Files
aevolution/docs/STATUS.md
T
dingo 8b54e7447a
Foundation / checks (client) (push) Successful in 4s
Foundation / checks (checks) (push) Successful in 5s
Foundation / checks (ecs) (push) Successful in 4s
Foundation / checks (deploy) (push) Successful in 5s
Foundation / checks (network) (push) Successful in 4s
Foundation / checks (tsan) (push) Successful in 4s
Foundation / checks (postgres) (push) Successful in 4s
Foundation / postgres-tools (push) Successful in 5s
Foundation / server-image (push) Successful in 5s
Foundation / image (push) Successful in 4s
Close M11.9 for now and review the overhauled cellular UI
2026-10-17 11:35:56 -04:00

1492 lines
91 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.
# Project status and next steps
## M11.9 closed for now; M12 continuation - 2026-10-17
The maintainer explicitly closes M11.9 for now and authorizes proceeding with M12.
The solo observations and incomplete product sample remain follow-ups, not a
continuing gate block. Reviewed the current microscope HUD at three sizes,
mutation cards and central death/restart presentation through production-panel
renders; see [closure and UI review](engineering/M11_9_CLOSURE_AND_UI_REVIEW.md).
The current reference is rules 8 with finite pools/bubbles and contact feeding.
M12.2 still requires cellular/cohort fidelity and target-host fallback evidence;
no automatic tier selection follows from the M11.9 closure.
## M11.9 solo feedback - 2026-10-16
[The maintainer's eight answers](playtests/2026-10-16_M11_9_SOLO_FEEDBACK.md)
are recorded. Returning to a living cell is blocked by an existing-life message
with no cell visible; diagnose that resume/routing/view path first. The solo loop
felt repetitive, mutation effects and predation need more testing, and fun remains
uncertain. Food distribution, pacing and presentation suggestions remain proposals.
M11.9 product acceptance is open; this is feedback, not completion of the sample.
## M12.2b experiment - 2026-10-16
A separate exact-reserve/age candidate matches 83 synthetic histories, including
40 held-out cases, at every tick on MSVC and GCC. The old failed aggregation
fixture and evidence remain unchanged. Compact active controls show that cost
benefits depend on compiler and workload; no production capacity is inferred.
See [the experiment report](engineering/M12_RESERVE_STRATA_EXPERIMENT.md).
Release tests pass 70/70, with clang-format/tidy 19 and Linux ASan/UBSan checks.
M12.2 still needs M11 cellular multi-genome qualification and target-host fallback
budgets; M12.3-12.5 and the unrecorded M11 product review remain open.
## M12 start - 2026-10-15
M12.1 defines ecological identity, viability evidence and time boundaries in
[ADR-0133](adr/0133-ecological-identity-and-time-evidence.md), including explicit
deferral of species algorithms, autonomous mutation, evolving behavior and
whole-world wall-time catch-up. Existing M11 biology and runtime formats remain
unchanged. The [M12 delivery plan](engineering/M12_DELIVERY_PLAN.md) records the
open M11 product review and M12.2 population qualification prerequisites.
Automatic tier selection remains disabled; 12.3-12.5 cannot proceed on M4's
failed aggregation model.
12.2a reproduces all 15 historical scenarios, including eight false extinctions
and zero qualified widths; Release CTest passed 69/69. See
[population qualification](engineering/M12_POPULATION_QUALIFICATION.md).
M12.2 remains open; no production aggregation or unattended scheduling was added.
## Client update - 2026-10-14 (v10)
The desktop cellular client now negotiates v10, exposes mutations with funded
costs, engulf/bud feedback, reserves, death and restart, and retains unresolved
requests through reconnect. Automatic feeding, adjustable/precision steering,
food approach assistance and a responsive teal interface address playtest
usability. See [CLIENT](engineering/CLIENT.md) and the validation entry for
checks and remaining live-play/hardware qualification. No deployment or world
state change is part of this client update.
## Current gate - 2026-10-11 (M11.3 exploratory probe)
The maintainer reports that the Git checks passed, `evo-region` is running, and
the client can connect. This establishes deployment/connectivity, not that the
cellular actions or their usability have been exercised. The M11.3e one-player,
single-region exploratory probe is the next gate, before expanding mutation in
11.4; its instructions are in
[`M11_3_CELLULAR_PROBE.md`](playtests/M11_3_CELLULAR_PROBE.md). The v8 cellular
path, first-life result, steering, feeding and the player's observations still
need confirmation. M11.2's owner lookup, pre-claim redirect, client-follow path
and PostgreSQL restart test are implemented and pushed; Linux/PostgreSQL CI has
not yet been confirmed, so multi-region qualification remains open.
## Current gate - 2026-10-09 (M11.2)
ADR-0098 records the maintainer's decisions for the first playable cellular loop:
- **Budding:** the parent keeps control and funds one wildlife bud.
- **Mutation:** earned during life, offered as seeded candidates and applied through
funded development.
- **Heredity:** exact copy; wildlife does not mutate in M11.
- **Logout:** biology freezes at once; the organism lingers, frozen but vulnerable, then
suspends.
- **Resume:** in place, or at the nearest valid spot within a bounded radius.
- **Feeding:** absorption and engulfment by a size ratio contested by membrane
thickness.
- **Birth cost:** a first bud fundable within a session (a tunable playtest target).
- **Friend recognition:** deferred.
11.1's ratified engineering half, time-based biological rates, is complete (ADR-0099;
the maintainer chose directional rounding):
- **Segment 1 (done):** `accounting/rates.hpp` converts a per-second rate to a per-tick
amount without saved state. Costs round up and capacities down, so rounding never
favours the organism. The migration's exact ×20 inverse refuses inexact values.
- **Segment 2 (done):** genome schema 2 stores rates per second. Schema 1 records are
read ×20 and never rewritten. Development version 1 converts at its own 50 ms
quantum, mutation keeps a lineage's schema, and an archive of schema 1 records keeps
its bytes. No behaviour or golden changed.
- **Segment 3 (done):** environmental community rules 3 states irradiance, growth,
intake and decay per second. At 50 ms it steps identically to its rules 2 twin. The
dev world's rules 2 opening is unchanged. Development version 1 outputs stay tied to
their declared 50 ms quantum. The tick-abstract M2 fixtures are not migrated. ADR-0089's
rate migration is complete.
- **Development version 2 and the primitive founder (done; tunable):** rates are per
second. A full-grown founder that pays full-speed movement throughout funds its first
bud in 5, 10 and 15 minutes when feeding 100%, 50% and 33% of the time (ADR-0099).
11.1 is complete.
- **In progress:** 11.2, primitive lives and persistent control. Server side only; client UI
parts of 11.x are deferred to the maintainer. Its
[delivery plan](engineering/M11_2_DELIVERY_PLAN.md) and accepted
[ADR-0100](adr/0100-primitive-lives-and-persistent-control.md) identify the new
checkpoint and atomic biological publication contracts needed. The maintainer
authorized finite spawn funding and implementation choices for later review.
11.2a adds in-memory funded creation: full-grown primitive founders, equal
energy-only reserves, conserved pool/body/heat and exact successful retries.
11.2b adds a bounded creation-state checkpoint codec with full inventory/archive
closure and exact retry recovery. 11.2c-1 adds durable identity reservations,
world-wide account exclusivity, and retained refusals under fenced account
control. A reservation spends nothing and is not a spawned organism.
Atomic body/checkpoint publication and life binding are in the adapter. The
separate profile-3 region host now recovers and durably publishes a new life,
without advancing biology. v5 authenticated session execution and primitive
profile selection are implemented. The migration-24 life-owner lookup,
pre-claim redirect and operator route catalog are now implemented; two-region
restart and client-follow tests are pushed but await CI confirmation. The v17
PostgreSQL and host integration tests have prior CI confirmation.
Open from M10 (BACKLOG):
- the maintainer's native-client observation, moved to the later 2k drill;
- the Windows client failing to connect to the dev region after a CA renewal.
## Current gate - 2026-10-03
M10.1 ratifies M10 timing and workload budgets (ADR-0089,
schemas/network/m10-budgets-v1.md): 50 ms tick, 1 s commit flush with final outcomes
p95 <= 1.5 s, responsiveness LAN p95 <= 150 ms and 100 ms RTT p95 <= 250 ms, server
budgets on one RKE2 worker and client budgets on the low-spec VM. Development-reference
traces (`evo-dev-session trace`, networking/M10_TIMING_TRACES.md) measured admission
p99 <= 5.2 ms, durable p95 about 1.4 s and 50 ms view cadence. Finding: genome v1 and
metabolism use per-tick rates, so the quantum fixes biological rates until M11.1
converts or ratifies them. M10.2a selects QUIC with msquic 2.6.2 and OpenSSL 3.5.9 LTS
(ADR-0090), pinned and building on Windows and Linux with a loopback smoke test. M10.2b adds
the bounded QUIC endpoint (datagram and reliable channels, exact CA pinning with the owned
check authoritative, minimum-MTU start) and the owned impairment relay, tested on both
platforms. M10.2c measures loss and head-of-line blocking through the relay: under 5-10 %
loss the ordered stream's p95 latency is 3-5.5x the datagram channel's and exceeds the
250 ms target at 100 ms RTT, while datagrams stay <= 67 ms, which confirms the channel split
(networking/M10_TRANSPORT_MEASUREMENTS.md). M10.2 is complete. M10.3a (ADR-0091) adds provider-independent account ids, ECDSA P-256
session grants verified with public keys only, the identity adapter and grant issuer
library, and per-region session control (newest grant wins and replaces the old session,
replay and stale grants refused, in-session refresh, attempt rate limits). M10.3b adds the
session login over QUIC (session-v1: Hello, Welcome, Refresh, Close with reasons) on both
the region and client side, tested end to end on loopback. M10.3c adds durable per-world
account control (schema v10): grant serials survive restarts and apply across regions,
cross-region takeover waits out the previous holder, and sessions close when their
control is fenced or cannot be renewed. M10.4a (ADR-0092) adds
intentions (datagram receipts and a cumulative simulated watermark) and account-scoped
actions with stable ids, receipts and final outcomes to session-v1, with admission
(window, rate, sequence, duplicates, unresolved bound) and client support; intentions and
actions have no biological effect yet. M10.4b makes
action outcomes durable: schema v11 records each batch's actions in the same statement as
its checkpoint, the host attaches actions to the ticks they ride, and outcomes are read
from the durable record after crashes before or after a commit, unknown publications and
cross-region moves. M10.4c adds the opt-in
evo-server QUIC listener (session control and action records under the host's lease) and
the headless `evo-session-probe`, with real-process tests including SIGKILL before and after
an action's commit. M10.4 is complete. Next: M10.5 bounded snapshots and spatial interest.
Deployment (2026-10-05): the maintainer deployed the M10 QUIC listener to evolution-dev
behind the ADR-0087 route; session protocol v1 is therefore frozen. The route admitted
only TCP 7400 while QUIC is UDP, so it is changed to UDP 7400 (ADR-0087 amendment). The
route is not yet verified end to end: the edge firewall rule (UDP 7400 from
172.10.1.0/24), internal DNS to 172.10.1.202 and a probe run through the VIP remain.
M10.5a (ADR-0093) adds session protocol v2 (interest, snapshot chunks, snapshot acks; v1 is
refused) and the replication core: server-capped interest with a perception boundary,
entity deltas against the acknowledged baseline, nearest first within 16 KiB per tick with
a slower distant tier, datagram chunks and client assembly. M10.5b wires it end to end:
evo-server replicates the region's environment for each session's interest (default: the
first habitat site), the client assembles and acknowledges snapshots, and the probe prints
them. M10.5c switches the engineering client to M10 sessions: snapshot-fed views, the camera
driving the interest after the user moves it, grant bundles (`evo-session-probe grants`,
evo-server `--m10-max-grant-lifetime` for dev), and `msquic.dll` in the Windows package.
M10.5 is complete. Deploying the v2 server to evolution-dev (the dev server still runs v1)
is a maintainer step. M10.6a (ADR-0094) adds the probe-motion experiment: one
non-biological marker per session moved by intentions with a shared integer rule, session
protocol v3 marker records (v2, deployed, is frozen and refused), and the client predictor
(bounded replay from the authoritative marker, corrections, smoothing, adaptive input lead).
M10.6b wires it: evo-server gives every session a marker in the habitat area, the probe
prints it, and the engineering client predicts and draws it (WASD), following it with
camera and interest. M10.6c traces it through the impairment relay: exact without loss,
94-99 % exact at 60-200 ms RTT with 2-5 % loss, convergence after stopping and a restart
resync of about 2-3 s (networking/M10_PREDICTION_TRACES.md). M10.6 is complete. M10.7a (ADR-0095) adds a drift-corrected tick clock, adaptive
interpolation of remote markers (2-6 ticks behind, at most 2 ticks extrapolated, then
stale; organisms never) and sub-tick own-marker presentation. M10.7b wires it into the
engineering client. A lost session's view stays frozen and labelled stale until the
reconnect's fresh baseline. In two-client traces, remote markers shown as current stay
within about 0.1 m of the true path, and outages show stale. M10.7 is complete. M10.8a (ADR-0096, the ratified handoff ADR) adds schema v12:
- durable probe records, one owning region per account;
- a fenced transfer state machine (prepared, staged, committed, aborted) whose every retry
returns the recorded outcome;
- an account-control handoff for committed transfers.
A PostgreSQL fault matrix covers restarts at each transition, stale epochs, aborts,
partitions and commit/abort races, checking a unique owner and the conserved test inventory.
Deploying v12 needs a rebuilt migrations image. M10.8b adds signed handoff tickets. A source
region's key signs one transfer for one account and destination; the ticket rides the Hello
grant slot and is verified against the neighbour keys registered here. The destination then
durably stages and commits the transfer and takes the account's control before admitting
the session. M10.8c-1 wires the regions:
- session protocol v4 (handoff offer, redirect, ghosts; v3 is deployed and frozen);
- sessions use the durable probe: opened where it was saved, redirected when it lives
elsewhere;
- a marker crossing into a neighbour's area is frozen, prepared and offered with a ticket;
- the destination shows arriving probes as ghosts (schema v13) and opens the marker where it
crossed.
M10.8c-2 has the engineering client follow crossings (ticket, no loading screen) and
redirects, with grants picked by region in rising serial order. evo-server gains
`--m10-area`, `--m10-handoff-key` and `--m10-neighbour`; the probe tool writes multi-region
bundles. M10.8d runs two evo-server processes on PostgreSQL through crossings and
crossings back, with SIGKILL of the source after prepare and of the destination after stage
and after commit, duplicate and stale ticket retries, a client that never arrives and a
destination that is down. After every scenario each probe has one owner, the inventory is
conserved and a record is frozen only under its open transfer. It found and fixed two
recovery gaps:
- a source restarted after a prepare now aborts overdue transfers from the database
(schema v14);
- a ticket replayed at a restarted destination after the probe moved on is refused.
M10.8 is complete. Deploying needs protocol v4 and schemas v12-v14 (a rebuilt migrations
image). M10.9a (ADR-0097) starts load qualification. Only runs on the evolution-dev reference
worker count as evidence. evo-server now measures each tick's work (host step plus session
layer) and feeds a load governor:
- above a p95 of 15 ms, or with tick debt, new grant logins are refused (`capacity`), while
crossings are still admitted;
- above 20 ms, replication sheds radius and far-update cadence in up to 3 levels;
- both relax after 5 s of calm.
`/metrics` exposes the work percentiles (total and session layer) and the governor's
state. M10.9b adds `evo-load`, a multi-client load generator. It runs bots on worker
threads that send intentions every tick, actions, interest moves and reconnects. It reports
per step pooled and per-client distributions, a starvation check, a budget verdict and the
region's gauges, ending with the sustainable bot count. In a local development reference
(not evidence), one region stayed within budget up to 24 bots and the governor began
refusing at 32 (work p95 about 14 ms), with responses about 103 ms and outcomes about
665 ms throughout. M10.9c-1 adds the abuse suite's first half:
- a libFuzzer target for session protocol v4, grants, tickets and snapshot assembly (23
million inputs, no findings), also run as a seeded smoke test on every platform;
- in-process floods (Hello, datagram), a slow peer and a connect/disconnect storm, during
which an honest session kept every snapshot and receipt.
`evo-session-probe` can now log in from a renewal's grant bundle. A dev login was welcomed
and its intention admitted (the dev region does not simulate outside drills). M10.9c-2
kills and drains a real evo-server under 24 `evo-load` bots. All of them log in to the
successor within 5.0 s (SIGKILL, after a 3.5 s lease wait) or 2.4 s (drain), with no failed
logins, and responses stay at 106 ms p95. M10.9d adds the evolution-dev load drill
(`tools/ops/load-drill.sh`). It turns synthetic inputs on for the run, drives bundle-mode
`evo-load` bots through the public route, and matches each step with the server's 10 s
`m10 load-report` log lines. Building it found that Mbed TLS (built without its threading
layer) raced when bots verified connections on several threads; one process-wide crypto
lock fixed it.
Reference runs on evolution-dev:
- **Third:** 24 sessions within the client budgets. It led to budgets v2 (commit ticks
≤ 50 ms, measured apart from the rest) and to watermark coalescing.
- **Fourth:** the first run whose actions committed. It exposed one synchronous store
round trip per action on the simulation thread. Lookups are now batched: one per poll
and one per commit, with no schema change.
- **Fifth:** clients within budget at 20, 24, 28 and 32 sessions. The server was within
budgets v2 at 24–32 (work p95 ≤ 12.9 ms, commit ticks ≤ 44.3 ms, no tick debt) and
reached the configured 32-session cap without saturating. At 20, the login ramp drew 4
refusals and per-session p99 was 0.554 ms (budget 0.5).
The maintainer closed 10.9 for now (2026-10-09):
- evo-server's session cap is raised to 2000, with the load governor as the working limit;
- drills stop for now. Saturation and native-client observation move to a later 2k drill.
ADR-0097 lists what that drill must solve first: single-thread CPU (the governor would
cap near 50–70 sessions), the region login rate (64 per 60 s), the 256 MiB pod limit and
linear session lookups.
## Previous gate - 2026-10-02
M9.8a: Dear ImGui 1.92.9b is pinned and qualified in an isolated Windows/Linux
SDL3 probe (ADR-0084). M9.8b adds bounded notification/dialog state, protected
external input capture and bounded session events with disconnect/drain/reconnect
backpressure. M9.8c adds the launcher (labelled local fixture vs dev region,
bounded errors, Retry/Cancel), legend, labelled inspector, click selection, session
panel with the intention outcome record, and visible notifications/dialog. M9.8d
adds rebindable controls (ADR-0085) with conflict checks and a validated file. M9.8e
builds a static-runtime Windows x64 ZIP locally with notices, identity, checksums and
credential guards (ADR-0086); Actions publication needs a Windows runner. M9.8f designs the dev session route (ADR-0087), with the local test-route inputs now
reserved in [VOID_SESSION_ROUTE_REPORT](engineering/VOID_SESSION_ROUTE_REPORT.md),
while the route remains disabled with fail-closed policy tests; nothing was applied to
the cluster. Tests,
visual review and limits are recorded in engineering/VALIDATION.md.
Next: implement and test server-enforced session windows, then include the prepared
ADR-0087 route and complete DNS, certificate, Argo, health-check-port and connectivity
validation. The local route inputs are recorded in engineering/VOID_SESSION_ROUTE_REPORT.md.
Windows packages will be built locally until a Windows runner exists. Testing is internal; 172.10.1.160 is the
RKE2 control-plane address. VM qualification and the cluster restart drill remain
outstanding, as agreed. See engineering/M9_8_DELIVERY_PLAN.md.
## M9.7 delivered scope
M9.7 connects the engineering client to an authoritative dev region
([ADR-0083](adr/0083-dev-region-sessions.md),
[DEV_SESSION_OPERATIONS](networking/DEV_SESSION_OPERATIONS.md)). evo-server runs a
non-blocking TLS 1.3 session server in its loop; clients see environment views from
the region's real field/terrain stores and an explicitly empty organism layer. An
admitted intention rides a synthetic input and returns a durable, no-effect outcome
after the host's commit. Reconnects consume fresh grants and start from a fresh
baseline; stale epochs, duplicates, foreign grants and floods are refused.
Proceeding with the maintainer's agreement although **M8-dev/stage is not passed**.
Local evidence is real (evo-server + PostgreSQL, SIGKILL + successor); the dev
deployment path (drill-time override, loopback listener, optional Secret,
`tools/ops/session-drill.sh`) is prepared but **not run**: no cluster access was
available. Next: **M9.8** packaging and presentation qualification.
M9.6 defined the dev-session protocol and TLS 1.3 channel
([ADR-0082](adr/0082-dev-session-transport.md)).
M9.5 added biology-derived organism rendering without controller labels
([ADR-0081](adr/0081-organism-presentation.md)).
M9.4 added owned revision-tagged environment presentations and explicit
missing/stale rendering ([ADR-0080](adr/0080-environment-presentation.md)).
M9.3 added local-mm pan/zoom camera mapping and bounded sequence/tick input
intentions ([ADR-0079](adr/0079-camera-and-input-intentions.md)).
The sections below retain earlier milestone delivery history.
Updated 2026-09-27. The repository contains an engineering foundation and a
deterministic headless proof. It does not yet contain cellular gameplay or a
persistent simulation server. The founding vision remains unchanged.
## Completed implementation
- Vision, architectural pillars, risks, open-question register, architecture
diagrams, biological contracts, engineering constitution and 58 ADRs (including the accepted M7.1 durability contract).
Proposed/deferred choices remain distinct from accepted proof decisions.
- M0–M4 task breakdowns with dependencies, acceptance evidence and risks,
plus M5–M14 gates in [ROADMAP.md](ROADMAP.md).
- C++23 CMake/Ninja library and CLI with no graphics, networking, database or
Kubernetes dependency. Versioned fixed ticks, stable entity IDs, component
lifecycle and explicit per-entity seeded randomness.
- Bounded canonical snapshots and restore, plus replay verification. An independent
reference fingerprint and Release-active tests cover lifecycle, malformed data,
overflow, seeded properties and CLI save/load.
- Formatter, static analysis, sanitizer presets, codec fuzz entry point, benchmark,
structured host logging and metrics output. Benchmark compiler metadata now
resets numeric formatting after printing its hexadecimal state hash.
## Delivery and verification
Git remote is `https://git.aeternalisgames.com/dingo/AEvolution.git`. The maintainer
added a working Gitea/Harbor workflow; run 1399 on `8ab48d3` successfully verified
and published. The foundation update extends that single workflow with Linux
analysis/sanitizers/fuzzing and runtime smoke verification. Images target
`registry.aeternalisgames.com/aeternalis/aevolution`.
[Run 1402](https://git.aeternalisgames.com/dingo/AEvolution/actions/runs/1402)
passed all jobs for source commit `4c2cc04`: Linux checks, separate TSan and image
smoke/publish. That verified image was previously pinned. The maintainer's
subsequent GitOps commit `20e609d` selects the run 1399 image at
`sha256:0c6c6e1ac28de2ce0648fd0992ca694963e303ef5c9414fe5ac668937136800d`;
the M1.6 change preserves that independent deployment choice.
Windows checks also pass; full commands and intermediate CI failures are recorded
in the validation report. M0's engineering proof is verified; dev cluster
execution is the remaining delivery check, not a claim of persistent-world readiness.
The maintainer confirms Argo is configured and deployment additions will sync
when a pod workload is ready. This workspace has no kubeconfig, so completed dev
Job execution is not independently verified. This is not a blocker for M1.
The repository supplies a restricted dev Job and bootstrap examples; image
publication does not demonstrate a Kubernetes rollout or world recovery.
See [VALIDATION.md](engineering/VALIDATION.md) for actual commands/results and
[deploy/README.md](../deploy/README.md) for image promotion and application setup.
Local `INITIAL_BRIEF.md` and `OpenAI_ChatLog.md` are maintainer-provided untracked
material and are not automatically included in commits.
## Next reviewable increments, in order
M1.5 progress: the opt-in row/EnTT4 experiment now checks canonical equality,
restore continuation, retired identities and allocator balance, and records
seven-sample sparse/dense/churn results. Windows Release, analysis and ASan checks
passed, as did Linux Release, analysis and ASan/UBSan in Gitea run 1404. All 56
sample fingerprints match across platforms. ADR-0005 retains rows
for the bounded proof; the general production ECS remains a later measured choice.
| Priority | Work | Completion evidence |
|---|---|---|
| 1 | M7.2 adopt storage dependencies and migrations | Pinned PostgreSQL client/server evaluation, guarded schema, real migration and error-path tests |
| Research | Improve aggregate fidelity and measure Linux/target-host costs | M4.6 found no qualifying width; see the aggregate budget report |
| 2 | Revisit the ADR-0031/0032 reproduction cost model | Reproduction is affordable only at the genome schema's floor; decide whether that is intended before M4 builds cohorts on it |
| Delivery gate | Add/update the dev workload when a deployable increment needs it; Argo is ready | Observe Job/pod completion separately from image publication; no World0 changes |
M1.6 now implements [the sequential scheduler contract](architecture/SCHEDULER_CONTRACT.md):
exact-tick admission, bounded ordered lifecycle commands, receipts and atomic
rollback on rejection. Windows Release, clang-tidy, ASan and formatting checks
pass. Permutations across 32 seeds, restored continuation, duplicate/late/future
inputs, limits and exhaustion are covered. The no-input golden and snapshot v1
are unchanged. Gitea run 1407 passed all Linux checks, ECS, TSan and image jobs;
evidence is recorded in the validation report.
Nonempty batches copy world state; this favors a clear transaction boundary for
the proof, not a measured realtime performance claim. No thread pool or gameplay
input protocol is introduced.
M1.7 implements read-only periodic neighborhood queries with stable IDs and
inclusive integer-distance boundaries. A nine-image oracle checks 2,304 seeded
queries across initial, scheduled and restored worlds. Boundary and full-capacity
tests supplement the oracle. The benchmark records uniform/clustered worlds up
to 100,000 entities. This is a linear correctness baseline, not a production
spatial index. See the spatial contract and validation report for scope/evidence.
Native Release/ASan/analysis/format checks passed, as did all jobs in Gitea run
1409. All 72 Linux benchmark result checksums and match counts equal Windows.
M2.1 adds an independent finite-inventory accounting library (ADR-0024). Typed
picojoules and C/N/P/other femtograms transfer atomically across bounded local
reservoirs. Opening totals are checked, repeated action IDs rejected, and failures
leave both balances and action watermark unchanged. One million seeded transfers
are checked against independently maintained balances and recomputed totals.
No organisms, biochemical rates, external replenishment, world snapshot format
or deployment changes are included. Conservation does not establish valid
chemistry or reusable heat; process restrictions are M2.2 work.
Native Release, clang-tidy, ASan and formatting passed; Gitea run 1411 passed all
Linux checks, ECS, TSan and image publication. Validation details are recorded
in the engineering validation report.
M2.2 implements a synthetic finite-source producer process owning its ledger.
It captures energy, sends conversion losses to heat, and funds whole material
recipes from reserves and finite nutrients. Fractional rates and efficiency
carry exact remainders; tick/remainders and balances commit together. Tests cover
hand-calculated outcomes, 32 seeded reference/replay runs, source and nutrient
exhaustion, invalid tick/config rejection and u64 arithmetic limits. No organism,
genome, database, process snapshot or deployment is implied by this fixture.
Native Release, analysis, ASan and format checks passed, and Gitea run 1413
passed all Linux checks, ECS, TSan and image jobs.
M2.3 adds a shared feeding community with one producer and up to 16 consumer
capabilities. One ledger owns all process resources, heat and waste. Ordered
intake-capped feeding, producer progression and terminal depletion commit as one
tick; a rejected request leaves the entire community unchanged. Tests cover
predation, resource contention, terminal actors, batch order/replay and conservation.
No spatial/perception validation,
organism biology, persistent checkpoint or deployment is included in this fixture.
Native Release, analysis, ASan and formatting passed; all four Gitea run 1415
jobs passed.
M2.4 adds terminal events with preserved corpse inventory, followed by a shared
bounded decomposition phase. Energy moves to heat, elemental matter returns to
nutrients, and next-tick producer growth can use those nutrients. Corpses retain
their process ID and cannot resume activity. Invalid ticks roll back death and
decomposition with all earlier phases. Default zero decay preserves earlier
fixture behavior. This is not death-cause biology, microbial chemistry, corpse
scavenging, persistent lineage history or a database-backed world.
Native Release, analysis, ASan and formatting passed, and all four jobs in Gitea
run 1417 passed.
M2.5 adds funded budding (ADR-0028). A bounded list of trusted reproduction
intents runs after feeding and before decomposition, so a consumer can fund
offspring from what it ingested and a process terminated in that tick cannot
reproduce. Parent eligibility is fixed on entry to the phase, so a child born in
a tick cannot itself be a parent until the next one. The intent names a
withdrawal from the parent; a fixed construction
fraction endows the child and the remainder becomes real heat and waste. Reserved
slots are pre-allocated at creation, so a birth never grows the ledger and a full
community rejects rather than funding a smaller child. Process identities are
never reused and a parent reference plus origin tick outlive both organisms.
A child duplicates its parent's configuration because no genome exists yet; this
is duplication, not heredity. Default zero capacity leaves every M2.4 result and
the particle goldens unchanged. Native Release, analysis, ASan and formatting
passed; Linux CI for this increment has not run yet. This is not a reproductive
strategy, fecundity rule, mutation model or funded founder admission.
M2.6 closes M2 with a deterministic ecosystem driver (ADR-0029). It owns one
community and one seeded stream, holds no inventory and emits only the intents
the earlier increments already accept, so no simulation contract changed. Two
limits shaped it: retained identities and capped slots mean one community is not
a population, and with no maintenance cost an uneaten consumer never dies. The
harness therefore measures many bounded episodes, with mortality from predation
and a configured lifespan delivered as a host terminal fact. The driver probes a
discarded copy each tick so it never submits a batch the community rejects; a
rejected tick is reported as a driver defect rather than absorbed as ecology.
`evo-sim --ecosystem` reports seeded episode ensembles: 3,000 episodes advanced
1,093,007 ticks, all reaching extinction, with deaths reconciling exactly against
founders plus births and conserved totals identical at every tick. That
demonstrates arithmetic and turnover, not ecological plausibility; strategy
constants are chosen inputs and must not become gameplay by having been measured.
Native Release, analysis, ASan and formatting passed; Linux CI has not run yet.
M3.1 opens M3 with the immutable genome record (ADR-0030) in a new `evo::biology`
library that nothing else links yet. A genome is validated once at construction
and exposes no mutator, so an invalid one cannot exist at runtime; `decode`
applies identical checks, so a decoded record is never less validated than a
constructed one. Schema 1 fixes the seven v0.1 capability kinds with units,
inclusive ranges, required flags and mandated developmental edges as schema facts
rather than record data. Identity is a nonzero world-scoped genome ID distinct
from every runtime identity, a nonzero origin event and zero or one parent, with
room reserved for recombination. The encoding is canonical, bounded and exact:
strict ascending order, a stored digest that is recomputed on decode, and a
trailing checksum, with the byte layout in schemas/genome-v1.md. No genome
affects any simulation; the M2 fixtures still duplicate founder strategies.
Development, phenotype, mutation and heredity remain M3.2 to M3.5. Native
Release, analysis, ASan and formatting passed; Linux CI has not run yet.
M3.2 derives a phenotype from that record (ADR-0031). `derive_phenotype` and
`plan_development` are pure integer functions of genome, growth and environment,
so a phenotype is recomputed from retained inputs rather than cached. Growth and
damage are runtime state that never reach the genome: damage changes the
phenotype and leaves the content digest identical, which is the risk this
increment was scoped against. The genome's development version selects a rule
set and an unknown version fails closed. Version 1 declares every constant:
the body is a cube of uniform declared density, membrane mass follows surface and
thickness, maintenance follows mass plus a membrane upkeep term, and uptake scales
with surface, so size trades uptake against maintenance. A capability the genome
does not carry yields zero. A plan advances growth by at most the rule step,
reduced to the largest step the available material funds exactly, and reserves
nothing: no inventory moves until M3.3. Omitted physics is a documented model
limitation, not a modifier, and the constants must not become tuning by having
been written down. Native Release, analysis, ASan and formatting passed 18/18;
Linux CI has not run yet.
M3.3 gives capability a cost (ADR-0032). `metabolic_budget` returns body
composition, maintenance, movement, construction, the per-tick intake limit and
the reproduction investment, in accounting types; `metabolism` is the single
translation layer and the genome record stays free of those types. Each budget
comes from a different phenotype quantity, so one parameter moves some and not
others: a thicker membrane costs more matter to build, more energy to hold and
more to haul, while leaving uptake, speed and assimilation identical. That
non-uniformity is asserted directly, because a capability that became a flat
multiplier is the failure this increment was scoped against. Composition is
declared per compartment, and splits give the remainder to the last channel so
parts sum back to the whole and a derived body can be spent without a rounding
leak. `processes::consumer_for` is the first path from a genome into the M2
fixtures, with opening inventory left to the caller so no matter is invented; a
derived consumer is shown driving a real feeding tick with the ledger conserving.
No existing fixture reads genomes yet and no organism spends any budget. Native
Release, analysis, ASan and formatting passed 19/19; Linux CI has not run yet.
M3.4 proposes successors (ADR-0033). A static catalog of four effects over the
seven capabilities is filtered by the M3.1 validation itself, so eligibility is
not a second rulebook: shedding a capability another declares as a prerequisite
fails without a rule written for it. Candidates are drawn without replacement by
rejection sampling on an explicit stream, so every eligible operator is exactly as
likely as any other, and the result is returned in canonical order with no score
anywhere in the record. The draw reads only the parent digest and a recorded
context, so the same context always returns the same candidates and a retry cannot
reroll, while a new event index earns a new draw. Selection regenerates the offer
and refuses a changed context outright rather than answering it with a different
mutation. A candidate reports what ADR-0032 would charge its successor and
reserves nothing. The uniformity test was confirmed to fail against a draw that
took the first eligible operators. Candidate count, mutation cost as a game rule,
inherited versus acquired expression, NPC mutation and recombination stay open;
no organism mutates. Native Release, analysis, ASan and formatting passed 20/20;
Linux CI has not run yet.
M3.5 keeps what M3.4 produces (ADR-0034). `LineageArchive` has no removal
operation at all, so retention is a property of the interface rather than a rule
someone has to remember, which is the only form of "keep extinct genomes
indefinitely" a later change cannot quietly undo. Admission requires the parent to
be present, so ancestry never dangles and the parent graph is acyclic by
construction; a duplicate identity is refused rather than overwritten. Organism
and genome identity stay separate: a death is terminal for the organism and
touches nothing about the genome, and ancestry, descendants and divergence all
resolve after every carrier has died. A lineage with no living carrier is reported
absent, never extinct, because terminal extinction needs a reconciled watermark
covering suspended members, eggs and in-flight transfers that this type cannot
see; absence is also reversible. Capacity is bounded and admission fails loudly,
since refusing to record is not deleting. NPC mutation stays undecided: the
archive records an origin event and never an actor. Nothing is serialized, so
durable retention and migration remain M7 and M8. Native Release, analysis, ASan
and formatting passed 21/21; Linux CI has not run yet.
M3.6 closes M3 by making the M3 pieces meet (ADR-0035). `strategy_for` derives a
`StrategyConfig` from ADR-0032's budget, so a strategy is what the genome already
implies rather than a second description of the same organism, and a parent
reproduces once it holds what it would invest. Founders are variants of one
ancestor diverged by ADR-0033's draw before any episode starts and admitted to
ADR-0034's archive with their parent links; every birth and death is recorded as
it happens, so the diversity a run reports has a provenance. Nothing mutates
during an episode, because autonomous divergence would settle NPC mutation, which
stays open. `evo-sim --ensemble` runs the grid and reports each cell's seeds in
two halves that are never merged. ADR-0029's driver is untouched and the M2.6
fingerprint still holds.
Two findings are recorded rather than smoothed over. Resource level never reaches
strategy derivation: founders are derived fully grown, so their construction
transition is empty and ADR-0031's per-channel `available` cannot move intake,
assimilation or investment; the resource level acts only as the nutrient pool
bounding producer growth. And the resource axis is not monotonic, so no test
asserts a direction. A sweep peaks near 8000 femtograms and falls away on both
sides, because ADR-0026 lets only one consumer bite a target per tick: the
community's income is capped at one producer bite regardless of the pool, scarcity
drives predation that concentrates mass into individuals that reach the threshold,
and abundance spreads the same income too thin to fund a birth. Separately, the
fixture's reproduction investment is pinned to the genome schema's floor because
intake scales with surface while investment scales with mass; at any larger
fraction nothing can earn a birth within a lifespan and both axes measure nothing.
That is a finding about the ADR-0031/0032 cost model, not a fixture detail.
Review of `89ad95b` found three defects, all now fixed with regressions confirmed
to fail without their fix. The harness recorded a death for every organism still
alive when an episode stopped, so a one-tick run reported no extinctions while
writing terminal records for founders that had not died; ending observation is not
death, survivors now stay living in the archive and their watched time is summed
apart from completed lifetimes. A draw that produced a genome without the
reproduction capability derived a zero investment, which ADR-0029's driver refused
outright, so seeds 10, 12 and 15 failed the whole run; the driver now runs such a
strategy and proposes no birth for it, because filtering those variants out would
have made the measured set the fertile subset chosen by a guard. And a half's
thermal shortfall was assigned rather than accumulated, so a field documented as a
total reported one episode's reading. The M2.6 fingerprint `2cd4a15f9fbe9949` is
unchanged, since only configurations the driver previously rejected behave
differently; the ensemble CLI pin was repinned because the reported facts changed.
Native Release, analysis, ASan and formatting passed 33/33; Linux CI has not run
yet.
Before biology, resolve the relevant heredity and energetic contracts rather
than treating proof particles as organisms. Exact game tick rate, NPC mutation,
logout treatment, species/extinction rules and multicellular transition remain
open decisions. Renderer/networking remain later milestones. Deployment
availability does not complete M7/M8: fenced ownership, durable outcomes, backup
restore and world-version migration require their own implementation/fault tests.
M4.1 opens M4 (ADR-0036) by defining the tier 2 record and who may act on it, and
nothing else: there is no collapse transaction, no population advance, no
materialization and no habitat geometry. A `Cohort` is validated on construction
and never mutated afterwards. Age classes each carry their own count and their own
reserve, so the record holds a distribution rather than the mean the milestone's
named risk warns about; within a class members are indistinguishable, which is the
record's declared resolution and a stated loss rather than a hidden one. No
accessor returns a mean. Totals are summed from the classes instead of stored
beside them, following the lineage archive's rule that a second tally is a thing
that can disagree. Lineage is deliberately not a field: ancestry is resolved from
the genome through the archive, so a cohort cannot contradict it or invent
parentage from membership. Residuals are integer numerators over a declared scale,
strictly below one member, and are never added to a count.
Ownership is enforced rather than described. `CohortRegistry` admits at most one
record per habitat and genome, so ECOLOGY_MODEL's "no inventory is spendable in
two tiers" is unrepresentable rather than forbidden, and replacement is refused
under a different worker, an older epoch, or a changed habitat or genome. Epochs
make a stale writer fail locally and say nothing about fencing between workers,
which stays M7 and M10. The approximation bound is carried and never computed:
M4.5 measures resolution equivalence, and carrying a bound is not meeting one.
Nothing is serialized, so `schemas/population-v1.md` pins field meaning and
invariants rather than a byte layout. All nine construction and registry guards
were confirmed load-bearing by removing each one and observing the suite fail.
Native Release, analysis, ASan and formatting passed 34/34; Linux CI has not run
yet.
Review of `1c8b292` found two defects, both fixed with regressions confirmed to
fail without their fix. Replacement checked the ownership epoch but nothing that
ordered two writes 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 one had already spent. Cohorts now carry a
revision that strictly increases within an epoch, and the interval boundary is
checked independently, so neither advancing a revision nor re-acquiring ownership
rewinds simulated time. Separately, the implicit move emptied a cohort's age
classes while keeping its identities, and the registry then accepted through a
move exactly what `create` rejects by name; the type is now copy-only, as ADR-0030
already made the genome for the same reason. That precedent existed and should
have been applied when the type was written.
M4.2 moves individuals into that record (ADR-0037), and does only that: there is
no population advance, no materialization back into individuals, no transfer
between habitats and no persistence. Collapse is two phases over one sample.
`plan_collapse` reads the members once and returns an immutable plan that already
carries a validated cohort; `commit_collapse` applies exactly that plan and
derives nothing, so ECOLOGY_MODEL's rule that recovery replays a committed sample
rather than sampling again is a property of the types.
A controlled, suspended, dead or unsettled member refuses the whole plan rather
than being skipped. That is the answer to adversarial selection: filtering would
let a caller obtain a plan by omitting whoever was inconvenient, and the result
would be indistinguishable from an honest conversion. Ages are carried into
declared classes and a member fitting none refuses the plan, so a conversion
cannot quietly move anyone to a different class or discard one for having nowhere
to go. Reserves are summed per class, never redistributed.
Conservation runs through the ledger rather than beside it: one atomic transaction
of one leg per member, each moving that member's entire reserve, so the conserved
total is unchanged by construction. Dual representation is closed from both ends.
A member must be emptied exactly, and a reservoir holding anything other than what
was sampled makes the plan stale rather than partly applicable. The destination is
proved available before anything moves, so no rejection can strand inventory in a
reservoir nothing owns, and every refusal leaves the ledger and the registry as
they were. Committing one plan twice is refused without moving inventory again.
Two limits are stated rather than implied to be covered. The function cannot tell
whether the members it was given are all the members there were, so completeness
is a caller obligation. And emptying a reservoir is not withdrawing the right to
use it: the ledger has no removal, so nothing prevents a later transfer crediting
that reservoir again. Durable conversion sequences, escrow and replay across a
process boundary are M7 and M8. All ten construction and commit guards were
confirmed load-bearing by removing each and observing a failing run. Native
Release, analysis, ASan and formatting passed 35/35; Linux CI has not run yet.
Review of `1a59b63` found one defect, now fixed with a regression confirmed to
fail without it. Commit checked that the destination reservoir existed and that
the registry slot was free, but not that the reservoir was empty, so a conversion
into an occupied reservoir succeeded and mixed inventory it never sampled into the
cohort's backing store. The reported case reproduces exactly: the ledger holds 67
picojoules against a cohort recording 60, so the cohort's reserve understated what
backed it and the surplus belonged to nobody. A destination holding anything is
now refused, a cohort's reserve is the whole of its reservoir rather than a part
of it, and the success test asserts that equality directly.
M4.3 advances a cohort over one declared interval (ADR-0038) and deliberately does
not model one. ECOLOGY_MODEL keeps population maths an experiment and asks for
several candidates to be compared, so choosing one here would turn a
recommendation into a locked rule. A model proposes fluxes; this checks that the
declared result is exactly what they imply, for counts and for every inventory
channel.
Nothing is clamped. A loss larger than what is there refuses the interval and
names whether it was the population or the biomass, so a caller can subdivide and
retry, and subtraction is checked against what is present rather than performed
and repaired. That is the answer to the milestone's named risk: a clamp is how an
interval a model could not integrate becomes a plausible looking result, and how a
false extinction arrives. Counts carry whole members and owed fractions together,
so a population below one member is a number rather than a rounding decision.
Biomass has no birth term at all, so a birth cannot bring inventory in and
newborns are necessarily funded from what the cohort already holds; immigration
draws from a reservoir that already held it, so a late transfer is not new
production either. Conservation runs through the ledger as one atomic transaction,
and an interval in which nothing moves submits none.
Two amendments to M4.1's record were forced rather than chosen. A cohort now
records the reservoir backing it, because an advance cannot otherwise know where
the inventory is and passing it alongside would let a caller advance one cohort
against another's store, which comparing balances cannot detect between equal
reserves; the registry refuses two cohorts sharing a reservoir. And a cohort may
be vacant, through a separate factory so vacancy stays deliberate, because a
population declining to exactly zero would otherwise be unrepresentable and the
caller would have to clamp. A vacant cohort is not an extinction: the record, its
genome, its habitat and its provenance all remain and it stays installed.
Limits are stated rather than implied covered. Accounting is not fidelity: closure
is checked, stability is not, so a long interval whose arithmetic closes and whose
ecology is nonsense is accepted. The declared age structure is not verified to be
what ageing would produce, because that needs the within-class distribution the
record deliberately discards. Emigration escrows into a reservoir nothing settles
with, since `TransferPopulation` does not exist. All fourteen guards, including
both clamp mutations, were confirmed load-bearing by removing each and observing a
failing run. Native Release, analysis, ASan and formatting passed 36/36; Linux CI
has not run yet.
M4.4 turns whole members of a cohort back into identified individuals (ADR-0039),
and does nothing else: no transfer between habitats, no persistence and no
resolution-equivalence claim, which is M4.5. Materialization is instantaneous, at
the cohort's current tick, because advancing time is M4.3's interval.
The named risk is that re-materialization becomes free healing, and it has two
separable sources. The arithmetic one is closed outright: a drawn member receives
its class share by integer division rounded down, because rounding up would create
biomass, and a class drawn empty hands over everything it held because a class with
no members may keep no reserve. A residual is never drawable, so an expected
fractional density cannot become an individual. Age is taken as the oldest the
class admits rather than the youngest or a seeded choice within it: the exact age
is unrecoverable, so the question is which direction to be wrong in, and every
other answer lets an adversary who picks when to convert roll back towards youth.
A round trip can therefore only age someone, and oscillation is strictly lossy.
The second source is the declared resolution itself and is stated rather than
mitigated. Members inside one class are indistinguishable by construction, so
collapsing a starving member with a well-provisioned one and materializing both
returns two average members; the starving one has been healed at the other's
expense. The tests demonstrate both halves of this rather than only the flattering
one: a round trip is exact when each class holds one member, and a coarse class
conserves the total while averaging the individuals. The mitigation is the
caller's choice of classes when it collapses, not anything this code can recover.
Materialization produces exactly the record a collapse consumes, so the two
compose and are tested against each other instead of being described as inverses.
Eleven guards were confirmed load-bearing by removing each and observing a failing
run, including the two that are the risk itself: rounding shares up, and drawing
ages from the bottom of a class. Native Release, analysis, ASan and formatting
passed 37/37; Linux CI has not run yet.
Review of `380d609` found two defects, both fixed with regressions confirmed to
fail without them. Materialization treated an empty reservoir as an available one,
so it credited a store that another installed cohort owned while leaving that
cohort's record unchanged: the same inventory in two places, reached from the
opposite direction to the M4.2 defect. Destinations are now checked against the
registry's ownership as well as their balance, on both the collapse and the
materialize side. The collapse side had the worse form of it: the shared-reservoir
rule would have refused the install only after the transfer had run, stranding the
members' biomass in a reservoir nothing accounted for, so whichever registry step
will follow a transfer is now proved to succeed before it.
The second was a functional gap the oscillation test had been hiding by discarding
its registry each 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, making a full materialization a one-way door out of a habitat.
Collapse now repopulates a vacant record in place, continuing its ownership and
revision series rather than restarting them, and refuses a record that still holds
a population. The test keeps one registry throughout and asserts the revision
continues.
M4.5 closes M4 by measuring what aggregation costs (ADR-0040). It adds no library
code: one rule is written once and applied at both resolutions, and a
one-member-per-class control proves the two runs are the same rule before any
divergence is attributed to resolution. Without that control a measured gap could
equally be a transcription mistake, and nothing afterwards would separate them; a
deliberate one-unit drift in the aggregate rule is caught by it.
The result is reported rather than smoothed into a tolerance. At class widths of ten
and fifty the aggregate population survives ten thousand ticks and tracks the
individual run within 375 parts per thousand on population and 43 on energy. At
widths of twenty-five, one hundred and two hundred it dies out entirely while the
individual run sustains seventy-five members: a false extinction caused by
resolution alone, because a whole class starves the moment its average cannot pay
upkeep where individuals would have starved a few at a time and the survivors would
have kept the shared per-capita share up.
The collapse is not monotone in class width. Twenty-five destroys the population
while fifty sustains it, so a resolution cannot be chosen by assuming degradation
worsens smoothly as classes widen, and a width that works is no evidence that every
narrower width works. The tests assert that non-monotonicity, because it is
surprising and because the warning written around it would otherwise be able to
become wrong silently. They also assert the runs disagree, so equivalence can never
be read as identity, and the word is not used for what was measured.
Conservation was exact at every tick on both sides of every run, including the three
whose populations died out. Four of five runs conserve perfectly and three are
ecologically worthless, which is exactly ECOLOGY_MODEL's point that conservation is
insufficient evidence.
The comparison also found a real defect in M4.3, which is what a validation
milestone is for. `plan_advance` checked affordability with arrivals added before
losses were taken, but emitted the ledger legs with the outflows first, so an
interval that was affordable as a whole underflowed leg by leg and was refused at
commit. A fifty-tick interval shedding 10,480 against 6,950 held while drawing
13,080 in reproduced it. Arrivals are now the first leg, matching the order the
arithmetic was checked in, with a regression covering it. Native Release, analysis,
ASan and formatting passed 38/38; Linux CI has not run yet.
## M4.6 - aggregate budget review (2026-09-26)
Completed test-only shared model, CPU/retained-memory benchmark, and 15-scenario
scarcity sensitivity report. The partial `tests/fixtures/resolution_model.hpp`
was integrated with existing M4.5 controls, including contested parent allocation.
No production library, simulation format, deployment or gameplay rule changed.
See [the report](engineering/AGGREGATE_BUDGET.md) and ADR-0041. None of the five
widths passes the proposed fidelity gate across supplies 240/300/360; automatic tier
selection remains disabled. The next implementation milestone is M5's sparse voxel
prototype; aggregate-model improvement and Linux/target-host measurement remain
prerequisites for a production tier policy. See VALIDATION.md for checks actually run.
## M5-M9 planning expansion - 2026-09-26
[ROADMAP.md](ROADMAP.md) now breaks M5-M9 into 42 reviewable tasks, each with
module boundaries, dependencies, completion criteria, tests and risks. These are
planned gates, not implementation or deployment claims. M8 dev/stage recovery is
separate from World0 launch; M9 client integration can use dev/stage without
premature World0 promotion. M4.6's failed fidelity gate remains explicit.
That expansion identified 5.1 as the next task; it is now implemented as recorded below. No library, service, cluster or dependency changed in this update.
## M5.1 - bounded local voxel contract (2026-09-26)
Implemented dependency-free `evo::world` coordinate and cell-sample utilities,
[ADR-0042](adr/0042-local-voxel-contract.md), and
[LOCAL_VOXEL_CONTRACT.md](world/LOCAL_VOXEL_CONTRACT.md). Signed floor mapping handles
negative chunk seams; all public operations reject invalid bounds before arithmetic.
The experiment uses a 1m cell pitch and explicit edge candidates 8/16/32/64; it
selects no storage layout, renderer, spherical topology or production world extent.
Logical fill describes volume fraction, not surface shape or consumable inventory.
Tests independently enumerate chunk intervals, exhaust 299,520 local ordinals, check
metric/domain corners and hostile signed/unsigned extrema, and validate fill/material
consistency. Existing simulation goldens/formats are unchanged. No world data or
service is deployed. Next: M5.2's measured layout/edge comparison; actual chunk
allocation and generator work follow its decision. See VALIDATION.md for checks.
## M5.2 compact layout evidence
Compared three representations, four edges and five fixtures with seeded edit/oracle
checks. [Evaluation](engineering/CHUNK_LAYOUT_EVALUATION.md) records memory, costs,
encoding estimates and limits. ADR-0043 selects 16-cube uniform/palette/dense geometry.
No production container, generator, codec or deployment was added. Next: M5.3.
## M5.3 deterministic abiotic base
Added pure seed/address/version generation, a bounded inert material table and
16-cube logical output. [Contract](world/BASE_GENERATION.md) specifies rejection,
compatibility and storage boundaries. Twelve independent reference goldens and
signed seam/query-order tests freeze v1. No gameplay, persistence or deployment
was added. Next: M5.4 sparse active chunk ownership.
## M5.4 bounded active chunk ownership
Added fixed-slot, budgeted ChunkStore and immutable uniform/palette/dense CompactCells.
Revision tokens reject stale and foreign reads. Clean chunks regenerate; dirty pins
refuse eviction. [Ownership contract](world/CHUNK_OWNERSHIP.md) documents memory/lifetime
limits and the absence of cell edits or durability. Next: M5.5 accounted terrain edits.
## M5.5 transactional terrain excavation
Added up to 64 cuts across four loaded chunks, atomically transferring occupied
volume into finite per-material debris. Revision/digest/full-payload retry checks,
64-command retention and peak allocation backpressure are explicit. The
[edit contract](world/TERRAIN_EDITS.md) records excavation-only scope and persistence
obligations. No mass conversion or durable acknowledgement is claimed. Next: M5.6.
## M5.6 snapshots and edit journals
Added bounded canonical world snapshot v1 and edit journal v1, world/frame and base
version checks, full geometry/debris/retry restore, semantic validation and detached
all-or-nothing replay. [Codec contract](world/WORLD_CODEC.md) and schemas define
memory/trust limits. The M5.6 codec alone does not publish files or issue durable
acknowledgements. M5.7 local publication and recovery are implemented above.
## M5.7 local publication and recovery
ADR-0048 adds a single-writer local-filesystem adapter over the M5.6 codec. It uses
immutable checksum-named snapshot/journal blobs and two alternating canonical
manifest slots; replacing the inactive manifest is the publication point. Checkpoints
compact a committed journal only after a caller supplies the exact recovered
state. Recovery validates the selected snapshot and complete journal replay into a
detached ChunkStore. Tests cover lock exclusion, journal/restart, compaction, failpoints
before and after publication, malformed manifests and missing published payloads.
FNV remains a corruption detector, and this local adapter makes no host-loss,
multi-region or production durability claim. Filesystem orphan cleanup is deferred.
## M5.8 bounded world restart proof
Added evo-world-proof and a CTest workflow that launches fresh prepare, compact and
verify processes for uniform, sparse and heavily modified deep chunks. Independent
cell/debris oracles, exact canonical equality, frozen hashes, retained retries and
clean regeneration checks pass. [Measured costs and reproduction](engineering/WORLD_PROOF.md)
record five Windows Release runs and explicit accounting/durability limits.
No world format or simulation version changed. Next: M6.1 environmental units and
coupling contract; Linux golden execution remains a CI confirmation.
## M6.1 environmental units and habitat coupling (2026-09-27)
ADR-0050 and [the units contract](world/ENVIRONMENTAL_UNITS.md) define bounded
inventories, compartment identity, derived concentration/moisture, explicit habitat
depth and residual ownership. Pure checked helpers measure one-voxel habitat boxes,
form rational views and integrate energy-rate proposals without creating supply.
Dimensional examples, invalid conversions, overflow, split intervals, closed/open
accounting and refusal rollback are tested. Windows Release, clang-tidy and ASan
suites pass; see VALIDATION.md for exact checks. Existing formats/goldens are unchanged.
No fields or biological coupling are implemented yet. Next: M6.2 optional finite
environmental fields with versioned persistence and explicit opening inventories.
## M6.2 optional finite environmental fields (2026-09-27)
Added a bounded FieldStore with fixed habitat mappings, material water-capacity
defaults, optional scalar channels and uniform baselines with sparse 64-site pages.
Opening inventories are explicit; missing differs from zero. Same-channel local
transfers preserve finite supply and reject stale commands/overspend without change.
Field snapshot v1 preserves mapping, quantities, defaults, presence and watermark;
restore does not regenerate supplies. [Contract and memory evidence](world/ENVIRONMENTAL_FIELDS.md)
and the byte schema document trust, budget and compatibility limits. M5 terrain
formats remain unchanged; combined terrain/biology durability and rate sources are
not implemented. Next: M6.3 read-only derived environment views.
## M6.3 read-only environment views (2026-09-27)
Added [bounded footprint queries](world/ENVIRONMENT_VIEWS.md) under ADR-0052.
Views copy typed quantities, exact concentration/moisture and terrain samples with
habitat resolution, logical tick, field watermark and sampled chunk revisions.
Missing fields are distinct from zero; absent habitats/chunks fail without loading.
Freshness validation rejects wrong stores, changed tick/field state and stale chunk
tokens, including eviction/reload. No persistent format or simulation rule changed.
Views grant no inventory reservation; coherent tick/store pairing remains a host
responsibility. Next: M6.4 coordinated environmental transfers.
## M6.4 atomic environmental transfers (2026-09-27)
Implemented the [transfer contract](world/ENVIRONMENTAL_TRANSFERS.md) and ADR-0053.
A single owner stages and publishes fields and tagged organism resource accounts
atomically, validates view freshness and ownership, rejects capacity/stock/budget
violations, and retains 64 exact request receipts. Tests cover contention, late-leg
rollback, changed retries, independent compartment balances and conservation.
This is in-memory ownership; no coupled recovery format or durable ACK is claimed.
No schema, simulation version, dependency or deployment changed. Next: M6.5 biology
adapters, including explicit conversion and active/alive validation.
## M6.5 prerequisite review (2026-09-27)
Implementation is pending resolution of [ADR-0054](adr/0054-m65-source-persistence-prerequisite.md).
ADR-0051 requires persisted source rate/residual/interval state before rate execution;
ADR-0053 defers the coupled format until M6.8. Recommended resolution: bring the
minimal coupled buffer codec forward, preserving the earlier guarantee. The proposal
breaks M6.5 into codec, explicit biological conversion contract, rate-driven fixture,
and qualification segments. No source-rate code or public interface changed in this
review; M6.5 remains incomplete. M6.4's validated behavior is unchanged.
## Accepted M6.5 sequencing and M6.5a codec (2026-09-27)
The maintainer accepted ADR-0054: bring coupled snapshot/restore forward. This
supersedes the pending-decision status above. M6.5a now implements
[resource snapshot v1](../schemas/coupled-resources.md): terrain, finite fields,
organism accounts/capacities, adoption watermark and chronological full retry history
restore as one validated owner. Restore audits retained transfers backwards against
current balances and validates nested codecs and aggregate ownership/budgets.
Next: M6.5b explicit capture/growth/feeding conversion contract, followed by M6.5c's
complete source-rate/biology envelope and M6.5d qualification. No rates are enabled
by the resource codec; no source progress is invented during restore. M6.5 as a
whole remains incomplete. Durable publication and region recovery remain M7.
## M6.5b process contract (2026-09-27)
Completed the design segment under ADR-0055 and
[ENVIRONMENTAL_PROCESS.md](biology/ENVIRONMENTAL_PROCESS.md). The contract defines
finite area-bounded capture, explicit nutrient incorporation, retained water,
organic feeding with heat/waste, exact source/growth residual policies and the
process-owned checkpoint/retry boundary. Hand-calculated vectors make each flow
reviewable. No new runtime code, executable rate integration or schema changed.
Next: M6.5c implements the fixture and complete source-rate/biology snapshot before
exposing advance; M6.5d qualifies it. M6.5 remains incomplete. M2/M3 maintenance,
reproduction and mortality behavior is unchanged; resource starvation is reported
without inventing a new death policy.
## M6.5c persisted environmental fixture (2026-09-27)
Implemented EnvironmentalCommunity: area-bounded finite capture, recipe growth,
organic feeding/waste/heat, source exhaustion and temperature gates, plus whole-tick
rollback and exact retained retries. Rates execute only alongside the complete
[process envelope](../schemas/environmental-community.md), which restores primitive
config, residuals, lifecycle, resources and history and audits retained inputs by
replay from a historical anchor. Existing M2/M3 rules and codecs are unchanged.
The pasted GCC 14 CI log exposed a vector reserve/append diagnostic in the resource
snapshot writer. Both coupled writers now fill exactly sized buffers and check their
final cursor, retaining strict compiler warnings and unchanged resource wire bytes.
Native validation is recorded separately from Linux CI verification.
Next: M6.5d broader qualification (boundary/adversarial cases, long-run seeds and
cost evidence). M6.5 as a whole is not yet marked complete; no deployment changed.
## M6.5d native qualification (2026-09-27)
M6.5a-d are implemented and qualified natively. Added boundary/malformed-state
coverage and a deterministic long-run/cost harness. The extended run compared
1048576 logical ticks against continuous controls, including 329231 feeding
requests; balances and restored bytes matched. See the
[qualification report](engineering/ENVIRONMENTAL_QUALIFICATION.md) for evidence and
limits. Native Release/analysis/ASan suites passed 54/54; Linux CI remains a separate
required delivery check, not an observed pass.
Next: M6.6 terminal events, retained corpse biomass and bounded decomposition.
No maintenance, reproduction, player progression, simulation version or deployment
changed in this qualification segment.
M6.6 now retains corpse resources and lifecycle in opt-in rules/envelope v2.
Partial bounded decay conserves matter and energy, restores with exact retry
history, and can supply the surviving producer on a later tick. V1 compatibility
is retained. See ADR-0056 and VALIDATION.md for scope and evidence. This does not
add mortality causes, decomposer kinetics, durable publication or player sessions.
M6.7 now schedules bounded explicit local intervals, checks stale/foreign boundaries
and reports committed prefixes on failed catch-up. Pause/restore/resume uses the
existing process envelope; no wall-clock debt is inferred. M6.8 is next. See
architecture/ENVIRONMENTAL_SCHEDULING.md for caller responsibilities.
M6.8 completed the bounded environmental integration qualification. The extended
ensemble passed 5120 episodes across five regimes, including nutrient-fed regrowth,
conservation and direct-versus-restored equivalence. See
engineering/ENVIRONMENTAL_SLICE_QUALIFICATION.md for actual measurements and
limitations. M6 is natively qualified; Linux CI confirmation remains separate.
Next is M7.1: define complete region state, durable ACKs and the crash-point matrix.
M7.1 durability design is complete. ADR-0058 chooses full checkpoint batches,
atomic permanent outcomes and exact reconciliation after ambiguous commits.
REGION_DURABILITY.md inventories included/missing state and the crash-point tests.
No PostgreSQL connection, persistence adapter or region pod exists yet. M7.2 is
next; lineage closure remains an explicit M7.4/M7.9 prerequisite, not waived scope.
M7.2 complete (2026-09-28): optional Linux libpq adapter and PostgreSQL 18.6
source pin, transactional checksum-checked bootstrap, bounded parameterized I/O,
private database integration tests and Secret-mounted migration Job example.
Release, clang-tidy and ASan/UBSan PostgreSQL checks passed in WSL; Windows
headless suite passed 54/54. Docker/remote CI and cluster deployment are not
claimed. See infrastructure/POSTGRESQL.md and engineering/VALIDATION.md.
Next: M7.3 fenced ownership and narrowly granted database functions. The bootstrap
lease table alone does not establish authority or durable region publication.
## M7.3 fenced region ownership (2026-09-28)
ADR-0060 and migration v2 add guarded lease and head functions for a runtime role
that has no table rights. The portable `LeaseAuthority` fences locally from
send-time deadlines on CLOCK_BOOTTIME. Selected timings are 20 s lease, 5 s renew,
2 s margin, 2 s renew deadline and 5 s commit deadline. Local round trips took
about 0.6-0.7 ms p50 and about 1 ms p99. Real PostgreSQL covered contenders,
expiry, paused and queued stale writers, unknown renewal, exhaustion and bypass
denial. Three SQL mutations (`now()` instead of `clock_timestamp()`, a missing
lease check, a disabled busy check) were each detected by a failing test.
Windows ran 55/55; Linux Release, clang-tidy and ASan/UBSan PostgreSQL suites
passed in WSL. No director, host, gateway, payload publication or deployment
exists yet. The v2 head must not carry real world state before M7.4. Next: M7.4.
## M7.4a checkpoint publication (2026-09-28)
ADR-0061 and schema v3 publish environmental-v1 checkpoints. A fixed 280-byte
manifest is parsed and digested by PostgreSQL. Payloads live in an immutable,
content-addressed filesystem store and are hashed with OpenSSL SHA-256. The
database enforces the commit chain, retained builds and fixed versions/quantum,
and refuses to migrate while v2 heads exist.
Recovery re-digests the manifest, checks the build allowlist and fully restores the
payload. A restored state continues identically to the uninterrupted original.
Missing, corrupt or tampered data quarantines the region. Four SQL mutations and
one host mutation were each caught by a test.
Local fixture payloads are 23-37 KiB and publish in about 3.5-10.6 ms across
builds. Windows ran 56/56; Linux Release, clang-tidy and ASan/UBSan PostgreSQL
suites passed in WSL. M7.4 stays open for M7.4b (lineage archive codec and
history-capable binding). No real world was provisioned; there is no deployment,
host or process-kill test yet.
## M7.4b lineage checkpoint profile; M7.4 complete (2026-09-28)
ADR-0062 adds a canonical `LineageArchive` codec that decodes by replaying the
archive's own admission, birth and death rules. It also adds `LineageRegion`,
which:
- binds the two environmental accounts to archived organisms;
- keeps identity high-waters;
- enforces closure over lifecycle, the living set and ticks;
- records deaths in the tick that causes them.
Checkpoint profile 2 (`environmental-lineage-v1`) and schema v4 publish and recover
it. A region can never switch profile. Recovery restores retained genomes,
ancestry and dead organisms, and continues identically. A full archive encodes to
2,543,648 of at most 3,219,496 bytes.
While qualifying, WSL2 was measured stepping the wall clock back about 0.5 s every
~32 s. Four database-time tests now wait on the database's own expiry judgement
instead of assuming a local sleep equals database time; ADR-0060 records the
deployment requirement. Windows ran 57/57. Linux GCC Debug/Release, clang-tidy
and ASan/UBSan CTest each ran 57/57. The PostgreSQL suites passed 2/2/5 times.
Next: M7.5 durable decisions, history and outbox rows, and commit-ID resolution.
Retained ancestry still needs M7.5 history and M7.7/M7.9 kill-recovery evidence.
## M7.5 durable batch outcomes (2026-09-28)
ADR-0063 introduces manifest v2, which binds a per-tick validation trace object
and a decision block.
- Staging replays the trace from the committed predecessor's verified payload and
must reproduce every report and the candidate bytes.
- `RegionBatch` seals at a death or at 64 ticks and refuses ticks until the commit
is confirmed.
- Schema v5 records each checkpoint, its deaths and their outbox rows in one
statement:
- one terminal outcome per actor;
- committed history is immutable;
- exact retries return the original receipt;
- read-only resolution answers committed, absent or conflict.
- Recovery reconciles durable deaths with the restored lifecycle.
Real-PostgreSQL tests cover a lost response, a timeout that later committed, a
terminated orphan that was retried, changed bytes, repeated or partial or
reordered decisions, gaps, outbox redelivery, and history disagreement. Five
mutations were each caught. Windows and Linux ran 58/58; PostgreSQL suites passed
under Release, analysis and ASan.
Next: M7.6 region host. No host, broker, deployment or process-kill test exists yet.
## M7.6 region host and server process (2026-09-28)
ADR-0064 adds `RegionHost`, a single-owner, single-threaded state machine:
acquire and recover, run, resolve, drain and stop, or quarantine.
- Admission is provisional and bounded (64 inputs, 64 KiB) with backpressure.
- Acknowledgements and outcomes are final only after a confirmed commit.
- A death pauses the region and commits in the same step.
- Tick debt is measured, and catch-up is bounded to 8 ticks per step.
- Readiness fails closed.
`evo-server` hosts one region using a stdin input feed and stdout event lines,
with a readiness file and SIGTERM drain. Real process tests:
- SIGTERM: 80 inputs through the 64-input backlog were drained, committed and
released.
- SIGKILL: every acknowledged tick was committed, and the successor became ready
about 3.9 s later, after the 4 s development lease expired, from the committed
head.
Seven host scenarios pass against real PostgreSQL, and four mutations were each
caught. Windows ran 58/58; the PostgreSQL suites passed under Release, analysis
and ASan/UBSan.
Next: M7.7 recovery verification at every kill boundary. There is no gateway,
metrics or deployment yet.
## M7.7 recovery before readiness (2026-09-28)
ADR-0065: before readiness, recovery now replays the head's validation trace
from the predecessor's verified payload. It must reproduce the head bytes
exactly and derive exactly the durable decisions. A missing, corrupt or
non-reproducing tail, or a missing predecessor payload, quarantines the region,
and nothing older or empty is ever served. Schema v6 adds the read-only
`load_region_commit`.
Real `evo-server` processes were SIGKILLed at five publication and ACK
boundaries using a test-only `--crash-at` injector. Every result matched the
crash matrix, and each successor verified the head before becoming ready. A
SIGSTOPped owner resumed fenced, and damaged heads exited quarantined.
Tail-skip and replay-equality mutations were each caught. The PostgreSQL suites
passed under Release, analysis and ASan/UBSan.
Next: M7.8 operations endpoints and a dev image.
## M7.8 operations endpoints and server image (2026-09-28)
ADR-0066: `evo-server --ops-listen` serves three endpoints:
- `/livez`: the loop is responding; it is never a statement of authority.
- `/readyz`: 200 only with the current lease and a verified recovered head.
- `/metrics`: 30 fixed series whose only labels are a 6-value phase and a 4-value
drop reason.
The responder is dependency-free, runs on the host's own loop with no thread, and
is bounded: 8 connections, 4 KiB requests, 1 s deadlines, GET on exact paths
only. The new `server-runtime` Docker target is non-root with no credentials,
covered by a CI smoke script, and a Kubernetes example ships outside Argo.
Real process probes covered serving authority, an unavailable database (live,
503) and a resumed stale process (live, 503, fence counted). Readiness-as-liveness
and an unbounded connection pool were each caught by a test. Windows ran 59/59;
Linux ran 59/59 in four configurations; the PostgreSQL suites passed in three.
Docker itself was not run locally; an emulated image layout was checked instead.
Next: M7.9.
## M7.9a region clones and retained history (2026-09-29)
ADR-0067 adds a consistent clone into an isolated world:
- **Re-identification:** `rebind_world` changes only the recorded world, and the
byte round trip is proven.
- **Schema v7 clone genesis:** it starts at the source tick, names the source
commit and digest, and digests the inherited outcomes.
- **`evo-world-clone clone`:** it reads the source read-only, requires a separate
database and user, and writes the target in one owner transaction.
`audit_region_history` (also `evo-world-clone audit`) re-verifies every retained
link.
Real processes showed:
- after re-identification, the clone and its source followed a byte-identical
trajectory;
- the clone cannot write its source;
- a killed clone left the source untouched and the target not ready;
- a missing historical trace is detected even when head recovery succeeds.
Five mutations were caught. Windows ran 59/59; Linux ran 59/59 in four
configurations; the PostgreSQL suites passed in three.
Next: M7.9b, covering the upgrade barrier and old-writer fencing, schema
migration failure while a host is serving, and measured recovery and batching
latency.
## M7.9b upgrade barrier and durability qualification (2026-09-29)
ADR-0068 and schema v8 bind authority to builds:
- Acquisition names the host's build, and only a retained build that has not been
retired may hold a lease.
- No commit may name a retired build.
- `upgrade_barrier` retires a build and fences every holder in one transaction.
- Hosts built before M7.9b keep working until the first barrier.
A lease-holding writer survived a failed v8 migration (the database stayed at v7
exactly) and then the real migration. After the barrier it could never regain
authority. The new build recovered the whole history, including links written by
the retired build.
Qualification (engineering/DURABILITY_QUALIFICATION.md), under the production
lease:
- SIGKILL to successor readiness: 19.5–19.6 s against the proposed 60 s target;
- acknowledgement latency with a 1 s flush: p50 549 ms, p99 1048 ms;
- a death was acknowledged in 44 ms;
- no acknowledged input was lost.
**M7 exit criteria are met locally.** Container, CI and cluster execution have not
run. Host-loss durability is M8.
Next: M8.1, the host inventory and budgets.
Review fixes (2026-09-29): schema v9 serializes build retirement with acquisition,
leases and provisioning under one lock order. It also serializes clone-world
classification per world. The operations listener's accept work per loop pass is
now bounded. Regression tests and five mutations cover each fix.
## M8.1 inventory (in progress, 2026-09-29)
docs/infrastructure/INVENTORY.md records:
- the maintainer's facts: an R730xd with about 128 GB, 32 GB worker nodes,
automatic publish to Harbor and automatic Argo CD deployment, and the existing
`aevolution-db` Secret. The Secret's single login conflicts with the required
18.6 pin and with the separate migrator/runtime roles;
- the measured workload profile: `evo-server` peaks at about 9.7 MiB and 0.3 % of
one core;
- a local storage baseline, and a proposed budget model.
Still needed:
- run the read-only `tools/inventory/collect-cluster.sh` with cluster access;
- hypervisor, VM and disk details from the maintainer;
- the storage probe on each candidate StorageClass, after approval.
M8.1 cluster inventory collected (2026-09-29, docs/infrastructure/INVENTORY.md):
- RKE2 v1.32.5: one control plane and three 32 GB workers on one hypervisor.
- `local-path` is the only storage (node-pinned, reclaim Delete).
- `aevolution-dev` syncs manually.
- AEvolution's PostgreSQL is a separate VM outside Kubernetes (`aevolution-db`),
so it is not in the worker budget.
- A proposed budget keeps AEvolution dev well under 1 GiB on one labeled worker
(not worker-3, whose disk is 87 % full).
M8.1 complete (2026-09-29) for the current disk layout. The per-worker
`local-path` probe measured:
- 1.9–2.5 ms for synchronous 4 KiB writes;
- 29–36 ms mean (73 ms max) per object publication;
- 110–255 MB/s sequential.
These are well inside the lease deadlines. Re-probe after the planned RAID and
storage change. Host memory was raised by about 500 GB.
Next: M8.2, environment isolation and secrets.
## M8.2 environment isolation (2026-09-29)
ADR-0069 adds three environments, each with its own guardrails, service accounts,
secret references and database:
- `evolution-dev`, `evolution-stage` and `evolution-world0`;
- guardrails: quota, limit range, default-deny networking, restricted pod
security.
A native ValidatingAdmissionPolicy denies deleting World0 data, identity or Argo
objects to everyone, including cluster-admin and Argo, unless the requester is in
`aevolution:world0-break-glass`. The stage and World0 Argo projects are scoped to
one namespace each, and World0 permanently denies automated sync. The cleanup tool
refuses World0.
The offline policy check (now a CI suite) passes, and nine injected violations
were each caught. Applying the layer and running `tools/ops/verify-isolation.sh`
is waiting on the maintainer.
M8.2 applied and verified live (2026-09-29): 175 PASS and 2 SKIP. The only 3 FAILs
were a script defect in the project check, now fixed. The maintainer accepted
leaving Argo's `default` project unrestricted. Break-glass membership is in the
new docs/BACKLOG.md. Next: M8.3, durable stores in dev and stage.
## M8.3 dev/stage stores (2026-09-29)
ADR-0070 sets up the dev and stage stores:
- **Database:** each environment gets its own database and logins on the external
PostgreSQL VM.
- **Object store:** the filesystem store stays, on a node-local PVC in the new
Retain class `aevolution-local-retain`, pinned to one labelled worker. There is
no object server.
- **Capacity:** node-local volumes ignore their requested size, so the object
store now enforces a free-space reserve (`--objects-reserve-mib`, default 1024).
A full volume refuses new objects with a distinct error, and nothing is
acknowledged.
- **Qualification:** `evo-world-clone check-store`, now in the server image,
checks from inside the cluster: volume presence and retention, the reserve,
database version and TLS, and the runtime privilege boundary.
Not yet done:
- the in-cluster run (it needs Secrets with the documented key names and
published image digests);
- choosing a backup destination (docs/BACKLOG.md).
M8.3 delivery moved to Argo CD, at the maintainer's request:
- **Platform app:** `aevolution-platform` (self-heal, never prune, `Delete=false`,
protected by admission) owns the namespaces, guardrails, storage class, object
claims and World0 policy.
- **Store Jobs:** they become sync hooks in each environment's path once their
digests and runtime roles are supplied.
- **Outside Argo:** Secrets and the store node label.
## M8.4 region host (2026-09-29)
The maintainer synced the dev, stage and platform applications through Argo CD,
and they are connected to PostgreSQL, with digests `661e433e…` (server) and
`9abb06ef…` (tools).
ADR-0071 deploys `evo-region` as one replica per environment:
- the Recreate strategy, since a rolling update would deadlock on the lease;
- readiness `/readyz`, liveness `/livez`;
- no CPU limit;
- a 20 s drain inside a 45 s grace period.
Each sync runs, in order: egress, migration, provisioning (idempotent, from the
new `development_opening`, which is byte-identical to the qualification fixture),
the host, then the store check. The image digest is also the build identity, and
the policy check enforces this. Dev is world 100 and stage world 200, both region 1.
Next: apply `deploy/argocd/dev/project.yaml`, sync, and run
`tools/ops/region-drill.sh dev`.
## M8.5 observability (2026-09-30)
ADR-0072 adds an AEvolution-only monitoring stack in `aevolution-monitoring`:
- **Components:** Prometheus 3.13.3 and Alertmanager 0.34.1, pinned by digest and
delivered through Argo.
- **Scraping:** static targets, so no Kubernetes API access. The region's port
9100 is reachable only from Prometheus.
- **Alerts:** 16 rules covering authority, lease, commits, backlog and object
capacity, plus a Watchdog. Each is proven firing and silent by promtool
injection tests in CI. They route to a null receiver until a destination is
chosen.
- **Metrics:** two new capacity gauges (32 fixed series).
No broker is adopted: the database outbox stays. Logs are bounded by kubelet
rotation. Monitoring costs about 1,300 series, roughly 11 MB a day.
Next: sync the monitoring app, confirm the targets, and run the `EvoRegionDown`
injection. M8.4's drill results are still to be recorded.
## M8.6 fault drills (2026-09-30)
ADR-0073 adds a drill-only synthetic driver, `evo-server --synthetic-inputs
development-v1`:
- it feeds deterministic inputs from the committed head after every recovery,
including the tick-20 death;
- its request IDs are `SYN1‖region‖tick`;
- it is off in Git, and the policy check enforces that.
`tools/ops/fault-drill.sh` puts dev or stage under load and runs these scenarios:
- **kills:** three forced kills, measuring RTO and checking that no acknowledgement
is lost;
- **full:** a full volume, which must acknowledge nothing;
- **empty:** an empty volume, which must quarantine and never start a fresh world;
- **db-outage:** an operator-blocked database, after which a stale epoch never
resumes.
It restores the Git configuration on exit.
Earlier (M8.4 follow-up): `REGION_RETAINED` keeps earlier builds recoverable
across image upgrades.
## M8.7 backup and restore (2026-09-30)
ADR-0074 adds world backup bundles and a verified restore:
- `tools/backup/backup-world.sh`: an online backup cut by one exported database
snapshot (MANIFEST heads and `pg_dump` read the same snapshot), then the
immutable objects, each checked against its digest; `SHA256SUMS` over all of it.
- `tools/backup/restore-world.sh`: checksums first, then an owner-only restore into
an empty database, exact head comparison and a full history audit of every region;
any failure removes what it restored.
- `tools/backup/verify-backup.sh`: the restore on a private throwaway cluster with
fresh roles, then (`--isolate WORLD`) a clone into an isolated world that serves
under synthetic load and drains.
- `tools/backup/cluster-backup.sh <dev|stage> --to DIR`: an operator Pod (read-only
objects volume, migrator login) whose bundle is streamed off the cluster.
- The postgres-tools image now carries `pg_dump` and `pg_restore` from the pinned
18.6 build; the database egress policy admits the `backup` component.
The destination, encryption, schedule, retention and WAL archiving wait on the
maintainer (docs/BACKLOG.md).
## M8.8 World0 gate and host-loss recovery (2026-09-30)
ADR-0075:
- **World0 launch is blocked.** Everything runs on one physical host, so an
acknowledged death could be undone by host loss. The deploy policy check now
rejects any World0 workload, a resolved World0 Application path, and World0
overlays; any future World0 pin must equal stage's digest.
- **Dev/stage objectives ratified:** worker loss RPO 0 and RTO <= 60 s (measured
~20 s); host loss rewinds to the last verified bundle, explicitly.
- `restore-world.sh --in-place --runtime-role ROLE --accept-rewind` restores a world
under its own identity with its grants, fences restored leases, audits runtime
rights and prints the rewind.
- `fence-database.sh` supersedes the old database through the upgrade barrier: a
still-running old host quarantines, and no host can acquire it again.
- `tools/ci/dr-check.sh` rehearses it in every PostgreSQL build.
## M8.9 upgrade and growth (2026-09-30)
ADR-0076:
- **Compare before upgrading.** `evo-server --synthetic-until TICK --accelerate N`
(drill-only, refused without the synthetic driver) runs clones to an exact tick
faster than real time; `evo-world-clone compare` (library:
`compare_region_histories`) audits both and requires the same world: same
versions, identical rebound head state and identical outcomes. Builds may differ.
- **Failed migration and barrier rehearsed** with real hosts
(`tools/ci/upgrade-check.sh`): a failing migration leaves schema v9 and the host
serving; the barrier fences the old build (exit 4), the new one resumes under a
newer epoch, and the history across both builds audits.
- **Dev/stage barrier:** `tools/ops/upgrade-barrier.sh <env> --retire HEX` retires an
old build only when it is in `REGION_RETAINED` and the new digest is already Ready.
- **Nodes:** `aevolution.io/stores` (one worker; every objects-volume pod selects it)
and `aevolution.io/compute` (preferred only; Alertmanager first).
`tools/ops/scheduling-check.sh <env>` verifies placement read-only.