refactor(review): delete the vestigial timeout sentinel (review feedback)

1be5742→main·

peyton-alt·1w ago·3 files·+38 added/-71 removed

Review feedback asked whether the two timeout helpers could collapse to max(). Tracing every ReviewerTimeout reader shows the truth is stronger on one side and subtler on the other:

• resolveReviewerTimeoutArg is deleted outright, not simplified. Its -1 'disabled' sentinel was load-bearing in the old three-state world (positive/default-20m/disabled); with the default gone, both remaining consumers (reviewerTimeout, judgeTimeoutArg) use a plain > 0 check, so -1 and 0 were indistinguishable everywhere and the mapping did nothing. The --timeout flag value now flows into RunConfig unmapped. This also removes the stale doc comment Copilot flagged ('the flag's default is nonzero').

• judgeTimeoutArg stays as a named function — the point is its contract (the judge is one bounded call and must never be uncapped, whatever the reviewer setting says), not the arithmetic — but its body is now max(reviewerArg, 0) as suggested (builtin max; math.Max is float64). A negative can still arrive via an explicit '--timeout -5m'.

• types/reviewer.go's ReviewerTimeout doc still described the deleted defaultReviewerTimeout three-state; rewritten to the two-state contract.

Tests updated to pin the flag->RunConfig two-state flow directly.

Co-Authored-By: Claude Fable 5 noreply@anthropic.com

Sessions

01KX0NQY37TYQ2B8DF5CGBA8FJView transcript

Changes

3

• cmd/entire/cli/review

  • Mcmd.go+13/-29

  • Mrun_test.go+20/-38

• types

  • Mreviewer.go+5/-4

222 unmodified lines

... // remainder of the content follows similarly...