docs(phi-scan): the branch-and-merge fixture went red on a committer identity, not an empty index - #111
Merged
Merged
Conversation
…identity, not an empty index The note and the test comment covering the unmerged-index case both explained the red CI run as `git ls-files -s` coming back with no records for the path at all. Measured FALSE. The run's own output reads `expected [ Array(1) ] to have a length of 3 but got 1`, and the `gitIn` calls above it assert exit 0 on the add and on the commits, so a stage-0 record necessarily existed. The draft handed its `git merge` no committer identity. The merge died before it touched the index and left the ordinary stage-0 record the last commit wrote, and the premise assertion, `.not.toBe(0)` written to mean "the merge really conflicts", accepted that crash as a conflict. A premise assertion that accepts any non-zero exit accepts a crash, which is the transferable lesson the false clause was displacing. Also drops "so the next failure names its own cause" from the same paragraph: the premise captures `listed.stdout` only and never its status or stderr, so a failing `ls-files` still hides its own cause. Disclosed, not grown. No behavior change. `scripts/phi-scan.ts`, the fixture and every assertion are byte-identical; the test edit is comment text only.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What
Deletes one false clause from the two carriers that explained why the first (branch-and-merge) version of the unmerged-index fixture went red in CI, and states the measured cause instead.
Both carriers said the run came back with no records for the path at all:
documentation/agent-notes.mdtest/scripts/phi-scan.test.ts(comment)Why it is false
The red run (
31478670886, both runners) printed:One record, not zero. It is also impossible on its own terms: the
gitIncalls above the merge assert exit 0 ongit add c.xmland on all the commits, so a stage-0 record necessarily existed.The real cause
The draft handed its
git mergeno committer identity (the commits got one via-c, the merge did not). A merge with no resolvable identity dies before it touches the index, leaving the ordinary stage-0 record the last commit wrote. Reproduced here: exit 128,Committer identity unknown, one stage-0 record, for both conflicting and cleanly-auto-merging sides.The premise assertion
expect(merged.status, "premise: the merge really conflicts").not.toBe(0)passed on that 128. A premise assertion that accepts any non-zero exit accepts a crash - that is the transferable lesson the false clause was displacing, andrepoWithConflictalready hands its own merge an identity for exactly this reason.Also
Drops "so the next failure names its own cause" from the same paragraph. The premise captures
listed.stdoutonly and never its status or stderr, unlike every other call in the case, so a failingls-filesstill hides its own cause. Disclosed, not grown: the assertion is deliberately unchanged.Scope
No behavior change.
scripts/phi-scan.ts, the fixture and every assertion are byte-identical; the test-file edit is comment text only.CLAUDE.mduntouched.conformance-refuterran and returned REFUTED on the first draft of the replacement prose (two introduced claim defects: a wrong statement that the file never passes identity to a merge, and a prose record count). Both were cut and re-checked against the file and the CI log firsthand before this commit.