odd/gate/transitions: add inflected forms for geminating verbs#122
Merged
Conversation
Stemmer limitation surfaced during Phase 2 smoke testing: 'shipping' stems to 'shipp' not 'ship', and 'stepped' stems to 'stepp' not 'step'. The Porter-style stemmer in workers/src/bm25.ts drops -ing and -ed but doesn't reverse consonant gemination. Fix via canon rather than stemmer: add 'shipping, shipped' to the execution-to-completion row and 'stepped back, stepping back' to the execution-to-exploration row so these common forms match by direct vocabulary. Notes section updated with honest description of the stemmer's reach and why these few inflections are listed explicitly. Non-geminating verbs (deploy, build, start, reconsider, etc.) continue to match their inflections via stemming alone. Refs klappy/oddkit P1.3.2 Phase 2 smoke preview.
This was referenced Apr 20, 2026
klappy
pushed a commit
that referenced
this pull request
Apr 20, 2026
Ledger at odd/ledger/2026-04-20-p1-3-2-gate-canary-landed.md captures the 2026-04-20T01:21Z-03:20Z session that shipped oddkit 0.20.0. Mirrors the P1.3.1 ledger structure (Summary, What Shipped, What the Validator Actually Checked, Patterns, Cleared, O-opens, Session Mechanics, Handoff, Provenance). Two handoffs flipped to status: superseded with superseded_by pointing at the new ledger: - odd/handoffs/2026-04-21-p1-3-2-gate-canary — the original two-phase handoff, superseded by the full P1.3.2 ship. - odd/handoffs/2026-04-20-p1-3-2-phase-2-gate-code-refactor — the mid-session Phase 2 forward handoff, superseded on arrival when Phase 2 shipped in the same session. Left in canon as documentation of the 'same-session handoff anti-pattern' called out in the ledger. Honest accounting in the ledger of the tradeoff between this session's smoke-heavy attestation (9 runs + live self-call) and P1.3.1's Sonnet 4.6 5-corroboration validator pattern. Neither is strictly superior; they catch different classes of issue. Refs klappy/oddkit#118 (merged 260492c), #119 (promotion 1308245), #120 #121 #122.
This was referenced Apr 20, 2026
klappy
added a commit
that referenced
this pull request
Apr 20, 2026
…flip predecessor handoff to superseded
Honest closeout for P1.3.3.
Two halves, both load-bearing:
1. Technical work that shipped:
- D5 stemmed prereq matcher (replace per-prereq regex with stemmed
set intersection + 4 structural-test side-paths)
- D9 cache removal (drop cachedChallengeTypeIndex; inline
buildBM25Index per cache-fetches-and-parses)
- cache-fetches-and-parses graduated tier-2 canon (klappy.dev#125,
merged 3726073)
- All landed in oddkit 0.21.0 (PR #120 merged 33ca5bf,
PR #121 merged 25ad719)
2. Process failure that demonstrated why release-validation-gate
needed to exist:
- Orchestrator merged PRs #120 and #121 with Cursor Bugbot still
in_progress, treating it as non-blocking
- Skipped Sonnet 4.6 validator dispatch despite P1.3.2 ledger
warning that smoke-only should not become the default
- Bugbot subsequently posted 2 findings: medium-severity prod
regression breaking the strictly-additive invariant the PR
description claimed (stop-word filter dropping 'from' from
source-named vocab), low-severity DRY violation in
BasePrerequisite (re-listed PrereqMatchVocab fields manually
instead of intersection)
- Approximate prod-regression window: 04:11Z to 05:09Z + warmup
≈ 1h 39m on 'from'-keyword source-named matches
3. Structural fix:
- canon/constraints/release-validation-gate.md (tier 1, 168 lines)
— three binding rules: no merge with active reviews in_progress,
no promotion without independent fresh-context validation when
PR touches load-bearing surface, canon outranks any
session-scoped recommendation
- canon/principles/contract-governs-handoff-drift.md (tier 2,
128 lines) — graduated on third deciding-argument recurrence
(P1.3.1 implicit, P1.3.2 explicit, P1.3.3 explicit-via-failure)
- canon/bootstrap/model-operating-contract.md (+4 lines) —
'Before Shipping Code' section as discoverability hook
- All in klappy.dev#126, merged ee9aee4
- Captures O-open P11: oddkit_gate enforces release-validation-gate
mechanically at execution → completion transitions
4. Fix-forward applied the new canon end-to-end:
- oddkit#122 merged d17bc0c (0.21.1 fix branch, both Bugbot
findings addressed, +2 regression smoke assertions)
- Bugbot wait respected on PR #122 (completed/success)
- Sonnet 4.6 validator dispatched (sesn_011CaERPjHi1CV4TrvKW68jB)
against fix branch + 0.21.0 prod state + canon PR #126
- Validator verdict: CONDITIONAL PASS (conditional only on canon
PR #126 merging first; resolved before promotion)
- oddkit#123 merged 2c5d652b (promotion to prod)
- Bugbot wait respected on PR #123 (~225s, completed/success)
- Prod 0.21.1 verified live; live self-call confirms 'from' fix
Meta-instructive moment: Bugbot caught a within-canon DRY violation
on the bootstrap hook in PR #126 itself. The first application of
release-validation-gate's discipline caught me in another canon
violation (dry-canon-says-it-once) before I could merge it. Working
as designed.
Predecessor handoff (odd/handoffs/2026-04-20-p1-3-3-challenge-revisit.md)
flipped to status: superseded with explicit supersession_note naming
the 'Option A is fine for P1.3.3' line in its Validation Plan section
as the recommendation that produced the incident. Future sessions
reading the superseded handoff will be redirected here.
Carry-forward O-opens updated:
- P11 NEW: oddkit_gate mechanical enforcement of release-validation-gate
- P12 NEW: tokenize() audit pass (any caller using canon vocab/input
should explicitly pass stop-word set)
- P2/P3/P5/P6/P8/P9/P10 carried from prior sweeps unchanged
klappy
added a commit
that referenced
this pull request
Apr 20, 2026
…flip predecessor handoff to superseded (#127) Honest closeout for P1.3.3. Two halves, both load-bearing: 1. Technical work that shipped: - D5 stemmed prereq matcher (replace per-prereq regex with stemmed set intersection + 4 structural-test side-paths) - D9 cache removal (drop cachedChallengeTypeIndex; inline buildBM25Index per cache-fetches-and-parses) - cache-fetches-and-parses graduated tier-2 canon (klappy.dev#125, merged 3726073) - All landed in oddkit 0.21.0 (PR #120 merged 33ca5bf, PR #121 merged 25ad719) 2. Process failure that demonstrated why release-validation-gate needed to exist: - Orchestrator merged PRs #120 and #121 with Cursor Bugbot still in_progress, treating it as non-blocking - Skipped Sonnet 4.6 validator dispatch despite P1.3.2 ledger warning that smoke-only should not become the default - Bugbot subsequently posted 2 findings: medium-severity prod regression breaking the strictly-additive invariant the PR description claimed (stop-word filter dropping 'from' from source-named vocab), low-severity DRY violation in BasePrerequisite (re-listed PrereqMatchVocab fields manually instead of intersection) - Approximate prod-regression window: 04:11Z to 05:09Z + warmup ≈ 1h 39m on 'from'-keyword source-named matches 3. Structural fix: - canon/constraints/release-validation-gate.md (tier 1, 168 lines) — three binding rules: no merge with active reviews in_progress, no promotion without independent fresh-context validation when PR touches load-bearing surface, canon outranks any session-scoped recommendation - canon/principles/contract-governs-handoff-drift.md (tier 2, 128 lines) — graduated on third deciding-argument recurrence (P1.3.1 implicit, P1.3.2 explicit, P1.3.3 explicit-via-failure) - canon/bootstrap/model-operating-contract.md (+4 lines) — 'Before Shipping Code' section as discoverability hook - All in klappy.dev#126, merged ee9aee4 - Captures O-open P11: oddkit_gate enforces release-validation-gate mechanically at execution → completion transitions 4. Fix-forward applied the new canon end-to-end: - oddkit#122 merged d17bc0c (0.21.1 fix branch, both Bugbot findings addressed, +2 regression smoke assertions) - Bugbot wait respected on PR #122 (completed/success) - Sonnet 4.6 validator dispatched (sesn_011CaERPjHi1CV4TrvKW68jB) against fix branch + 0.21.0 prod state + canon PR #126 - Validator verdict: CONDITIONAL PASS (conditional only on canon PR #126 merging first; resolved before promotion) - oddkit#123 merged 2c5d652b (promotion to prod) - Bugbot wait respected on PR #123 (~225s, completed/success) - Prod 0.21.1 verified live; live self-call confirms 'from' fix Meta-instructive moment: Bugbot caught a within-canon DRY violation on the bootstrap hook in PR #126 itself. The first application of release-validation-gate's discipline caught me in another canon violation (dry-canon-says-it-once) before I could merge it. Working as designed. Predecessor handoff (odd/handoffs/2026-04-20-p1-3-3-challenge-revisit.md) flipped to status: superseded with explicit supersession_note naming the 'Option A is fine for P1.3.3' line in its Validation Plan section as the recommendation that produced the incident. Future sessions reading the superseded handoff will be redirected here. Carry-forward O-opens updated: - P11 NEW: oddkit_gate mechanical enforcement of release-validation-gate - P12 NEW: tokenize() audit pass (any caller using canon vocab/input should explicitly pass stop-word set) - P2/P3/P5/P6/P8/P9/P10 carried from prior sweeps unchanged
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.
Small canon follow-up surfaced during Phase 2 smoke testing of klappy/oddkit#gate-governance-source-envelope.
Observation: the Porter-style stemmer at
workers/src/bm25.tsdrops-ingand-edsuffixes but does not reverse consonant gemination. Soshippingstems toshipp(notship) andsteppedstems tostepp(notstep). The CHANGELOG and tool description I wrote for 0.20.0 claimed these forms work via stemming — they don't.Fix via canon, not stemmer. Adding a handful of explicit inflections to the detection vocabulary is cheaper and more auditable than extending the stemmer, and keeps the adjustment editable at canon rather than at the code level. Non-geminating verbs (
deploy,build,start,reconsider, etc.) still match their inflections via the stemmer alone — this PR only adds the forms that gemination blocks.Changes:
execution-to-completionrow:ship, deploy, ...→ship, shipping, shipped, deploy, ...execution-to-explorationrow:... step back, reconsider→... step back, stepped back, stepping back, reconsiderRefs: klappy/oddkit Phase 2 branch
gate/governance-source-envelope(PR pending after this merges and the matching MINIMAL_TRANSITIONS update lands).Note
Low Risk
Low risk: canon-only documentation/vocabulary tweak that broadens match phrases, with no code or data-handling changes.
Overview
Updates
odd/gate/transitions.mddetection vocabulary soexecution-to-completionandexecution-to-explorationalso match geminated verb inflections (e.g.,shipping/shipped,stepped back/stepping back) that the current stemmer doesn’t normalize.Rewrites the Notes to accurately describe the stemmer’s limitations around doubled consonants and explicitly justifies listing these inflections in canon rather than changing the stemmer.
Reviewed by Cursor Bugbot for commit a3c95a0. Bugbot is set up for automated code reviews on this repo. Configure here.