fix: candidate-extract.jq's overlap() is now a symmetric, multiplicity-free Jaccard (c9a4) - #46
Open
talgolan wants to merge 1 commit into
Open
fix: candidate-extract.jq's overlap() is now a symmetric, multiplicity-free Jaccard (c9a4)#46talgolan wants to merge 1 commit into
talgolan wants to merge 1 commit into
Conversation
…y-free Jaccard (c9a4)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
overlap()inhooks/lib/candidate-extract.jqhad a numerator/denominator mismatch: the numerator counted one side's title words with multiplicity while the denominator was the deduped union, sooverlap(A;B) != overlap(B;A)and short/similar titles scored above the 0.7 dedup threshold when they shouldn't (vitest/jestscored 0.909). A newdedup_keyalso strips the retry-burst heuristic's shared title-template suffix before scoring, so two distinct retry-bursts sharing that suffix no longer over-merge (was 0.722).c9a4.This was originally Task 1 of a 6-task Determinism Phase 5 plan (
meta/superpowers/plans/2026-09-07-determinism-phase-5-backlog-mechanics.md). Tasks 2-6 were abandoned mid-run: they builtbacklog-item.sh's tag-mint/renumber mechanics and BACKLOG.md-tag-convention wiring inprimer.md/end-session.md, all premised on.session-continuity/BACKLOG.mdstaying the backlog store — moot now that the backlog store is migrating to GitHub Issues. This commit is unrelated to backlog storage either way and stands on its own.Full task + fix-round + re-review cycle completed via subagent-driven-development: task review approved with one Important, plan-mandated finding (a test that never actually exercised
overlap()'s merge path), fixed in a follow-up round, re-review confirmed addressed with no new breakage.Test plan
zsh meta/superpowers/validation/2026-09-01-candidate-extract-smoke.zsh— 34/34 passing (31 pre-existing + 3 new c9a4 regression assertions: suffix-over-merge stays distinct, symmetric scoring, genuine near-duplicates still merge).meta/superpowers/validation/*.zshsuite run against this branch: all green except two pre-existing failures unrelated to this diff (stale count-pins inlearnings-index-smoke.zsh/render-smoke.zshagainst realBACKLOG.md/LEARNINGS.mdentry counts — confirmed by diff scope that this commit never touches those files).🤖 Generated with Claude Code