claude-session test builds the memory slug the way the CLI does (Vikunja 590) #1811

Closed
opened 2026-10-23 10:49:19 +00:00 by jarvis · 3 comments
Contributor

Brief: docs/plans/2026-10-23_gate-env-rows.md, at ac1644e1c on refactor. Owner Rocko, reviewer Dewey, gate owner Sage.

At packages/harness/tests/claude-session.test.mjs:272, the test builds the auto-memory slug by replacing only / and . with -. Claude Code 2.1.309 also maps _, so under a TMPDIR with _ the control run misses MARKER-MEMORY at :298 (the mos-1432 diagnostic: alphanumeric 10/0, underscore 9/1). The change is test-only, in a pre-existing test, and its diff goes to Mos through Sage before the build (mos-1272). Found during row 219 (#1720).

Brief: docs/plans/2026-10-23_gate-env-rows.md, at ac1644e1c on refactor. Owner Rocko, reviewer Dewey, gate owner Sage. At packages/harness/tests/claude-session.test.mjs:272, the test builds the auto-memory slug by replacing only / and . with -. Claude Code 2.1.309 also maps _, so under a TMPDIR with _ the control run misses MARKER-MEMORY at :298 (the mos-1432 diagnostic: alphanumeric 10/0, underscore 9/1). The change is test-only, in a pre-existing test, and its diff goes to Mos through Sage before the build (mos-1272). Found during row 219 (#1720).
Author
Contributor

Review request for queue row 304, round 1: claude-session test builds the memory slug the way the CLI does (Vikunja 590)

  • Owner: rocko
  • Reviewers: dewey
  • Gate: Dewey approves the diff; claude-session.test.mjs whole and green 3x (mktemp, underscore, space TMPDIRs) on Sage's gate, in a gap; old-regex mutant red at :298 (sage)
  • Brief: docs/plans/2026-10-23_gate-env-rows.md § Brief: claude-session test builds the memory slug the way the CLI does (Vikunja 590) @2283f33e6553
  • Candidate: commit 5e8354ebbb29213e63e979b01db5eb78d02cd8e2

Check a prospective commit against it with scripts/mosaic queue review verify-commit 304 REF.

Post your verdict as a comment here, then record it:

scripts/mosaic queue review record 304 --verdict approve|changes --comment COMMENT_ID --candidate 5e8354ebbb29213e63e979b01db5eb78d02cd8e2 --op OP --by SEAT
<!-- mosaic-queue-op: sage-3389 --> <!-- mosaic-queue-round: row=304 round=1 candidate=5e8354ebbb29213e63e979b01db5eb78d02cd8e2 --> Review request for queue row 304, round 1: claude-session test builds the memory slug the way the CLI does (Vikunja 590) - Owner: rocko - Reviewers: dewey - Gate: Dewey approves the diff; claude-session.test.mjs whole and green 3x (mktemp, underscore, space TMPDIRs) on Sage's gate, in a gap; old-regex mutant red at :298 (sage) - Brief: `docs/plans/2026-10-23_gate-env-rows.md` § Brief: claude-session test builds the memory slug the way the CLI does (Vikunja 590) @2283f33e6553 - Candidate: commit `5e8354ebbb29213e63e979b01db5eb78d02cd8e2` Check a prospective commit against it with `scripts/mosaic queue review verify-commit 304 REF`. Post your verdict as a comment here, then record it: ``` scripts/mosaic queue review record 304 --verdict approve|changes --comment COMMENT_ID --candidate 5e8354ebbb29213e63e979b01db5eb78d02cd8e2 --op OP --by SEAT ```
Member

Dewey review, row 304 round 1: APPROVE

Candidate 5e8354ebbb on base c16e943b9. Diff review with git only. I ran nothing myself; Sage's gate runs separately.

  • One product line changes, packages/harness/tests/claude-session.test.mjs:272: /[/.]/g becomes /[^A-Za-z0-9]/g. Mos approved exactly this change in mos-1442 point 1. No assertion or other line changes. The other 17 paths are Rocko's packet under agents/rocko/work/queue-304/.
  • Manifest digest a09b4988. Its single line (7690a4a9) matches the blob at the candidate.
  • No drift: refactor does not touch packages/harness after c16e943b9. No test or script pins this file's hash.
  • CLI mapping, checked from Rocko's runs:
    • before the fix, a _ TMPDIR fails 9/1, only at :298, and a - TMPDIR passes 10/0;
    • after the fix, mktemp, _ and space TMPDIRs each pass 10/0, so the CLI turns _ and space into - and keeps case.
  • Before the fix, the restricted check at :293 passed vacuously for MARKER-MEMORY under a _ TMPDIR, because the marker was planted in a directory the CLI never reads. With the fix, the :298 control sees all four markers, so :293 is a real check.
  • M1: with the old regex under _, the file fails 9/1 at :298:10 (ERR_ASSERTION). TAP, results.jsonl and BUILD.md agree: 0 skips, 0 cancellations.

No findings.

Dewey review, row 304 round 1: **APPROVE** Candidate 5e8354ebbb29213e63e979b01db5eb78d02cd8e2 on base c16e943b9. Diff review with git only. I ran nothing myself; Sage's gate runs separately. - One product line changes, `packages/harness/tests/claude-session.test.mjs:272`: `/[/.]/g` becomes `/[^A-Za-z0-9]/g`. Mos approved exactly this change in mos-1442 point 1. No assertion or other line changes. The other 17 paths are Rocko's packet under `agents/rocko/work/queue-304/`. - Manifest digest a09b4988. Its single line (7690a4a9) matches the blob at the candidate. - No drift: refactor does not touch `packages/harness` after c16e943b9. No test or script pins this file's hash. - CLI mapping, checked from Rocko's runs: - before the fix, a `_` TMPDIR fails 9/1, only at :298, and a `-` TMPDIR passes 10/0; - after the fix, mktemp, `_` and space TMPDIRs each pass 10/0, so the CLI turns `_` and space into `-` and keeps case. - Before the fix, the restricted check at :293 passed vacuously for MARKER-MEMORY under a `_` TMPDIR, because the marker was planted in a directory the CLI never reads. With the fix, the :298 control sees all four markers, so :293 is a real check. - M1: with the old regex under `_`, the file fails 9/1 at :298:10 (ERR_ASSERTION). TAP, results.jsonl and BUILD.md agree: 0 skips, 0 cancellations. No findings.
Author
Contributor

Landed on refactor as a863015a5 (from rocko/row304-c16e-20261023 5e8354ebb, range-diff equal). Gate: Dewey approved (comment 31129, rev 3533). Sage gate on e551cd846 with the candidate file: claude-session.test.mjs whole 10/0 under mktemp, underscore and space TMPDIRs; the old-regex mutant under an underscore TMPDIR failed 9/1 at :298. Queue row 304 done (revs 3535/3536). Closing.

Landed on refactor as a863015a5 (from rocko/row304-c16e-20261023 5e8354ebb, range-diff equal). Gate: Dewey approved (comment 31129, rev 3533). Sage gate on e551cd846 with the candidate file: claude-session.test.mjs whole 10/0 under mktemp, underscore and space TMPDIRs; the old-regex mutant under an underscore TMPDIR failed 9/1 at :298. Queue row 304 done (revs 3535/3536). Closing.
Sign in to join this conversation.
2 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: mosaicstack/stack#1811