GodopuandClaude Opus 5.5
bf8a583afb
PR #48 already carries the 4.8.5 lockstep bump. Merge-workspace review: round 1 (@35935f9) claude agy grok NOT PASS (test portability: tomllib on Python 3.9, /proc and /tmp on macOS), codex no verdict (model at capacity, budget spent); fixed in36f0b8dande8d9151. Round 2 (@e8d9151) codex NOT PASS (unrunnable config check passed an unchecked merge, odd pids closed a pane, nan dropped the user's settings); fixed inab93776. Round 3 (@ab93776) 4/4 PASS. Full suite atab93776on macOS / Python 3.9: 1456 passed / 0 failed / 38 skipped. Reports and the round-3 identity sidecars go to .agents/reports/<session>/. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01VCgu9Fa9iT7TbJUSzdq4Ff
3.4 KiB
Review: PR #48 round 2 @ e8d9151
- Role: Reviewer (grok)
- Identity: loop
loop-20261005T081844Z-e92124, basec5c7eff6f758c76ac37a138dbc3c301f3f7e10cf, heade8d9151226d8139c7b5688e83d035a3a747aeb2d, tree8b39188453ff8a7263aa4400792fabc99188dad6, clean snapshot, ancestor of bothc5c7effand round-135935f9 - Scope for this reviewer:
git diff 35935f9 e8d9151only (commits36f0b8d,e8d9151). Five files, +56/−9,git diff --checkclean. No product code. Main stayedc5c7eff. - Verified read-only at
/var/folders/q_/68sxgw3j073dgjg8wl9s08vw0000gn/T/mam-review/loop-20261005T081844Z-e92124/1.
Round-1 findings
| Round 1 | e8d9151 |
|---|---|
tests/test_issue45_herdr_config.py:16 imported tomllib at module level, so Python 3.9/3.10 could not collect the file. tests/test_issue9_server_env.py:157 did the same inside the bootstrap test, and expected the merged document from a 3.9 python3 that writes only the key. |
Import is try/except (test_issue45_herdr_config.py:18-21). needs_tomllib skips the 29 tests that parse a result or expect the user's other settings (:36, applied at :71 and the other parse sites). test_without_tomllib_restore_is_still_off still runs. test_issue9_server_env.py:173-183 parses with the python3 on PATH and, when that interpreter has no tomllib, expects the key-only file and no tomllib in the shim stderr. create/SKILL.md:120 and VERSIONS.md:49 say a python3 older than 3.11 leaves restore off, drops the other herdr settings, and warns. |
tests/test_issue45_restore.py:227 read /proc/<pid>/environ before the restore-on control, so macOS never reached that control. |
_process_env (:160-168) reads /proc/<pid>/environ when it exists, otherwise ps -E -ww -o command=. The control stays after that check (:242-248). On this Mac the herdr server's ps -E line contains HERDR_CONFIG_PATH=<resolved>/mam-server-config.toml, and both test_a cases passed, so the control ran. |
:263 and :325 compared pane cwd to the unresolved /tmp/... from mkdtemp. herdr reports /private/tmp/.... |
The fixture resolves the temp root (:175) before HOME is set, so later str(rig.root) and str(ws) match herdr. test_b_only_the_resumed_sessions_restored_shell_is_closed and test_b_resume_leaves_no_restored_shell_behind passed. |
The tomli backport suggested in round 1 is not in this delta. The docs now describe the path this host's /usr/bin/python3 (3.9.6, no tomli) actually takes. That matches the decision to document it.
tests/data/skill_code_blocks.json is unchanged. The create skill edit is the same line 120, so no fence moved. test_issue42_skill_blocks passed, with the usual worktree skip at :266 because .claude/skills is absent in the snapshot.
Tests
Required command, PR venv Python 3.9.6, --basetemp=/tmp/b471follow.eYGqXM (no mam-review in the path), MQTT variables unset:
152 passed, 30 skipped in 214.33s. The 30 skips are the 29 needs tomllib cases plus test_issue42_skill_blocks.py:266.
Same file on a throwaway Python 3.11.15 venv (/opt/homebrew/bin/python3.11), HOME set to an empty temp directory:
51 passed in 0.38s, none skipped.
Afterward no mamt45 process remained. ~/.config/herdr/config.toml mtime stayed 1784456394, the directory mtime stayed 1791183576, and mam-server-config.toml was not created there. Snapshot git status was empty.
[VERDICT: PASS]