chore(vendoring): move the t3code pin to the fork commit actually checked out - #291
Merged
Merged
Conversation
…cked out `pin.contractSource` is `fork`, so `pin.commit` is not a label — it is the commit the vendored contract was generated from, and three separate guards compare live state against it. The fork checkout has been at `26b4c2dc09f0` since issue #272's sidebar change landed there, while `pin.json` still named `2f64a1b0ee2b`. That gap is what turns two suites red for every builder on this machine: - `spec-250-vendoring-identities.test.ts:1215`, "describes the fork commit that is actually checked out", compares the criterion 8b evidence against `git -C $T3CODE_FORK_ROOT rev-parse HEAD`. - `t3-server.mjs verify` exits 1 with `FORK_AHEAD_OF_CONTRACT`. **Why this is not the one-file fix it looks like.** Updating only `250-criterion-8b-evidence.json` swaps one red test for another: `collect-spec-250-evidence.mjs:74-79` fails `STALE_RUN` when the criterion 8b or hierarchy wire evidence names a fork commit the pin does not, line 81 does the same for the rebase drill's `forkHead`, and `spec-250-evidence-collector.test.ts:95` runs that collector with `--check` against the real repository tree and requires exit 0. The pin and the three runs that describe it are one fact and have to move together. So this is `tools/t3-codegen/REFRESH.md` steps 3-8, minus nothing: - `pin.commit` / `commitDate` moved. `upstreamBase` is untouched — upstream has not moved, and `classify-churn --upstream-movement` reports the same 2 undecidable and 3 source-only commits it did before. - Regenerated. The output is byte-identical apart from the sha, which is what a Sidebar-only fork commit should produce: the pinned closure is `packages/contracts/src` and nothing there changed. Checked rather than assumed. 0 unrepresented, so nothing Codev consumes lost its schema. - The four evidence runs re-run against the new head, because REFRESH.md is explicit that this is the step that gets forgotten — a fork commit touching no closure file changes nothing but a sha, so regeneration looks like the whole job while the acceptance evidence goes on describing the previous fork. criterion 8b passed; hierarchy wire 8 of 8; the rebase drill still reports `conflicts` on the same 3 files; `collect --check` exits 0. - 35 patches re-exported, because they are cut against `pin.commit` and exist to be read by a reviewer without the private repository. `250-acceptance-evidence.md` gains a note naming which prose rows still describe `3786b840e1a4` or `2f64a1b0ee2b` and were **not** re-run. Rewriting a commit inside a historical record without re-running it would turn a record into a claim. Verified on this branch before opening: `t3-server.mjs verify` exits 0 on both identities, `collect-spec-250-evidence.mjs --check` exits 0, and `spec-250-vendoring-identities.test.ts` + `spec-250-evidence-collector.test.ts` are 83 passed. Refs #272, #250. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
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 freeto join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Unblocks the shared suite for every builder on this machine.
pin.jsonnamed2f64a1b0ee2bwhile the fork checkout has been at26b4c2dc09f0since #272's sidebar change landed there, socriterion 8b, the kill testandt3-server.mjs verifywere red for everyone.Why it is not the one-file fix it looks like
Updating only
codev/research/250-criterion-8b-evidence.jsonswaps one red test for another:tools/t3-server/collect-spec-250-evidence.mjs:74-79failsSTALE_RUNwhen the criterion 8b or hierarchy wire evidence names a fork commit the pin does not; line 81 does the same for the rebase drill'sforkHead.packages/codev/src/__tests__/spec-250-evidence-collector.test.ts:95runs that collector with--checkagainst the real repository tree and requires exit 0.The pin and the three runs that describe it are one fact and move together.
What is here
tools/t3-codegen/REFRESH.mdsteps 3-8, minus nothing.pin.commit/commitDate26b4c2dc09f0.upstreamBaseuntouched — upstream has not movedpackages/contracts/srcand nothing there changed. 0 unrepresentedconflictson the same 3 filespin.commit250-acceptance-evidence.mdgains a note naming which prose rows still describe an earlier fork head and were not re-run. Rewriting a commit inside a historical record without re-running it would turn a record into a claim.Verified on this branch before opening
node tools/t3-server/t3-server.mjs verify→ exit 0, both identitiesnode tools/t3-server/collect-spec-250-evidence.mjs --check→ exit 0spec-250-vendoring-identities.test.ts+spec-250-evidence-collector.test.ts→ 83 passedRefs #272, #250.
🤖 Generated with Claude Code