M11.5c (ADR-0125): v10 cellular actions, the canonical input and the trace carry
target kind 3 (bud), matched to durable actions; below rules 4 the owner
ignores it. The v5 version bundle is admitted in C++ and by schema v36, which
also admits kind 3 in SQL action inputs; the provisioner offers cellular-v3.
A PostgreSQL test publishes a bud from a fed founder and recovers it after a
host restart. The publication tests now use commit ids no other suite produces.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ADR-0124: checkpoints pin cellular bundle v4 (rules 3, owner envelope 4,
revision archive) with the revision history profile and format-3 traces.
A format-4/rules-3 opening provisions on v4, and evo-world-clone accepts
--opening cellular-v2 for it. Migration 34 widens canonical action input
validation to target kind 2 (engulf, no coordinates), admits v4 manifests
through the existing v3 parser chain, and lets birth binding read
format-3 traces for v3 and v4. The migration runner and its checks extend
to version 34, with SQL and provisioning tests for v4.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ADR-0124: byte 12 of the canonical cellular input, the CellularAction
wire message and the batch trace record becomes a target kind (0 none,
1 patch, 2 engulf with zero coordinates). Kind 2 exists only on v10,
which no server admitted before; cellular servers now admit v10 with v9
behaviour plus engulf, and still refuse a v10 mutation choice until
11.4c. Admission, the session-batch adapter, trace codec, matcher and
action-block validator carry the kind.
The client session library accepts v10 and an engulf flag, and its send
no longer reports a message its version cannot encode as sent.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ADR-0124's engulf profile starts a region on owner format 4 and rules 3,
with the revision-format archive. Ticks pass genome membrane thickness and
source to the region step, record each corpse's cause and predator, insert
detritus patches and validate rules-3 corpses as empty. Format-4 snapshots
carry the cause and predator and refuse an unknown cause, a predator on a
starvation corpse or a non-empty corpse. Mutation selection accepts both
genome-revision formats (3 and 4).
Persistence (version bundle, schema 34), protocol v11 and provisioning
follow.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ADR-0124 implements ADR-0098's predation and corpse rules for the
cellular trial. Under rules 3 the pure region step adds an engulf phase
after every life's step: in ascending life id, a predator takes the
lowest-id contacted survivor whose total mass, times 1.5 per micrometre
of its membrane, is at most the predator's. The prey's whole inventory
moves once; the predator keeps what its remaining per-tick intake allows
and the rest, like every starvation corpse, becomes detritus in the
patch at the dead cell's 1 mm cell. Corpse candidates report their cause.
Rules 2 is unchanged and ignores engulf intents.
Owner, protocol v11, persistence and recovery wiring follow in the next
11.3f steps.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The 2026-10-13 probe in DEV region 100/3 showed steering and feeding
working end to end after the session fixes, and not yet enough game to
judge fun. Record its findings and inputs for 11.4-11.8, and mark the
remaining 11.3 work (engulfment and corpse absorption) in the roadmap.
Also refresh the dev session window (3 h).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Promote server image 5e13e009 (Gitea run 365, source e9b010e): cellular
actions admitted every tick, births ticked only for held accounts, and
life outcomes valid on the wire. Retain faeb26fc, which wrote region 3's
genesis. Also refresh the dev session window (3 h).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A fresh client launch sends a new birth request. For an account that
already has a cell, the cellular host reports existing_life with the
reserved identity, and RegionSessions forwarded it although the codec
accepts an identity only with born. The client closed with protocol and
retried with the next grant until the account was rate-limited, so no
account with a cell could log in from a new launch. Clear the identity
and commit on every other result at the session boundary, and test that
an existing_life refusal arrives valid while the session stays open.
The client now names the cause of a protocol close on stderr, which is
how this was found. Record it in ADR-0123. Also refresh the dev session
window (3 h).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A second account's queued birth outlived its session. Every following
batch failed check_life_control with EVL02, which the host reads as a
lost region lease, so region 100/3 fenced, reacquired and failed again
every 20 s, refusing logins. Include, and tick for, only births whose
account holds a session; the rest stay queued until the account returns.
Record the incident in ADR-0123. Also refresh the dev session window (3 h).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The first playable run showed two server faults. The client sends one
cellular action per tick (an idle tick carries no movement or intake),
but the default probe admission allowed 4 per 8-tick window, refusing
every other one; the cellular bridge now admits two per tick of window.
Region 2's opening starved the cell: only patch energy feeds the reserve,
and founder intake emptied its 4M and 16M pJ patches in seconds. DEV
moves to region 100/3 with 200M and 50M pJ patches (ADR-0123).
Also refresh the dev session window (3 h).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The camera started at 0.05 px/mm, the metre-scale lineage default, so the
whole 10 mm cellular arena drew inside half a pixel and the player could
not find their cell. In the cellular trial, frame about 12 mm around the
own cell (organism id == life identity) once, then keep it centred while
keeping the player's zoom. Draw food patches as their true 1 mm footprint
under the organisms, so contact with a patch is visible.
Also refresh the dev session window (3 h).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Promote server image faeb26fc (Gitea run 360, source 5e45be3), whose
provisioner resolves a cellular genesis under its recorded build, and
retain 1dd640eb. Also refresh the dev session window (3 h).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Promoting the dev server image changed --build, and the cellular
provision rerun re-encoded the genesis manifest with the new build, so it
refused region 100/2 as a different genesis. The check could also never
pass once the head moved past genesis. Match an unadvanced cellular
genesis under its recorded build, and accept profile and frame past
genesis as the other openings do. A different commit or opening is still
refused. Add a PostgreSQL process test and record the finding in ADR-0123.
Also refresh the dev session window (3 h).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Promote server image 1dd640eb (Gitea run 358, source 56e7525), which
adds evo-world-clone provision --life-allocator-start (ADR-0123), and
retain the previous build 566c2932 that wrote region 2's genesis. The
provision Job now creates world 100's life allocator at
1,000,000,000,000. Also refresh the dev session window (3 h).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
clang-tidy's bugprone-unchecked-optional-access could not follow the
optional cell through the indexed row in the provisioner and its test,
failing the PostgreSQL analysis build (run 357). Read the cell through a
checked reference instead. Also refresh the dev session window (3 h).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The DEV cellular region reported every first-life request as unknown:
world 100 had no life_allocators row, and life-reservations-v1 leaves
that row to the owner with no runtime initializer. evo-world-clone
provision now accepts --life-allocator-start N with the primitive-v1
and cellular-v1 openings. It creates the allocator once before the
genesis check, leaves it untouched on reruns, and refuses a start above
the existing counter.
ADR-0123 records the DEV cellular region bring-up: region 100/2, schema
v33, the trial opening, the allocator start and the 300 s grant ceiling.
Also refresh the dev session window (3 h).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The cellular-v1 provision hook failed with PublicationError::rejected:
the dev store Jobs still pinned the schema-v11 postgres-tools image,
so PostgreSQL refused the profile-4 genesis manifest. Pin the
postgres-tools image from Gitea run 350 (source ae3b2b4, migrations
through 0033), the same run that built the promoted dev server image.
Also refresh the dev session window (3 h).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The dev overlay now serves region 2, so the selftest's literal
REGION_ID "1" anchor no longer exists and setup aborted. Anchor the
injected listener patch on REGION_WORLD instead, and refresh the dev
session window (3 h).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Region 100/1 holds lineage history and cannot host the cellular profile,
so dev provisions a fresh cellular-v1 genesis at region 2 and serves it
with --profile cellular and its --life-route catalog. Region 1's stores
are left untouched; it is no longer served.
The base Deployment and provision Job take REGION_PROFILE,
REGION_LIFE_ROUTES, REGION_OPENING and REGION_OPENING_ARGS. Their
defaults keep stage on the lineage profile with the development opening.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Compile the PostgreSQL stages with PG_BUILD_JOBS=6 (was 2), including
PostgreSQL itself. postgres-check.sh gains --full (release),
--sanitizer (ASan: C++ suite plus backup/DR/upgrade process checks that
run the instrumented binaries) and --suite-only (clang-tidy build), so
SQL-only checks run once. Each run prints per-phase timings, and the
postgres job logs runner CPU, memory and disk headroom.
Also refresh the dev REGION_SESSION_WINDOW.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
apply.sql matched v1..v21 against the full checked-in history but v22..v33
only against the final entry, so a changed checksum for any earlier
migration was admitted once a database reached v22. postgres-check's
checksum-drift rehearsal caught this (zeroed v14 sha256 migrated cleanly).
Build the exact v21 history once and extend it per version.
Verified on PostgreSQL 18.6: fresh migrate and re-run, v30 -> v33 upgrade,
and refusal of a fake v34 and of zeroed checksums for v2, v14, v21, v22,
v30 and v33, including a drifted v30 database. Refresh the dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
postgres-check expected granting find_region_action_record to the
runtime role to fail the role audit, but migration 19 grants it and the
audit lists it as fenced, so the grant was a no-op and the audit passed.
Exercise the internal check_cellular_action_block validator instead,
mirroring the existing check_action_block case. Refresh the dev session
window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The upgrade rehearsal asserted the schema stayed at version 23 after a
deliberately failing migration, but migrations now reach 33, so it
reported "schema version moved" although nothing changed. Record the
version before the failing migration and compare against that. Refresh
the dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The primitive_life_bindings foreign key and evidence trigger checks
selected a bare boolean, which libpq returns as "t", but compared it
with "true" (region migration test line 216). Cast to text like every
other check in the file. Refresh the dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Migrations 31-33 were added without updating the history the migration
test expects after running migrate-postgres.sh (region migration test
line 214). postgres-check.sh already expects 1..33. Refresh the dev
session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
encode_manifest refuses unsupported cellular version bundles, so the
migration test could never produce the manifest it sends to the SQL
parser (region migration test line 587). Encode the valid v1 manifest
and bump the archive version byte at offset 188 instead. Refresh the
dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The primitive provisioning test provisioned world 7 region 87, which the
synthetic load suite provisions again later in the same disposable
cluster, failing with region_leases_pkey SQLSTATE 23505. Move it to the
unused region 89. Refresh the dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The primitive provisioning fixture passed a 66-hex-digit --build, which
evo-world-clone rejects as a usage error, so the first owner provision
never ran (region store test line 322). Match the 32-byte identity the
fixture already retains. Refresh the dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CI run 136 failed network: Phenotype::rate_microseconds (added with development version
2) is not carried by either session protocol, so the organism view no longer re-encoded
canonically. The phenotype keeps its replicated shape; the period follows from the
development version (development_rate_microseconds) and MetabolicBudget carries it.
No protocol change; first-bud times unchanged. Refresh the dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Development version 2 is defined per second; phenotype/budget rates carry their period
(rate_microseconds: 50 ms for v1, 1 s for v2) and the M2 fixture adapters accept only
v1. Tunable v2 constants (maintenance 10, membrane upkeep 15, movement 50 pJ/s per
1000 fg) and the primitive founder (schema 2: 1 um membrane, 8e6 fg, 200 um/s,
574,000 fg/s ingestion, 500 per-thousand bud) give a first bud in 300/600/901 s when
feeding 100/50/33 % of the time at full-speed movement (bud_funding_seconds; never at
10 %), the maintainer's 5-15 minute target. Tests pin the range; v1 is unchanged.
11.1 complete. Refresh the dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Rules 3 is rules 2 with irradiance, growth, intake and decay per second. The community
derives its per-tick form once at creation and restore (rationals exact via the tick in
numerator and denominator, caps rounded down); steps and the snapshot layout are rules
2's; the codec reads formats 1-3. At 50 ms a x20 per-second twin steps identically
through feeding, death and decay and across a snapshot; at 25 ms caps halve. Rules 1/2
and the development-v1 opening are unchanged. Development version 1 outputs stay tied
to their 50 ms quantum and the tick-abstract M2 fixtures are not migrated (recorded).
environmental-qualification's "unknown rules" case moves from 3 to 4.
Refresh the dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Movement and ingestion are stored per second in schema 2 (ranges x20). Schema 1 records
stay valid, are never rewritten, and read as exactly x20 via Genome::per_second; drafts
default to schema 1 so existing literals keep their meaning. Development version 1
converts genome rates at its own 50 ms quantum (round down), so phenotypes are
unchanged. Mutation keeps the parent's schema and step; the lineage archive header names
exactly the highest schema held (all-schema-1 archives keep their bytes); the
development-v1 opening names schema 1. No golden changed; two tests that used 2 as an
unknown schema now use the next one. Refresh the dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Rates are stored per second in their integer base units and converted at the tick
boundary with no carried state: costs round up, capacities down (maintainer decision),
error below one unit per tick. per_second_from_per_tick is the migration's exact inverse
(x20 at 50 ms) and refuses inexact values. New rate-contracts test. No biology uses it
yet; genome v2, development v2 and process fields follow. Refresh the dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- A stale_grant before Welcome spends only that grant: the session tries the next one
in the bundle and ends as exhausted when none is left (other refusals still stop).
- UI scale (ADR-0084 amendment): embedded scalable font (ProggyForever), scale from
logical window height (720 reference, 1.0-2.5) times a Ctrl+= / Ctrl+- nudge; panel
sizes and table columns in logical units; movable panels placed again per scale.
- --size W,H for engineering review screenshots.
- Tests: client_remote_session (stale grant, exhausted), client_ui_state (scale).
Refresh the dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Maintainer decisions (2026-10-09) for the first playable loop: budding (parent keeps
control, one funded wildlife bud); funded in-life mutation from seeded candidates;
exact-copy heredity, no NPC mutation in M11; logout freezes at once, lingers vulnerable,
then suspends; resume in place or nearest valid spot within a bounded radius;
absorption plus engulfment by membrane-contested size ratio; first bud fundable within a
session (tunable); friend recognition deferred. Updates OPEN_QUESTIONS 7/10/14/31/33,
REPRODUCTION_MODEL, ROADMAP, STATUS; BACKLOG gains the open M10 native-client items.
Refresh the dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The connect form paired the dev endpoint with the local test PKI's name
(dev-region.evolution.test), so a default connection to the dev region failed the
certificate name check. client_ui_model checks the new default; the tester doc names it.
Refresh the dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Maintainer decision (2026-10-09): the region session cap becomes a safety bound of 2000
(plus 64 connections of headroom); the load governor is the working limit. Library bounds
go from 1024 to 4096 in SessionControl, IntentAdmission and ReplicationServer (defaults
stay 32). Drills stop for now: saturation and native-client observation move to a later
2k drill, whose prerequisites ADR-0097 lists (single-thread CPU, region login rate,
256 MiB pod limit, linear lookups). Declared: >= 32 sessions within every budget on the
reference worker. Budgets v2, STATUS, ROADMAP, VALIDATION. Refresh the dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Image c597c0ba (batched durable lookups). Clients within budget at 20-32 sessions; server
within budgets v2 at 24, 28 and 32 (the configured session cap): work p95 <= 12.9 ms,
commit ticks <= 44.3 ms, session p99 <= 13.3 ms, no tick debt. At 20, four refusals during
the login ramp and per-session p99 0.554 ms. Saturation beyond the cap and native-client
observation remain open. Refresh the dev session window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>