The knowledge base's review knows its round: from round 3 only a wrong record asks changes, at round 5 the operator decides (issue 784) #648

2026-11-09T22:24:20Z

mesh-tools commented

2026-11-09 22:24:20 +00:00

Builds ADR 0665 (accepted 2026-11-09, approved by the operator) for issue 784: the knowledge base's reviewer did not know which round it was in, so it kept finding one more thing (research 086 took about twenty versions tonight).

The review's question names the round, lists the earlier rounds' findings, and from round 3 asks changes only for a finding that makes the record wrong (ADR 0665 §1, §2).
At round 5, changes become an ask of the operator; no review runs before the answer; a Decline restarts the count (§3).
A revision of a pending version, and a closing, keep the count (before, a revision restarted it, which let research 086 loop); a revision or a close is refused while a round-5 ask waits.
get shows the round of a version under review, and Waiting names a round-5 ask.
Tests per §4 in review_round_test.go; mesh-knowledge has 145 tests (budget 500).

Quick fix vs thorough: the round-3 rule is held by the review's question, not by code (the judge decides); enforcing it in code would change the review's answer schema. Chosen because the operator's ask at round 5 bounds what a judge that ignores the rule can cost.

Note for the operator before the delivery: records that already have more than five reviews since their last acceptance, research 086 among them, go to the operator at their next review that asks changes.

Reviewed-by: mesh-code-reviewer for fix-1965@shanks at c12f1a16db

https://claude.ai/code/session_01JG73KK8HD1gufnqyFxMFdW

Builds ADR 0665 (accepted 2026-11-09, approved by the operator) for issue 784: the knowledge base's reviewer did not know which round it was in, so it kept finding one more thing (research 086 took about twenty versions tonight). - The review's question names the round, lists the earlier rounds' findings, and from round 3 asks changes only for a finding that makes the record wrong (ADR 0665 §1, §2). - At round 5, changes become an ask of the operator; no review runs before the answer; a Decline restarts the count (§3). - A revision of a pending version, and a closing, keep the count (before, a revision restarted it, which let research 086 loop); a revision or a close is refused while a round-5 ask waits. - `get` shows the round of a version under review, and `Waiting` names a round-5 ask. - Tests per §4 in review_round_test.go; mesh-knowledge has 145 tests (budget 500). Quick fix vs thorough: the round-3 rule is held by the review's question, not by code (the judge decides); enforcing it in code would change the review's answer schema. Chosen because the operator's ask at round 5 bounds what a judge that ignores the rule can cost. Note for the operator before the delivery: records that already have more than five reviews since their last acceptance, research 086 among them, go to the operator at their next review that asks changes. Reviewed-by: mesh-code-reviewer for fix-1965@shanks at c12f1a16db83ba0cb2c33dda286e386b31464d04 https://claude.ai/code/session_01JG73KK8HD1gufnqyFxMFdW

mesh-tools added 3 commits 2026-11-09 22:24:20 +00:00

The knowledge base's review knows its round: from round 3 only a wrong record asks changes, at round 5 the operator decides (issue 784, ADR 0665) 62644aa012

Claude-Session: https://claude.ai/code/session_01JG73KK8HD1gufnqyFxMFdW

A closing keeps the review round, get says the round, a close waits for the operator's round-5 answer (issue 784, ADR 0665) 09ea4a40bf

Claude-Session: https://claude.ai/code/session_01JG73KK8HD1gufnqyFxMFdW

get shows a round only for a proposed version (issue 784, ADR 0665)

mesh/delivery merged: source.main moved to novox/mesh-catalog@9c42423d5821

Details

mesh/check checking c12f1a16db83 on 9c42423d5821

c12f1a16db

Claude-Session: https://claude.ai/code/session_01JG73KK8HD1gufnqyFxMFdW

mesh-tools commented

2026-11-09 22:24:26 +00:00

Delivery novox/mesh-catalog@c12f1a16db83 — merged since 2026-11-09T22:27:02Z

Source — source.main moved to novox/mesh-catalog@9c42423d58. Whether it reached the machines is mesh-delivery-view's, derived from source.main, the wants and the haves.

Transitions

2026-11-09 22:24 (new) → proposed (announced): novox/mesh-catalog#648's head announced
2026-11-09 22:24 proposed → ready (accepted): a verdict names this commit and the trunk's head as its trunk commit, the gate passed or warned, and the repository's own check did not fail; or the check gates no merge (check-gates-merge false)
2026-11-09 22:27 ready → merged (merged): on the trunk its modules follow: merged there

The commit's note under refs/notes/mesh-plan keeps every transition of this change up to its merge: git log --notes=mesh-plan.

**Delivery** `novox/mesh-catalog@c12f1a16db83` — **merged** since 2026-11-09T22:27:02Z **Source** — source.main moved to novox/mesh-catalog@9c42423d5821. Whether it reached the machines is `mesh-delivery-view`'s, derived from `source.main`, the wants and the haves. **Transitions** - 2026-11-09 22:24 (new) → proposed (announced): novox/mesh-catalog#648's head announced - 2026-11-09 22:24 proposed → ready (accepted): a verdict names this commit and the trunk's head as its trunk commit, the gate passed or warned, and the repository's own check did not fail; or the check gates no merge (check-gates-merge false) - 2026-11-09 22:27 ready → merged (merged): on the trunk its modules follow: merged there The commit's note under `refs/notes/mesh-plan` keeps every transition of this change up to its merge: `git log --notes=mesh-plan`.

mesh-tools merged commit 9c42423d58 into main

2026-11-09 22:26:59 +00:00

mesh-tools deleted branch fix/784-review-round

2026-11-09 22:26:59 +00:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: novox/mesh-catalog#648