You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[BUG] Correct VS Code LM measurement dates in test titles and comments #27
The measurement-date wording should distinguish when values were captured from when the change was prepared. This is a low-priority wording follow-up; it does not request changes to model budgets or runtime behavior.
Exact deferred finding
Round 2, green finding from PR #26, "[vscode-lm] Add measured model budgets and Astra safety margin":
Test titles and comments say "2026-10-28/29", but every value comes from 2026-10-28 captures.
Status supplied by the review handoff: not addressed.
Reviewed commit: a0c495d.
Evidence
Verified: gh pr view 26 --repo simurg79/Roo-Code --json body,headRefOid,headRefName,baseRefName,url returned the reviewed commit above and this review summary:
Remaining 🟢: test titles say "2026-10-28/29" although every value comes from 2026-10-28 captures.
This records the existing review finding, not a new measurement run.
Acceptance criteria
Correct the affected test titles and comments to identify 2026-10-28 as the capture date, checking against the retained captures before editing.
If 2026-10-29 remains relevant, label it separately as the change-preparation date rather than a capture date.
Leave model values, assertions, and runtime behavior unchanged.
The test title and model-table comment now identify 2026-10-28 UTC as the capture date, with Fable 5 separately identified as 2026-10-29 UTC. The supplied capture evidence places the claude-fable-5 run at 2026-10-29T00:00:02.313Z, just after midnight UTC; its date is therefore a capture date, not a preparation date. This corrects the original finding's assumption that every capture occurred on October 28. Model values, assertions, and runtime behavior are unchanged.
Verified: git diff-tree --no-commit-id --name-only -r HEAD lists only the two requested files. git ls-remote fork refs/heads/feat/vscode-lm-measured-model-budgets-20261029 returned 5056fb87b286faf329493e31068e88e801aee7c4, matching local HEAD.
pnpm exec prettier --check packages/types/src/__tests__/vscode-llm.spec.ts packages/types/src/providers/vscode-llm.ts: All matched files use Prettier code style!
Actual pre-push hook: Tasks: 10 successful, 10 total.
The supplied handoff reports code review/build PASS, tests 85/85 PASS, and runtime PASS under the user's waiver; those historical gates were not rerun in this submission task.
Activity
fix(vscode-lm): correct capture-date wording (#27)
simurg79 commentedon Oct 29, 2026
Fixed in existing PR #26 by commit 5056fb87b286faf329493e31068e88e801aee7c4.
The test title and model-table comment now identify 2026-10-28 UTC as the capture date, with Fable 5 separately identified as 2026-10-29 UTC. The supplied capture evidence places the claude-fable-5 run at
2026-10-29T00:00:02.313Z, just after midnight UTC; its date is therefore a capture date, not a preparation date. This corrects the original finding's assumption that every capture occurred on October 28. Model values, assertions, and runtime behavior are unchanged.Verified:
git diff-tree --no-commit-id --name-only -r HEADlists only the two requested files.git ls-remote fork refs/heads/feat/vscode-lm-measured-model-budgets-20261029returned5056fb87b286faf329493e31068e88e801aee7c4, matching local HEAD.Verified checks (all exited 0):
pnpm --dir packages/types lint:eslint src --ext=ts --max-warnings=0pnpm --dir packages/types check-types:tsc --noEmitpnpm exec prettier --check packages/types/src/__tests__/vscode-llm.spec.ts packages/types/src/providers/vscode-llm.ts:All matched files use Prettier code style!Tasks: 10 successful, 10 total.The supplied handoff reports code review/build PASS, tests 85/85 PASS, and runtime PASS under the user's waiver; those historical gates were not rerun in this submission task.