Record v3.11.2, and what it says about these records - #77
Merged
Conversation
Written after its own tag, as the document requires, and on the not-re-audited pattern: no four-corpus run, because nothing in the release touches detection or conversion policy. The new claim this record can make is that the nine GUI phases ran inside the release job, against the published executable, in the same run that produced the archives it quotes. Until now every record could say only that they passed somewhere before the tag. The suite was evidence a person chose to gather; it is now evidence the pipeline cannot skip. The record also says what this release means for the two before it. The drive-root defect was live while both were written, both reported it closed, and neither caught it, because both trusted a summary rather than the source. Nothing in the audit method would have caught it either: no corpus run reaches the plan boundary and no phase covers a drive-root base path. An audit document that cannot say this about itself is not worth keeping. Four limits stated: no corpus measurement, code signing skipped so the suite drove an unsigned binary, one fix no test can demonstrate, and a refusal covered by a unit test rather than the GUI phase that was attempted and abandoned. Commit and both archive digests were checked against the tag object and the live release rather than copied from earlier output. Co-Authored-By: Claude Opus 5 <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 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Adds the v3.11.2 record to
docs/SAFETY-AUDIT.md, after its own tag as thedocument's rule requires — any commit changes the assembly, so a record written
before the tag would describe a build nobody shipped.
Written as "not re-audited"
No four-corpus run: the checklist requires one for a release that changes
detection or conversion policy, and this changes neither. The record says so in
its first line, quotes no assembly hash, and disclaims the v3.11.0 figures.
The claim no earlier record could make
The nine GUI phases ran inside the release job, against the published
executable, in the same run that produced the archives the record quotes — 646
tests and nine phases, between packaging and publication. Every earlier record
could say only that they passed somewhere before the tag.
The suite was evidence a person chose to gather. It is now evidence the pipeline
cannot skip.
What the record says about itself
The drive-root defect shipped broken in v3.11.0 and v3.11.1. Both of those
records were written while it was live, both reported it closed, and neither
caught it — because both trusted a summary of what had been fixed rather than
the source.
Nothing in the audit method would have caught it either: no corpus run reaches
the plan boundary, and the GUI suite has no phase for a drive-root base path. It
was rediscovered by accident during an unrelated review.
That is in the record. An audit document that cannot say this about itself is
not worth keeping.
Limits stated
drove an unsigned published executable. The claim that it drives the signed
binary remains unproven.
code purely to make one fail.
was attempted and abandoned over a driver defect, and both the defect and the
fix were reproduced by hand.
Checked, not copied
The commit was read from the tag object and both digests from the live release:
Documentation only — no code changes.
🤖 Generated with Claude Code