Files
aevolution/docs/engineering/M12_DELIVERY_PLAN.md
T
dingo 627411983f
Foundation / checks (ecs) (push) Successful in 2m41s
Foundation / checks (tsan) (push) Failing after 18m12s
Foundation / server-image (push) Canceled after 0s
Foundation / image (push) Canceled after 0s
Foundation / checks (network) (push) Canceled after 20m41s
Foundation / checks (postgres) (push) Canceled after 20m43s
Foundation / checks (checks) (push) Canceled after 20m49s
Foundation / checks (deploy) (push) Successful in 5s
Foundation / checks (client) (push) Failing after 17m4s
Foundation / postgres-tools (push) Canceled after 0s
M12.3: publish and recover durable exact populations
2026-10-17 17:10:17 -04:00

6.0 KiB

M12 delivery and prerequisite conflicts

Requested scope: ecological identity, population qualification, durable automatic tier transitions, unattended scheduling/migration and environmental coupling. Commit and push each validated scoped segment separately.

Gate status (2026-10-17)

Subset State Evidence or condition for completion
12.1 Engineering evidence policy and explicit deferrals documented ADR-0133, identity/time contract and worked histories; no new game rules or runtime versions.
12.2 Exact local model selected and paired qualification implemented (12.2d); production cost gate open ADR-0139, exact canonical equality across 196,608 paired ticks per platform. Regional approximation disabled; no compression claim. Deployment-host capacity and fuller interaction coverage remain open.
12.3 Exact population profile/trace/store and opt-in host integration implemented (12.3b) ADR-0142, schema 44; exact publication/recovery, mutation race guards, ambiguous replies and process kills. Executable/network population fault drills and target-host qualification remain open.
12.4 Current-worker unattended demand and confirmed-head pacing implemented (12.4a) ADR-0141. Population durable intervals and biological migration escrow remain unimplemented.
12.5 Blocked on 12.2-12.4 Finite M6 process fixtures do not qualify coupled persistent cellular/cohort atmosphere/water budgets.
12.6 Existing local archive retained; integrated population history open World-level durable event/query/client path remains required.
12.7 Evidence policy defined; reconciler open Complete durable world watermark and every viable/in-flight store remain required.
12.8 Not qualified Actual multi-day integrated ecology/restart/clone evidence remains required.
12.9 Not accepted Returning-player product evidence and expansion review remain required; World0 stays separately blocked.

The remaining-gate audit identifies the exact publication boundary and concrete integration work. This is not an approval hold or a request to reopen M11.9. Do not mark portable recovery, current-worker pacing or a short research run as completion of the dependent durable ecology/product gates.

The maintainer explicitly closes M11.9 for now on 2026-10-17 and authorizes M12 continuation; see the closure decision. The recorded solo sample and fun/variety questions remain incomplete empirical evidence and a follow-up backlog. They no longer block M12 engineering under that decision. Current work must use the overhauled microscope UI and implemented rules 8 (ADR-0136 step A, ADR-0137), not the older square-patch/rules-6 baseline. Proposed colony, infall and photosynthesis mechanics remain outside this implementation.

Executed 12.2a results and remaining limits: M12_POPULATION_QUALIFICATION. Release CTest passed 69/69 outside the filesystem sandbox. No production model was adopted.

12.2b now implements the separately versioned exact-reserve/age research candidate proposed below: experiment results, ADR-0134. It removes the synthetic false-extinction mechanism but often retains one row per individual. Release CTest passed 70/70, with Linux analysis and ASan/UBSan checks. The rules-8 control below supplies the cellular reference. Supported-state/model review remains ahead; neither exact synthetic fidelity nor workstation costs qualify a production tier.

12.2c implements the current rules-8 cellular reference: results and scope, ADR-0138. Eight full cases compare 65,536 individual ticks and 70,656 replay boundaries, retaining mutation, genealogy, inventory and offline state across encoded replay/checkpoint restore. MSVC/GCC non-timing outputs agree; native Release CTest passes 71/71. This is an exact reference control; the paired cellular/cohort model, fuller predation/capacity matrix and target-host budgets remain open. No automatic tier selector is enabled.

Architectural conflict and resolution (updated 2026-10-17)

The maintainer delegates M12.2 decisions, subject to later revision. ADR-0139 selects exact retained-member local cohorts instead of the failed average model. Paired results validate biological equality with conversion oscillation and local fallback samples up to the existing 256-cell ceiling. This permits implementing durable exact ownership without enabling a production selector. Target-host/scaling gates remain explicit; the historical conflict below still prohibits deploying approximate aggregation.

Automatic transitions or regional aggregation now would conflict with ADR-0041 and ECOLOGY_MODEL: the measured model can erase viable populations while exactly conserving inventory. The M4 test-only model is not the M11 spatial cellular model; wrapping its in-memory conversion primitives in persistence would not qualify it. Under the engineering constitution, stop expanding the affected implementation.

The isolated 12.2b candidate retains the failed baseline and exact age/reserve variation; it does not yet address cellular/genome variation. Before running it, declare population/biomass/diversity trajectory bounds, extinction criteria, supported cellular states and hold-out seed/regime splits. Compare against the accepted M11 controller, contact, contested intake, budding, death and offline rules. An exact retained-member control can check the instrument but cannot by itself establish useful aggregation or acceptable target-host cost.

Measure bin/interval sensitivity, conversion and full-region costs on the Linux target host, including active fallback capacity and backpressure. Passing the M4 synthetic envelope alone is insufficient. Preserve the scoped M11 closure above. Only then accept the model/tier ADR and proceed to 12.3, followed by owned durable intervals/escrow in 12.4 and versioned finite resource coupling in 12.5. No World0 deployment or promotion follows from these gates.