chore(wolf): sync anatomy/cerebrum/memory/hooks/token-ledger bookkeeping

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AgYRU3zDTqNGV5Jk8aiKLg
This commit is contained in:
Rongsong Shen 2026-08-25 09:59:07 +00:00
parent c2690cbf95
commit c3d6cfa4cc
No known key found for this signature in database
GPG key ID: 5BCDAAE0123E9FE2
5 changed files with 1884 additions and 34 deletions

View file

@ -3981,3 +3981,21 @@
| 02:06 | Edited src/controller/testsuite/test_commit_barrier_reply.c | 2→2 lines | ~39 |
| 02:13 | Edited src/controller/testsuite/test_commit_barrier_reply.c | removed 10 lines | ~35 |
| 17:35 | bug-1687 FIXED: WAL_JOB_START now carries job_pidinfo_t+start_time through both apply seams (mirrors bug-1678's WAL_JOB_DISPATCH pattern) — job.c (record_job_start_fact, restore_job_start_state), persist.c (capnp_encode_pidinfo/decode_pidinfo, persist_submit_job_start_fact, apply_wal_entry+apply_committed_entry WAL_JOB_START handling), persist.h/wal_types.h decls+docs, new testsuite/test_wal_job_start_fact.c (6 tests). RED (4/6 fail pre-fix) → GREEN (6/6) independently verified via git-stash. Also found+fixed a real deadlock my larger payload exposed in test_commit_barrier_reply.c's cbrp_count_jobstate_entries_ex (exact-size `==` match undercounted → downstream test spun forever; fixed to `>=`) plus applied bug-1452's teardown-ordering precedent to cbrp_teardown as defense-in-depth. Full suite: 1146 Tests, 2 Failures (pre-existing bug-1561 overlayfs flake only), zero hangs | src/controller/job.c, persist.c, testsuite/test_wal_job_start_fact.c, testsuite/test_commit_barrier_reply.c, testsuite/controller_test.c, meson.build, src/include/persist.h, wal_types.h, .wolf/buglog.json (bug-1687) | GREEN, full suite verified | ~large |
| 02:24 | Session end: 84 writes across 30 files (persistence.md, SKILL.md, plugin.h, plugin.c, driver.c) | 63 reads | ~674849 tok |
| 02:35 | Created src/controller/testsuite/test_wal_job_finish_fact.c | — | ~6904 |
| 02:35 | Edited src/controller/testsuite/controller_test.c | 2→3 lines | ~36 |
| 02:35 | Edited src/controller/testsuite/controller_test.c | 2→3 lines | ~30 |
| 02:36 | Edited src/controller/meson.build | 2→3 lines | ~34 |
| 02:36 | Edited src/include/wal_types.h | expanded (+17 lines) | ~353 |
| 02:36 | Edited src/include/persist.h | expanded (+20 lines) | ~418 |
| 02:36 | Edited src/controller/persist.c | added error handling | ~259 |
| 02:36 | Edited src/controller/persist.c | added 2 condition(s) | ~320 |
| 02:37 | Edited src/controller/persist.c | modified persist_submit_job_finish_fact() | ~580 |
| 02:37 | Edited src/controller/persist.c | 16→15 lines | ~124 |
| 02:37 | Edited src/controller/persist.c | added error handling | ~560 |
| 02:38 | Edited src/controller/persist.c | added 4 condition(s) | ~587 |
| 02:38 | Edited src/controller/job.c | added error handling | ~453 |
| 02:38 | Edited src/controller/job.c | expanded (+18 lines) | ~416 |
| 02:41 | Session end: 98 writes across 31 files (persistence.md, SKILL.md, plugin.h, plugin.c, driver.c) | 67 reads | ~704342 tok |
| 09:56 | bug-1689: fixed WAL_JOB_FINISH silent field drop (exit_info/finish_time/run_time) on replay, mirroring bug-1687's WAL_JOB_START pattern exactly (native trailer, no capnp needed for job_exitinfo_t) | job.c, persist.c, persist.h, wal_types.h, test_wal_job_finish_fact.c (new) | RED (4 fail, git-stash-verified) -> GREEN (6/6 new tests, 1152 suite tests, only pre-existing bug-1561 flaky failures) | ~large |
| 09:56 | Logged bug-1689 to buglog.json (re-checked max id twice, renumbered bug-1688->bug-1689 after sibling-session collision); updated cerebrum.md Key Learnings (2 entries) | buglog.json, cerebrum.md, anatomy.md | done | ~medium |