P08-03: 配置の通常操作受入

This commit is contained in:
Agent Control Runner
2026-10-14 04:53:00 +09:00
parent 8e3d79d5f2
commit 4a17d0e0f5
15 changed files with 14249 additions and 56 deletions
@@ -57,7 +57,7 @@ and saves outside private profiles. The mock collector supplies a controlled
disk-space input; the production disk-admission gate is unchanged.
Current native/OS/full regression/review/integration remain pending.
## r02 whole-model operation repair (current candidate)
## r02 whole-model operation repair (retained earlier submission)
The retained submitted HEAD is `c2b59f04f94dda0a61991ec3bb7f0e946196f995`.
`NATIVE_R02_READER_OPERATION_DELTA.json` and `native-r02-reader-stderr.log`
@@ -134,3 +134,35 @@ No reused P08-02 PASS is claimed for changed sources. P08-01 and the whole
P08 package remain open. Windows same-distribution/device evidence remains
open; Mac/Linux results cannot substitute for it. No outside-scope service
defect or required permission change has been demonstrated by this worker.
## r04 initial OS load oracle repair (current submitted candidate)
The supplied OS_R04_INITIAL_LOAD_ORACLE_FINDING.json and
os-r04-writer-engine.log identify the bootstrap-only simple equality failure.
The registered build contains 2153 files; its complete read-only inventory is
recorded in registered-r04-evidence-manifest.json. Historical raw/PASS evidence
is unchanged and no changed-source acceptance is inherited.
Passive setup now uses the existing verify_operation oracle with bootstrap-
prefix. Complete typed JSON and raw Variant formal, checkpoint (postload),
returned Dictionary, expected-operation and operation-actual files, plus a
separate comparison JSON, remain distinct from writer/reader final proofs.
The same source-defined notification logic checks all state, previous events,
next event ID/counter and every payload field; no event masking or omissions.
The checker binds bootstrap formal/returned models to the recorded read-only
inspection and first Runtime return, reconstructs the expectation and checks
whole typed state and raw expected/actual identity.
Static inspection within the allowed fixture/recorder scope found no other
postload-versus-formal simple equality oracle. Remaining exact comparisons
check presave/formal identity or read-only preservation and are intentional.
Seven added semantic controls cover missing/extra notifications, counter,
payload and unknown-field drift, missing raw and overwritten bootstrap raw;
all 49 earlier tests remain. Final actual results and exact source/input
hashes are recorded in local-validation.json and local-source-manifest.json.
Only pure Python validation uses the ordinary 900-second subprocess timeout.
Root native/OS collection for this candidate, final full regression after
freeze, separate Opus review and integration/current final verification remain
open. All five conditions and the 30/88 denominator are unchanged. No Runtime,
UI, SaveService, collector, specification or historical evidence edits.