docs: merge authority stays with the owner (OD-0020); record the unfinished test pruning - #178
Merged
Merged
Conversation
…t pruning Two things from the 2026-07-29 session that would otherwise be lost. Merge authority (OD-0020). The 2026-07-18 grant was real and was used once, on PR #163, but it lived only in an untracked memory file while CLAUDE.md and roster.json kept listing merging as ask-first. The estate held two contradictory answers for eleven days with nothing able to detect it. The owner resolved it by narrowing the grant: he retains merge authority while studio-2.0 is the staging target, because the risk is a correct-looking change landing on the wrong base. A per-task authorization expires with that task and is never read forward. He will revisit once studio-2.0 lands in main. Everything up to a reviewed, mergeable PR stays do-then-report, so this costs one message, not velocity. Unfinished test pruning, added to REMAINING_TASKS.md. About 40 low-value tests from the 2026-07-10 audit are still in the suite. The audit landed and most of its execution branches merged, but three did not, and they are on origin/studio2/phase-12.5-style rather than lost. Cherry-picking is not the way in: all three collide in the same three VoiceLab test files, which have moved on since. The entry says to redo the deletions against today's files and re-check each against R1-R4 first, since a test that was low-value in July may be load-bearing now. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
…nciliation Reconciling 32 abandoned branches, a first pass matched commit subjects against studio-2.0's history and reported all 32 as unlanded. Nearly all had landed. rev-list --cherry-pick fails the same way, since a squash rewrites the patch-id of the whole branch into one commit, and reachability inflates the count because N branches sharing one squashed trunk each read as N distinct bodies of work. Only content answers it: read what the commit changed, then read those same files as they stand on origin/studio-2.0. 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.
Two things from the 2026-07-29 session that would otherwise be lost when it ends. This one is yours to merge (which is rather the point of the first half).
Merge authority (OD-0020)
The 2026-07-18 grant was real, and it was used once, on PR #163. But it lived only in an untracked memory file, while
CLAUDE.mdandroster.jsonboth kept listing merging as ask-first. So the estate carried two contradictory answers for eleven days with nothing able to detect the conflict, until the retrofit surfaced it.You resolved it by narrowing the grant rather than denying it: merge authority stays with you while
studio-2.0is the staging target, because the risk being managed is a correct-looking change landing on the wrong base. A per-task authorization covers that task and expires with it, and is never read forward as standing permission. You'll revisit oncestudio-2.0lands inmain, and the OD says plainly that revisiting means asking you, not assuming.Worth noting the cost is small: everything up to a reviewed, mergeable PR is still do-then-report, so this is one message at the end rather than a slower pipeline.
The corresponding memory file has been rewritten to match, so a future session can't read the old grant and act on it.
The unfinished test pruning
Added to
REMAINING_TASKS.mdso it survives the session. Roughly 40 low-value tests from the 2026-07-10 audit are still in the suite. The audit landed and most of its execution branches merged, but three never did. They're onorigin/studio2/phase-12.5-style, so the work is pushed and safe, just unmerged.The entry records the trap: don't cherry-pick, because all three collide in the same three VoiceLab test files, which have moved on in three weeks. Redo the deletions against today's files, using each commit as the list of what to remove, and re-check each against
testing-standards.mdR1-R4 before deleting (a test that was low-value in July may be load-bearing now). Roughly an hour, and worth it since the vitest suite leaks memory.How to verify
Docs and config only, no code.
roster.jsonparses, the decision index regenerates to 20 records, and OD-0020'sDecision:field quotes the new ask-first line verbatim.