before multiplication, and the cleared record initializes all fields.
Windows Debug owner/step tests, protocol test, session-server target build and
format check passed locally. Full CI follow-up is pending.
Run 261 was triggered for commit `15ad04f` after the bounded-perception fixtures
were formatted. At the last successful Actions query, client, deploy and network
jobs had passed, while checks, ECS and PostgreSQL were still running; TSan and
the dependent image jobs had not started. The Gitea host then returned HTTP 503
for the run page, workflow page and a Git remote dry-run. The final result could
not be verified in this environment. No full-CI pass is claimed for the 11.4b
migration/recovery work; rerun or inspect this gate when the host is available.
For the M11.4c archive-revision and v3 checkpoint-migration changes, clang-format
and `git diff --check` passed at that commit. The targeted Windows Debug build for
lineage, cellular lives and checkpoint tests again produced no output for 60
seconds and was stopped. Earlier direct compiler setup lacked the Visual C++
standard-library environment, and WSL is unavailable in this session.
The next M11.4c segment adds owner-side application and deterministic replay of
mutation selections, v3 trace decoding, a separate canonical mutation-action
codec, and SQL v3 trace bounds and birth binding. No local compile or PostgreSQL
test completed: both targeted Ninja invocations stalled without output and were
stopped. The C++ and SQL publishers deliberately refuse batches containing
mutation selections until durable action reservation, publication and recovery
are implemented. Formatting and Linux CI remain pending for this segment.
Run 265 compiled the M11.4c owner and trace code, then stopped in
`tests/lineage_tests.cpp`: a decoded archive was held through a const expected
before a negative death-mutation assertion. The fixture now keeps that decoded
archive mutable. Local clang-format and `git diff --check` passed; the targeted
Windows lineage-test build again produced no output for 60 seconds and was
stopped. Full CI follow-up is pending.
CI run 266 compiled the corrected lineage test, then failed
`region-checkpoint-contracts` at the assertion that rejected a v2-to-v3 cellular
profile migration. That upgrade is now supported for the M11.4c genome revision
history, so the test now verifies the resulting v3 versions and history profile.
The remaining CI jobs had not completed when this correction was prepared; local
Windows builds have repeatedly stalled during Ninja without producing output.
The correction needs a fresh CI run.
The M11.4c mutation-choice v10 wire-codec slice adds message type 25 with a
fixed 64-byte canonical payload and strict decoder checks; v1-v9 remain frozen.
It does not enable v10 in the server or client runtime. The focused codec test
and schema are included, but local C++ builds stall during Ninja and
`clang-format` is not installed in this Windows environment. Linux CI is the
required compile, format, and test gate for this slice.
CI run 268 compiled the v10 codec and passed all 68 CTests, including the
checkpoint migration and protocol tests. Its format-check then rejected the
new checkpoint test call layout at `tests/region_checkpoint_tests.cpp:478`;
that wrapping is corrected for the next run. The run is not a green gate.
The M11.4c schema-v29 reservation slice adds a world-allocated successor
identity reservation keyed by authenticated account and action ID, exact-retry
conflict handling, one pending choice per account, lease-fenced recovery reads,
and PostgreSQL integration coverage. It remains unverified until the pushed CI
run completes. The local PostgreSQL target reconfigured, then produced no build
output for 60 seconds and was stopped; `psql` is unavailable in this Windows
environment. `clang-format --dry-run --Werror` passed for the changed C++ files.
The user-directed validation path here is CI.
CI run 270 compiled the reservation slice and passed all 68 CTests, then failed
the format gate on the wrapped call in `tests/region_checkpoint_tests.cpp`.
That call now matches local clang-format output; the follow-up CI run is needed
to validate the PostgreSQL migration and remaining jobs.
CI run 271 passed the client, deploy, ECS and network jobs, then the PostgreSQL
image's clang-tidy analysis rejected four unchecked mutation-offer optional
accesses in `cellular_lives.cpp` and `cellular_batch.cpp`. CI run 272 showed
that the analyzer also rejects `.value()` after a preceding presence guard.
Those paths now copy through `value_or()` after the guard, preserving the
existing rejection behavior while avoiding unchecked optional access. Run 272
passed all 68 regular CTests but failed analysis before PostgreSQL
migration/tests. Its network job also failed because the protocol test marked
version 10 as future even though the v10 mutation codec now accepts it. That
test assertion remains unchanged in the pushed source. Runs 274 and 275 later
reported all 10 jobs green, including network and PostgreSQL; the reason run
272's network failure did not recur is unresolved and should be checked if the
network test fails again.
CI run 274 for commit `376b322` completed successfully across all 10 jobs:
the seven checks matrix jobs, `postgres-tools`, and both image builds. This
validates the pushed mutation-offer lint repair and schema-v29 reservation
slice on the configured CI path. The 11.4c mutation publication, atomic action
result binding and restart recovery are still incomplete; this green run does
not claim a live mutation path.
## 2026-10-11 — M11.3 human probe blocked by server image availability
The maintainer reports there have been no successful server image builds since
2026-10-10 and cannot update `evo-region` to the current cellular server. The
11.3 movement/feeding human probe therefore remains unrun; existing service
connectivity does not establish that the current cellular client/server pair
works. Resume the probe once a deployable server image is available. CI image-job
success is not deployment evidence and does not satisfy this gate.
## 2026-10-11 — Server image pipeline diagnosis
The latest relevant M11 code run (#272) failed in the checks matrix, so its
dependent `server-image` job was canceled. The next code run (#273) was canceled
by a subsequent push before its dependent image jobs started. Runs #274–#278
were green but their changes were documentation-only; `ci-paths.sh` therefore
skipped the server-image build. This explains why those green runs produced no
new image. Harbor publication has a separate gate: the main-branch job pushes
only when `HARBOR_SERVER_REPOSITORY` is configured. That repository variable is
not visible from this workspace, so its current value is unverified.
Run #280 (commit `bfa0e1d`) exercised the workflow after adding a Harbor-gate
diagnostic. Its network suite failed at `tests/network_control.cpp:375`: the
negative test still treated protocol version 10 as a future unsupported
version, although M11.4c defines the v10 mutation-action codec. The failed
checks matrix blocked `server-image` and `image`, so that run attempted neither
an image build nor Harbor publication. The future-version test now uses v11;
the M11.4c reservation-to-tick adapter is included in the next CI run so the
network, native, server-image and Harbor gates can all be evaluated together.
The local focused Windows build stalled without output and was stopped;
`git diff --check` passed. CI run #285's network suite passed, but the main
checks build failed clang-tidy on `tests/cellular_step_tests.cpp:53`:
`bugprone-unchecked-optional-access` rejected `intake.value()` despite the
preceding presence guard. The test now uses a guarded optional dereference.
The image selector also explicitly includes M10 session-control and
cellular-session-batch contract tests, so those changes select the
`server-image` smoke/publish job. The next main-branch run must confirm the full
checks matrix and Harbor publish gate before a playtest image is available.
CI run #286 built all regular targets and passed its 68 CTests, but the clang-tidy
stage rejected an unchecked mutation-offer candidate access in
`tests/cellular_lives_tests.cpp:316`. The guarded test reads now use safe defaults
after assertions, preserving the assertions' failure behavior. The server-image
selector includes this cellular lifecycle contract test, so the next push reruns
the complete validation and image publish path. The local Bash selector fixture
could not start: Windows returned `Bash/Service/CreateInstance/E_ACCESSDENIED`;
CI's deploy suite runs that fixture.
CI run #287 passed client, deploy, ECS and network, then clang-tidy found one more
unchecked nested optional read at `tests/cellular_lives_tests.cpp:289`. The earned
mutation offer is now copied through a safe default after the assertion still
checks the offer has candidates. Its check suite failed, so the dependent server
image did not start; rerun is needed to reach Harbor publication.
CI run #288 found another unchecked optional access in cellular checkpoint
recovery at `src/persistence/cellular_checkpoint_store.cpp:568` while building
the PostgreSQL-enabled analysis target. Recovery now copies the optional record,
checks its presence, and reads it through `value_or`, keeping a missing committed
action classified as a history failure. The PostgreSQL job failed, so the
dependent server-image build and Harbor publication did not start.
CI run #289 confirmed that record-access repair and found a second nested optional
access at `src/persistence/cellular_checkpoint_store.cpp:550`: the stored action
block inside the successful database result. Recovery now validates that the
block is present and decodes a `value_or` copy. The PostgreSQL job failed on this
diagnostic, so this run also did not start server-image or Harbor publication.
CI run #290 confirmed both action-record repairs and found a third nested optional
access in the cellular profile-migration recovery check at
`src/persistence/cellular_checkpoint_store.cpp:464`. Recovery now copies the
optional migration record through `value_or` and validates the copy after
preserving the missing-record history failure. PostgreSQL failed on this
diagnostic, so the image and Harbor jobs remained blocked.
CI run #291 confirmed the migration-recovery repair and found another unchecked
optional in the PostgreSQL reservation-row decoder at
`src/persistence/region_lease_store.cpp:1322`. The decoder now passes a safe
empty-string default to the hex parser; absent or malformed input still fails
the existing validation path. The PostgreSQL job failed on this diagnostic, so
the dependent server-image and Harbor jobs did not start.
CI run #292 confirmed the reservation-input decoder repair, then clang analysis
reported additional optional-value and ineffective-move warnings in
`src/persistence/region_lease_store.cpp` while building the PostgreSQL target.
The affected action-record, migration-record, commit-status, publication-action,
and life-owner parsers now use safe optional defaults or checked copies; missing
mutation input retains its prior absent-value behavior. The PostgreSQL job failed
on these analyzer diagnostics, so the dependent image jobs did not start.
Run #293's main checks job stopped at format-check before compilation because the
new optional-return expression wrapped at line 499. It has been kept on one
line to follow the 100-column clang-format configuration. PostgreSQL analysis
continued independently, but was superseded before completion by the format fix
push so the full matrix could rerun on one commit.
CI run #294 passed the main, client, deploy, ECS, network and TSan jobs, but the
PostgreSQL-enabled analysis reached `src/server/cellular_region_host.cpp` and
reported unchecked optionals at the cellular publish grant and life-query result,
plus an intentionally best-effort lease release expressed as a discarded value.
The image jobs were skipped because that matrix job failed. The host now passes
safe grant/outcome values after presence checks and captures the shutdown release
result using the existing RegionHost convention.
CI run #295 passed the main checks matrix, including the full cellular host build,
but PostgreSQL analysis next found checked-optional and ineffective-move warnings
in `src/server/session/postgres_action_ledger.cpp`. The single and batched
lookups now copy present rows via `value_or`, and the small durable result records
are copied without ineffective `std::move`. The PostgreSQL job failed on these
new diagnostics, so image publication remained skipped.
CI run #296 confirmed the action-ledger repair, then PostgreSQL analysis rejected
an unchecked `Options::primitive_frame` access in `apps/evo-world-clone/main.cpp`
while compiling the primitive-region provisioning path. The provision command now
uses a validated frame value for primitive and cellular openings. The default
Windows debug preset has PostgreSQL disabled and does not contain this target, so
PostgreSQL analysis and integration checks remain delegated to CI. The Visual
Studio LLVM `clang-format` executable is available locally.
CI run #297 stopped the main checks job at clang-format on the primitive
provisioning call after the PostgreSQL frame validation change. The call now
follows the repository formatter output; local `clang-format --dry-run --Werror`
and `git diff --check` pass. The PostgreSQL matrix job continued independently,
but its result cannot unblock image publication. Run #298 must rerun the full
matrix and server image/Harbor gates.
CI run #298 passed the main build/test job and network suite, then the
PostgreSQL-enabled Clang analysis rejected the provision CLI's assignments from
`*n` into optional fields at `apps/evo-world-clone/main.cpp:270-282` with
`bugprone-optional-value-conversion`. The validated parser value is itself an
optional, so these destinations now copy `n` directly. PostgreSQL analysis
failed before the dependent server-image build and Harbor publication could
start; the next pushed run must confirm the analyzer and image publish gates.
CI run #299 confirms the clone parser now builds in the PostgreSQL-enabled
release job, and the network suite passed. The client matrix job passed its
13 CTest cases under SDL's dummy driver and the standalone X11 client/UI probes,
but its additional X11 `evo-client-lifecycle-tests` invocation failed while
opening the final toolkit-capture application (`tests/client_lifecycle.cpp`).
That failure gates `server-image`; the test now reports its typed open error and
checks SDL teardown after each test section so the next CI run can identify the
failing lifecycle boundary. The PostgreSQL integration and static-analysis job
is still running; no Harbor publish has started.
CI run #300 did not rerun the client suite: the path selector omitted
`tests/client_lifecycle.cpp`, so it treated the diagnostic-only commit as
irrelevant. The client selector and its offline regression fixture now include
that test source. The next CI run must execute the diagnostics before its client
result can be used to unblock the server image.
CI run #301 passed the client lifecycle suite and deploy selector fixture, but
the PostgreSQL-enabled analysis rejected three unchecked cellular provisioning
optionals at `apps/evo-world-clone/main.cpp:509`. The execution path now uses
safe `value_or(0)` reads for those parsed values; valid `cellular-v1` options are
still required by the parser. The run failed before the dependent server-image
and Harbor publication jobs could start. CI must validate this repair on the
next push.
CI run #302 compiled that cellular change, then PostgreSQL clang-tidy found three unchecked primitive-provisioning optionals and a SQLSTATE `string_view` created without its fixed length in `apps/evo-world-clone/main.cpp`. The run failed before the dependent server-image and Harbor publication jobs could start. These four analyzer findings are now addressed; the next pushed run must validate them and publish the image.
CI run #303 stopped at the format gate on the SQLSTATE expression from the preceding fix, before reaching clang-tidy. The continuation is now laid out for clang-format; local clang-format is unavailable, so CI must verify formatting along with the analysis and Harbor publish gates.
CI run #304 rejected both wrapped layouts of the SQLSTATE expression at the formatting gate. The complete statement is now joined onto one line (within the configured 100-column limit); CI must confirm the exact formatter output.
CI run #305 passed formatting and most independent gates, but PostgreSQL clang-tidy rejected `site()` in `apps/evo-world-clone/main.cpp`: `string_view::data()` may be null for an empty view passed to `from_chars`. The parser now rejects an empty site argument before accessing its data. The dependent PostgreSQL tools and server-image/Harbor publication jobs did not run; CI must validate the guard and complete publication.
CI run #306 passed formatting, deploy selection, ECS, and TSan, and built the PostgreSQL test targets. Its client lifecycle probe failed at `native_input()` while opening the SDL application (line 174); the test reported only the source line and an empty SDL error, so the startup error was not identifiable. It has been updated to report the typed application error and SDL initialization state. PostgreSQL static analysis was still running when this follow-up was prepared; Harbor publication did not start.
CI run #307 passed the format/deploy/ECS/network/TSan gates. The instrumented `native_input()` open succeeded, but the next application open in `toolkit_capture()` failed (line 277 in that run), before PostgreSQL analysis completed; Harbor publication did not start. This second open now reports its typed error and SDL initialization state too, so the next run can distinguish an SDL window failure from toolkit initialization.
CI run #308 passed client, deploy, ECS, network and TSan, including the X11 lifecycle and AddressSanitizer client probes. PostgreSQL clang-tidy again rejected `site()` in `apps/evo-world-clone/main.cpp`, this time because `bugprone-suspicious-stringview-data-usage` treats `string_view::data()` on the split coordinate text as potentially unterminated even though `from_chars` receives a bounded range. The parser now passes the split views' begin/end iterators instead. The PostgreSQL tools and server-image/Harbor jobs did not start.
CI run #309 passed the client, deploy, ECS, network and TSan jobs, and the regular 68-test suite in the main checks job. PostgreSQL's full build then rejected 16 unchecked `std::optional` accesses in `tests/cellular_publication_tests.cpp` under warnings-as-errors. The affected test reads now use `optional::value()` after their existing checks, with committed snapshots stored once where repeated reads were previously unchecked. PostgreSQL tools and server-image/Harbor publication did not start.
CI run #310 passed client, deploy, ECS and network. The main checks stopped at the formatting gate because three new `CellularBatch::header` initializers were not laid out exactly as clang-format 19 expects. The test source is now formatted with clang-format 19.1.5 and passes its local dry-run check. PostgreSQL analysis was still building when this formatting correction was prepared, so run #310 does not yet validate the optional-access fixes; Harbor publication did not start.
CI run #311 passed client, deploy, ECS, network and TSan, and compiled the PostgreSQL test target. The PostgreSQL analysis build then still reported 17 unchecked accesses: clang-tidy does not treat the `CHECK(...)` macro or `optional::value()` by itself as proof of engagement. The test now uses a helper with an explicit `has_value()` guard before dereferencing those optionals. This version awaits CI validation; PostgreSQL tools and Harbor publication did not start.
CI run #312 passed client, deploy, ECS and network, but the PostgreSQL-enabled test build found that the new `checked_value` helper only accepted `std::optional`, while the SHA-256 helper returns `std::expected`. The test now provides the same explicit engagement check for `std::expected`; run #312 did not reach PostgreSQL tools or server-image/Harbor publication.
CI run #313 passed client, deploy, ECS, network and TSan; both PostgreSQL compilation configurations completed. The PostgreSQL integration run then failed its final runtime-role audit because the cellular profile-migration publisher and lookup are explicitly granted to the runtime role but were missing from the audit's fenced-function set. Those two lease-fenced functions are now listed, and the online migration test now expects the current schema version 29 after upgrading its v7 fixture. PostgreSQL tools and Harbor publication did not start.
CI run #314 passed client, deploy, ECS, network, TSan, PostgreSQL compilation,
and the runtime-role audit, then failed the first cellular birth publication
integration assertion. The shared life-binding table still had its original
foreign key to primitive-only publication evidence, so a profile-4 birth could
not reference its cellular publication. Schema migration 30 retargets that
foreign key to the common region commit and adds a deferred guard requiring
profile-specific publication evidence, matching region and a live reservation.
The migration test checks the deferred constraints; the cellular publication
integration exercises an actual birth. Run #314 did not start the PostgreSQL
tools or Harbor image jobs; CI validation of migration 30 is pending.
CI run #317 selected the server-image gate after `tests/region_migration_tests.cpp`
was added to the path selector. Client, deploy, ECS, network and TSan passed, but
the PostgreSQL matrix failed the cellular birth publication integration at its
post-publish assertion, so dependent image jobs remained blocked. The assertion
did not expose the sanitized database error category/SQLSTATE; the cellular host
publish result now carries those diagnostics to the test without server messages
or credentials. Validation of the diagnostic and migration 30 remains pending.
CI run #321 exposed the exact PostgreSQL error: PL/pgSQL's local record `b`
shadowed the `b` aliases in the duplicate-birth queries in migration 23. The
fix is a forward migration 31 so existing migration checksums remain intact.
Run #322 compiled the PostgreSQL targets and passed the earlier publication
test after migration 31, then exposed an obsolete test fixture that created a
non-empty profile-4 tick-0 genesis from a primitive life. ADR-0100 and ADR-0114
require empty genesis and allocator-backed births at a committed tick. The
fixture now creates empty legacy-v1 genesis, publishes its first life through
the normal control and birth path, then tests the v1-to-v2-to-v3 no-tick
migrations while carrying that living state. PostgreSQL integration cannot be
run in the local Windows environment, so this correction is awaiting CI.
CI run #323 passed the normal build and reached the PostgreSQL integration
suite, where the migration fixture collided with the primitive publication
fixture: both used world 9755 in the same disposable database, so the existing
region lease rejected the second provisioning call. The migration fixture now
uses its own world id (9759). PostgreSQL integration remains CI-only here; the
next push must validate this isolation fix and continue to the Harbor image job.
CI run #324 passed that region-lease setup, then found the fixture's genesis
commit ID also collided with an existing region-lease test: commit IDs are
unique across the database, and that fixture uses sixteen `0x50` bytes. The
cellular migration genesis ID now has a distinct first byte while preserving
the remaining test pattern. CI must rerun the PostgreSQL gate and downstream
Harbor publish before a server image is available.
CI run #325 reached the v2-to-v3 cellular profile migration and rejected the
normalized manifest as malformed. The v3 parser was replacing a four-byte
history-profile field with the eight-byte result of `manifest_u64le(3)`, which
increased the 360-byte manifest to 364 bytes. Migration 32 now intercepts v3
manifests, writes the four-byte little-endian value directly, and delegates to
the preserved v27 parser. The PostgreSQL migration integration test covers the
v1-to-v2-to-v3 path. CI validation and Harbor image publication are pending.
CI run #326 compiled the PostgreSQL targets but failed while applying migration
32: it tried to rename the current parser to `parse_checkpoint_manifest_v27`, a
name already created by migration 28. The pending migration now uses the unique
`parse_checkpoint_manifest_v31` alias. The PostgreSQL gate and Harbor image
publication are pending another CI run.
CI run #327 applied migration 32 and built the PostgreSQL test targets, then
failed in the cellular publication integration test while recovering the
migrated checkpoint. The fixture inspected `committed()` before calling `step()`
on the newly created host; host recovery and lease acquisition happen during
`step()`. The test now checks readiness first, matching the established restart
test. PostgreSQL CI and Harbor image publication must rerun before a playtest
image is available.
CI run #328 passed that recovery check, then reached a second obsolete fixture
that constructed a tick-0 cellular genesis from a primitive life born at tick 1.
`CellularLives::create` correctly rejects that impossible chronology. The test
now provisions an empty v1 genesis and publishes its first life through the
allocator-backed control and birth path, retaining the action publication and
restart assertions. PostgreSQL CI and Harbor image publication remain pending.
CI run #329 verified the test compiles but rejected replay because the newly
empty nursery had no energy reserve to fund its first birth. The fixture now
keeps the canonical primitive endowment while starting with no cells. PostgreSQL
CI and Harbor image publication must rerun.
CI run #330 passed all 68 regular tests, then failed the formatting gate on two
lines in `tests/cellular_publication_tests.cpp`. That gate failure prevented the
PostgreSQL, server-image, and Harbor publish jobs from running. The fixture lines
are now formatted to match clang-format; CI must rerun the PostgreSQL gate and
publish path before a playtest image is available.
CI run #331 again passed all 68 regular tests and accepted the primitive fixture
format, but clang-format rejected the line wrapping of its account-control call.
The call now follows the surrounding fixture's multiline argument layout. The
PostgreSQL integration and Harbor publication jobs remain gated on the next run.
CI run #332 again passed all 68 regular tests and accepted the fixture setup,
then rejected the account-control call's argument grouping. Its first line now
ends after the account ID, with generation and timeout arguments together on the
continuation. PostgreSQL and Harbor jobs remain blocked until this format gate
passes.
CI run #333 passed all 68 regular tests but requested the control call's full
argument list on one line; it fits the configured 100-column limit. The fixture
now uses that layout. PostgreSQL integration and image publication must still
complete successfully before the playtest image is ready.
CI run #334 passed the regular tests and compiled the PostgreSQL targets, then
failed in `cellular_publication_tests.cpp` because its fixture assigned the new
cell's first action to the same batch that created it. Durable action matching
is against the batch's predecessor lives, so the test now publishes the birth
first and the neutral action in the following one-tick batch. The run's
PostgreSQL gate failure blocked the server-image and Harbor jobs; this correction
must pass CI before the playtest image is available.
CI run #335 contains only the publication-fixture correction and this validation
record. The server-image path selector did not include
`tests/cellular_publication_tests.cpp`, so a green run would not rebuild the
server image from this commit. The selector now includes that integration test;
CI must confirm both the server-image build and Harbor publication. Harbor
publishing also requires the `HARBOR_SERVER_REPOSITORY` repository variable.
CI run #336 fixed the build selector and ran the PostgreSQL publication test,
which exposed an off-by-one in migration 20: `manifest_uint` reads from a
zero-based offset, but the publisher supplied 25 instead of 24 for each
canonical action tick. This made the durable tick disagree with the 24-byte
input, so PostgreSQL correctly rejected the action row. Migration 33 replaces
the publisher with the corrected offset while retaining migration 20's recorded
checksum. The migration runner's accepted latest history and upgrade fixtures
now extend through version 33. Local PostgreSQL is unavailable on this Windows
environment; migration, publication, and server-image checks are awaiting CI.
## 2026-10-13 — DEV cellular region and life-allocator provisioning (CI pending)
ADR-0123. Bringing up the M11.3 probe moved DEV to cellular region 100/2,
migrated the DEV schema from v11 to v33 (postgres-tools from run 350), and
showed that world 100 had no `life_allocators` row, so every first-life request
resolved `unknown`. `evo-world-clone provision` now accepts
`--life-allocator-start N` with the primitive-v1 and cellular-v1 openings. It
creates the allocator once, leaves it untouched on reruns, and refuses a start
above the existing counter. `region_store_tests` adds a PostgreSQL process case
for first create, idempotent rerun, lower and higher starts, and refusal with
the development opening.
Executed locally: g++-14 `-std=c++23 -fsyntax-only -Wall -Wextra -Wconversion
-Werror` on both changed files, and clang-format 19.1.5. The PostgreSQL process
test runs only in CI. The DEV overlay adopts the flag only after CI publishes
the server image that contains it.
The first sync with the promoted image showed the allocator created
(`life-allocator world=100 next=1000000000000`). It then refused region 2,
because the cellular genesis check re-encoded the manifest with the new
`--build`. Provisioning now matches an unadvanced cellular genesis under its
recorded build, and accepts profile and frame once the head is past genesis.
`region_store_tests` adds a cellular case: a rerun under a new build is
already-provisioned, while another commit or opening is refused. Checked
locally with g++-14 `-Werror`, clang-tidy-19 `--warnings-as-errors=*` and
clang-format 19.1.5; the PostgreSQL process tests run in CI.
The first playable run (client camera fixed in 532f45b) showed two server-side
faults. First, the client sends one cellular action per tick, but the default
probe admission allowed 4 per 8-tick window, so every other action was refused;
the cellular bridge now admits two per tick of window. Second, region 2's
opening starved the cell. Only patch energy feeds the reserve, and founder
intake (1,148,000 pJ/s) emptied its 4M and 16M pJ patches in seconds. DEV moves
to region 100/3 with 200M and 50M pJ patches (ADR-0123). Checked locally:
g++-14 `-Werror` and clang-tidy-19 on `apps/evo-server/main.cpp` with
`EVO_SERVER_M10`, plus the deploy policy suite.
Region 3 then livelocked. A second account's birth was queued, the account
disconnected, and every following batch failed `check_life_control` with
EVL02, which the host reads as a lost region lease. It fenced, reacquired and
failed again every 20 s, refusing logins with `shutdown`. A restart of
`Deployment/evo-region` (permitted by the agent contract) cleared the
in-memory queue, and the region held `ready 1` at epoch 8 from 20:20:44 UTC.
`serve_cellular` now includes, and ticks for, only births whose account holds
a session. Checked locally: g++-14 `-Werror`, clang-tidy-19 and clang-format.
Logins from a fresh client launch then failed in a loop: welcome, then the
client closed with `protocol`. A diagnostic in `RegionSessionClient` (now
kept: protocol closes name their cause on stderr) showed
`message decode detail=6 bytes=33`, a primitive-life outcome the codec
rejects. A new client launch sends a new birth request; for an account that
already has a cell, the cellular host reports `existing_life` with the
reserved identity, and `RegionSessions` forwarded it although the codec allows
an identity only with `born`. The session layer now clears identity and commit
on every other result. `quic_sessions` adds an `existing_life` refusal that
must arrive valid while the session stays established. Locally (MSVC, client
preset) the case failed with the same 33-byte decode error before the fix and
passed after it.
## 2026-10-13 — M11.3f-1 cellular engulfment and detritus resolver (CI pending)
ADR-0124. `propose_cellular_region_step` takes `CellularRegionRules`; rules 3
adds an engulf phase after every life's step (ascending life id, one prey per
predator, contact plus `predator_mass × 1000 ≥ 1500 × prey_membrane_um ×
prey_mass`), bounds the predator's gain by its remaining per-tick intake, and
deposits detritus and starvation corpses into the patch at the dead cell's
1 mm cell (a new patch at an empty cell, the nearest patch when full). Corpse
candidates report a cause and predator. Rules 2 results are unchanged.
Executed locally: MSVC Debug full build and `ctest -R cellular` (6/6, with five
new region-step cases: capacity split and region-wide conservation, size,
membrane and contact refusal, one prey per predator per tick, empty-cell
detritus, rules-2 no-op and invalid rules-3 intents); g++-14 `-Werror`,
clang-tidy-19 and clang-format 19.1.5 on the changed files.
## 2026-10-13 — M11.3f-2 rules-3 cellular owner (CI pending)
`CellularLives::create(..., CellularLivesProfile::engulf)` starts a region on
owner format 4 / rules 3 with the revision-format archive. Rules-3 ticks feed
membrane thickness and source to the resolver, keep each corpse's cause and
predator, add detritus patches, and validate rules-3 corpses as empty (rules 2
keeps the body-matter floor). Format 4 snapshots append a cause byte, seven zero
bytes and the predator id to each corpse record; restore refuses an unknown
cause, a predator on a starvation corpse or a non-empty corpse. Mutation
selection now accepts formats 3 and 4.
Executed locally: MSVC Debug full build and `ctest` (68/68, including a new
owner case: engulf-profile genesis, equal-founder `no_eligible_prey`, refused
engulf-plus-intake tick, starvation detritus into the patch, per-tick
conservation, format-4 round trip and tampered corpse causes); g++-14
`-Werror`, clang-tidy-19 (the one remaining `cellular_session_batch.cpp:110`
report predates this change) and clang-format 19.1.5.
## 2026-10-13 — M11.3f-3 engulf intent on the wire (CI pending)
Byte 12 of the canonical cellular input, the v10 `CellularAction` and the batch
trace intent record is now a target kind (0 none, 1 patch, 2 engulf with zero
coordinates). v7-v9 still refuse kind 2. `RegionSessions` admits v10 for
cellular servers with v9 behaviour (actions, datagrams, perception); a v10
mutation choice still closes with `protocol` until 11.4c. Admission, the
session-batch adapter, trace codec, trace/action matcher and action-block
validator carry the kind. The client session library accepts v10 and an
engulf flag, and its `send` no longer reports an empty (refused) encoding as
sent.
Executed locally: MSVC Debug `ctest` 68/68; client-network build with
`cellular-intent-protocol` (new v10 engulf case), `cellular-session-batch`
(engulf intent, trace round trip and stripped-flag mismatch) and
`quic-sessions` (a v10 client's engulf action admitted and carried; a v7 client
cannot send one) passing; `motion-trace-smoke` failed once under `-j 6` load
and passed alone. g++-14 `-Werror`, clang-tidy-19 (two existing reports in the
mutation test remain) and clang-format 19.1.5.
## 2026-10-13 — M11.3f-4 cellular v4 bundle and schema v34 (CI pending)
`cellular_v4_versions` (rules 3, envelope 4, archive 2) joins the allowed
cellular bundles with the revision history profile; a helper selects
format-3 traces for v3 and v4 in publication, preparation and recovery.
Provisioning maps a format-4/rules-3 opening to v4, and `evo-world-clone`
accepts `--opening cellular-v2` for it. Migration 34 replaces
`valid_cellular_action_input` (kind 2 with zero coordinates), wraps
`parse_checkpoint_manifest` to admit v4 through the v3 chain, and lets
`cellular_births` read format-3 traces for v3 and v4. `apply.sql`,
`migrate-postgres.sh` and `postgres-check.sh` extend to version 34.
Executed locally: MSVC Debug `ctest` 68/68, including region-checkpoint
contracts for the v4 genesis (round trip, restore, wrong history, v3 bundle
refusal, successor); g++-14 `-Werror` and clang-tidy-19 on the persistence,
provisioner and PostgreSQL test sources; `bash -n` on both scripts. The SQL
migration, `region_migration_tests` (v34 history, kind-2 input validation) and
`region_store_tests` (cellular-v2 provisioning under v4) run only in CI.
## 2026-10-13 — M11.4c-1 durable mutation publication (CI pending)
Migration 35 lets birth binding read format-3 traces that carry mutation
records, adds `cellular_mutation_selections`, `check_cellular_mutation_block`
and `publish_cellular_mutation_checkpoint` (binding each reserved choice to
the commit with its result), a matching resolver and a lease-fenced recovery
loader, and turns the old four-argument publisher into an internal helper
behind a wrapper that refuses traces with selections. The runtime privilege
audit lists the three new functions. `publish_cellular_batch` accepts mutation
actions, matches them to the trace, and takes their results from its replay;
prepared commits carry the block and results through retry and resolution;
recovery rebuilds the choices from the publication rows and compares results.
`cellular_publication_tests` adds a v4 region that feeds a cell to its first
offer, reserves and publishes a choice, recovers it, and restarts a host.
Executed locally: MSVC Debug `ctest` 68/68; g++-14 `-Werror`, clang-tidy-19
and clang-format 19.1.5 on the persistence, batch, host and test sources. The
migration and PostgreSQL tests run only in CI.
## 2026-10-13 — M11.4c-2 mutation choice admission (CI pending)
`CellularRegionHost::select_mutation` checks a v10 choice against the
committed offer, reserves it for head + 1 through `reserve_cellular_mutation`,
and queues it; `serve_cellular` passes the queue to the session batch builder
and publishes `mutation_actions()` with the batch; finalized choices go to the
owner's v10 session as `ActionOutcome`. Uncertain reservations, refused
publications and recovery reload pending reservations before the next publish.
`RegionSessions` dispatches `CellularMutationAction` to the new
`select_cellular_mutation` handler and adds `send_mutation_outcome`; the client
adds `send_cellular_mutation` (protocol only, no UI). Migration 35 gains
`cellular_mutations_due` and a `(world, region, target_tick)` index: both
publishers refuse a checkpoint that omits a reserved choice in its range.
Tests: `cellular_host_admits_mutation_choices` (PostgreSQL, world 9765) covers
invalid and stale choices, admission, an idempotent repeat, `busy`, reload
after a host restart, refusal of a checkpoint that skips the choice,
publication with its tick, the finalized outcome and its wire form, and the
final repeat. `quic-sessions` sends a v10 choice and receives the receipt and
outcome; a v7 session cannot send one.
Executed locally: MSVC Debug `ctest` 68/68; client-network QUIC/session 8/8;
g++-14 `-Werror`, clang-tidy-19 and clang-format 19.1.5 on the host, session,
server main and test sources. The PostgreSQL tests run only in CI.
## 2026-10-13 — M11.4d server half: owner offer replication (CI pending)
Session protocol v10 adds type 26 `CellularMutationOffer` (schema in
`schemas/network/session-v10-cellular-mutation-action.md`): the owner's
outstanding offer with up to eight unranked candidates in operator order and
each successor's capability value, maintenance and total mass. `offer_view`
builds it from the life; the cellular bridge sends it through
`RegionSessions::send_mutation_offer` when the offer or the owner's session
changes, and an empty view clears a consumed offer. The client keeps the
latest offer (`mutation_offer()`); the choice UI stays with the maintainer.
Tests: `cellular-intent-protocol` round-trips an offer and a clear, and refuses
it under v9, out of operator order, with a future offer tick, a non-empty clear,
an unknown capability, non-zero reserved bytes and a short payload.
`quic-sessions` delivers an offer and its clear to a v10 owner only;
`cellular_host_admits_mutation_choices` checks `offer_view` against a real
offer.
Executed locally: MSVC Debug `ctest` 68/68, network 74/74, client-network
88/88; g++-14 `-Werror` and clang-tidy-19 on the protocol, session, host,
client, server main and test sources (no new findings); clang-format 19.1.5.
## 2026-10-13 — M11.5a funded bud proposal (CI pending)
ADR-0125 and the M11.5 delivery plan define budding and wildlife descendants.
`evo::biology::fund_bud` is the pure proposal; `budding-contracts` covers
conservation on every channel, the largest-fit birth growth, the 30 s
provision, determinism, per-channel refusal with nothing spent, and the
not-grown, damaged and no-capability refusals.
Executed locally: MSVC Debug `budding-contracts`; g++-14 `-Werror`,
clang-tidy-19 and clang-format 19.1.5 on the new sources.
## 2026-10-13 — M11.5b owner budding and wildlife (CI pending)
The cellular owner gains format 5 / rules 4 (ADR-0125, schema doc updated):
bud intents, `fund_bud` funding, touching placement, region-namespace
identities with a durable sequence, archive birth of the parent's genome,
parent life on lives and corpses, and unowned wildlife without mutation
offers. `cellular-lives-owner` adds `rules_four_owner_buds_wildlife` and a rules-3
refusal of a bud intent.
Executed locally: MSVC Debug `ctest` 69/69; g++-14 `-Werror`, clang-tidy-19 and
clang-format 19.1.5 on the owner, region step and owner tests.
## 2026-10-14 — CI run 372 PostgreSQL failure and fix; M11.5c bud action and v5 bundle
Run 372 (187b1a7, M11.4c) failed only `checks (postgres)`: the M11.4c-1
publication test's first 64-tick feeding publish was refused client-side. The
PostgreSQL adapter bounded every parameter to 4 KiB of text (ADR-0059), and a
cellular trace is sent as one hex `bytea` parameter, so any trace over about
2 KiB failed as `bounds` and surfaced as `invalid`. A one-tick trace crosses that
at about 85 lives, which budding makes reachable. ADR-0126 raises the bound to
4 MiB, enough for the largest admitted trace. The same run would then have
failed on test commit ids: the new publication tests used uniform-byte commit
ids (`0x60`, `0x65`) that other suites also produce; they now use mixed bytes.
M11.5c adds bud target kind 3 (v10 actions, canonical input, trace intention,
action matching), the v5 bundle, schema v36 (kind 3 in SQL inputs, v5 bundle
normalization and birth binding), the `cellular-v3` provisioner opening, and
tests: protocol and canonical-input kind 3, session batch and trace round trip,
v5 checkpoint restore, v36 SQL checks, `cellular-v3` provisioning, and a
PostgreSQL bud publication with host restart and recovery.
Executed locally: `postgres-check.sh --full` (WSL g++-14 Release: C++ suite, life
reservations, backup, DR, upgrade, migration history, permission audit and checksum
drift; 489 s) passes; before that the C++ suite (`postgres-check.sh
--suite-only`, WSL g++-14 Release) passes, including the M11.4c-1/-2 and M11.5
publication tests; MSVC Debug 69/69, network 75/75, client-network 89/89 (the
two motion-trace smokes needed a serial rerun under load); g++-14 `-Werror`,
clang-tidy-19 (no new findings) and clang-format 19.1.5 on the changed sources.
## 2026-10-14 — M11.5d wildlife controller (CI pending)
`wildlife_intents` drives unowned lives (bud when fundable, else feed from the
nearest perceived energy patch, else wander) and `serve_cellular` records its
intentions in the trace baseline. `cellular-lives-owner` checks that only the
bud is driven, that two calls agree, that a cell touching a stocked patch feeds
without moving (or buds), and that the owner accepts the intention.
Executed locally: MSVC Debug `cellular-lives-owner` and `budding-contracts`;
g++-14 `-Werror -Wshadow` and clang-tidy-19 on the controller, budding, owner
tests and server main; clang-format 19.1.5.
## 2026-10-14 — M11.6a permadeath and restart (CI pending)
Schema v37 (ADR-0127) replaces "one life per account" with "one active life per
account": `ended_commit` on the reservation, a final-transition trigger, the new
unique index, and active-only reservation, owner lookup and birth publication;
`end_cellular_lives` records endings under the region lease. The cellular host
records endings after a death's publication and for every founder corpse on
recovery. `serve_cellular` refuses only the unusable carried action instead of
the whole tick. The publication tests build host configs as prvalues: copying
`CellularHostConfig` tripped a GCC 14 `-O3` array-bounds false positive.
Tests: `cellular_death_is_final_and_account_restarts` (world 9768) starves a
founder, checks the corpse and that the account has no active owner region,
starts a second life with a new identity, checks the first request still reports
its birth and a third request is `existing_life`, then restarts the host and
compares the recovered state.
Executed locally: `postgres-check.sh --full` (WSL g++-14 Release, 488 s) passes; g++-14
`-Werror -Wshadow` and clang-tidy-19 on the host, lease store, tests and server
main; clang-format 19.1.5.
## 2026-10-14 — M11.7 offline freeze, linger and resume (CI pending)
ADR-0128: the frozen step intent (trace intent byte 13, bit 0), owner format 6 /
rules 5 presence (active, lingering for 200 ticks, suspended) with resume in place
or within three extents, bundle v6, schema v38 and the `cellular-v4` opening. The
cellular server freezes owned lives whose account holds no session, refuses their
carried actions, keeps ticking while anyone lingers, and skips suspended cells in
replication and perception; the host refuses mutation choices for non-active lives.
Executed locally: `postgres-check.sh --full` (WSL g++-14 Release, 492 s) passes with
schema v38; MSVC Debug 69/69, ASan cellular/budding/mutation/lineage 10/10, network
75/75, client-network 89/89; g++-14
`-Werror -Wshadow` and clang-tidy-19 on the owner, region step, trace codec, host,
persistence, provisioner, tests and server main; clang-format 19.1.5.
## 2026-10-14 — CI run 374 TSan failure (test-only use after free)
Run 374 (e71147f) failed `checks (tsan)`: `cellular-lives-owner` read the bud
through a reference into the owner's previous state after the M11.5d controller
check advanced a tick. The test now reads the current life. ASan locally confirms
the fix; no product code was involved.
## 2026-10-14 — M11.8 server half: readable play signals (CI pending)
ADR-0129: v10 diagnostics carry engulf and bud results (bytes 77-78) filled from
tick feedback; v10 type 27 `CellularLifeEnded` (tick, life, cause) is sent once to
the owner when its life becomes a corpse; v9 views have the new bytes cleared. The
client stores the last life end. `cellular-session-protocol-v6-v10` covers the
codec (round trip, v9 refusal, unknown values, effects without tick effects,
type 27 bounds); `quic-sessions` delivers type 27 to a v10 owner only.
Executed locally: MSVC Debug 69/69, network 75/75, client-network 89/89; g++-14
`-Werror -Wshadow` and clang-tidy-19 on the protocol, sessions, client, tests and
server main (with the bridge); clang-format 19.1.5.