Skip to content

chore(release): fill v0.62's step-8 attestation with the measured result (0-line diff) - #1160

Merged
avrabe merged 1 commit into
mainfrom
chore/v062-attestation
Sep 6, 2026
Merged

chore(release): fill v0.62's step-8 attestation with the measured result (0-line diff)#1160
avrabe merged 1 commit into
mainfrom
chore/v062-attestation

Conversation

@avrabe

@avrabe avrabe commented Sep 6, 2026

Copy link
Copy Markdown
Contributor

Two pre-tag obligations for v0.62.0, batched so #1158's 64-job CI was not restarted for prose.

1. Step-8 attestation — measured, not asserted

loop_conformance_check labels PR-head-vs-merge ATTESTED in both pretag and retro mode — squash-merge head refs are not durably fetchable at audit time — and directs the result into the step-7 review record. So filling it is a deliberate act, and it is now filled with the real measurement:

PR head  59441a80d40853346a6cea0030fbe05de9394510
merged   569ca7559da38b1d2c421e7c3f7c451a06b60ba3
git diff <PR head> <merged>  =  0 LINES

What CI verified, what the independent cold review audited, and what the v0.62.0 tag will name are byte-for-byte the same tree.

Recorded beside it: the four merge-ritual checks as they actually stood (9/9 required contexts green by name from branch protection; baseRefOid == origin/main at 5e6bc4400297; zero non-advisory reds across 64 checks — 63 pass, 1 skipped; status_evidence_check exit 0, 2 delivery-shaped commits — 2 attributed) and the CONFORMS verdict that preceded the merge.

An attestation with nothing behind it is precisely the vacuity this release spent its scope finding — a floor 1,892 short of its declaration, a rule blind to an unattributed commit, a sentence nobody checked. Shipping those findings while leaving this slot empty would have been incoherent.

2. A number that goes stale at tag time

Caught on a final read-through. The notes said "the release commit itself adds a further +213/−51" — true when 0a8c1e1f was the only release commit, wrong once there were four, wrong again after this one.

The fix is not a fresher figure but no figure: state the scope (feature merges vs release-assembly commits) and drop a number that keeps moving until the instant of tagging. Same treatment as the 17-hour figure and the v0.59 subcounts — attributed or scoped, never asserted.

Gates: claim_check 58/58 · status_evidence_check exit 0.

Refs #1136, #1158

…ult, and drop a number that goes stale at tag time

Two pre-tag obligations, batched into one PR so #1158's 64-job CI was not
restarted for prose.

1. STEP-8 ATTESTATION, MEASURED. loop_conformance_check labels
   PR-head-vs-merge ATTESTED in BOTH pretag and retro mode (squash-merge head
   refs are not durably fetchable at audit time) and directs the result into
   the step-7 review record. Filling it is therefore a deliberate act:

     PR head  59441a8
     merged   569ca75
     git diff <PR head> <merged>  =  0 LINES

   So what CI verified, what the independent review audited, and what the
   v0.62.0 tag will name are byte-for-byte the same tree.

   Recorded alongside it: the four merge-ritual checks as they actually stood
   (9/9 required contexts green BY NAME from branch protection, baseRefOid ==
   origin/main at 5e6bc44, zero non-advisory reds across 64 checks,
   status_evidence exit 0 with 2 delivery commits / 2 attributed) and the
   CONFORMS verdict that preceded the merge.

   An attestation with nothing behind it is exactly the vacuity this release
   spent its scope finding — a slack floor, a rule blind to an unattributed
   commit, a sentence nobody checked. Leaving this one empty while shipping
   those findings would have been incoherent.

2. A NUMBER THAT GOES STALE AT TAG TIME. Caught on a final read-through of
   the notes. The entry said "the release commit itself adds a further
   +213/-51" — true when 0a8c1e1 was the ONLY release commit, wrong now that
   there are four and wrong again after this one. The fix is not a fresher
   figure but NO figure: state the scope (feature merges vs release-assembly
   commits) and drop a number that keeps moving until the instant of tagging.
   Same discipline as attributing the 17-hour and v0.59-subcount figures
   rather than asserting them.

Refs #1136, #1158

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01YJK5LZZEkV5smCY1jKn18L
@codecov

codecov Bot commented Sep 6, 2026

Copy link
Copy Markdown

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

@avrabe
avrabe merged commit fdcefa6 into main Sep 6, 2026
60 checks passed
@avrabe
avrabe deleted the chore/v062-attestation branch September 6, 2026 10:30
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant