Apply the review findings deferred from #20 #24

Open
opened 2026-10-03 17:28:14 +00:00 by jercik · 0 comments
Owner

Six low findings from the review of #20 that were deferred, each valid. Details are in the review summary comment on head 41cd661: #20

  • src/reconcile/conversation-markers.ts:77, "404 warning buries the reply-only fallback behind an unverified cause list". Lead the warning with the effect (resolution returned 404 at the endpoint, this pass posts replies only), then list route availability, a deleted comment, and token access as things to check, not as "the cause".
  • src/reconcile/dispositions.ts:130, "Superseded-thread comment incorrectly says labelled open threads get no write". Reword the comment: an open, unlabelled thread gets one label, any open thread is resolved when the API is available, and closed threads get no writes.
  • src/reconcile/inline.ts:49, "Leading-marker comment claims authorship that the code never verifies". Replace the comment with "Treat only a leading marker as an idempotency marker; a quoted marker is text. This check does not verify the author."
  • src/reconcile/dispositions.test.ts:281, "404 degradation test misses repeated failed resolution attempts". Record resolution attempts or assert the mock and the warning are called exactly once, keeping the second conversation so the latch is exercised.
  • src/reconcile/dispositions.test.ts:385, "Supersession tests bypass the marker check with resolved anchors". Give at least one already-marked anchor resolver: null and assert it gets neither a second reply nor a resolution.
  • src/reconcile/inline.test.ts:272, "Head-file fetch test cannot distinguish a path-anchor fetch". Give the path and snippet anchors distinct paths, or add a path-only case that asserts zero getRawFile calls.
Six low findings from the review of #20 that were deferred, each valid. Details are in the review summary comment on head `41cd661`: https://code.j4k.dev/j4k-oss/review-wrapper/pulls/20 - `src/reconcile/conversation-markers.ts:77`, "404 warning buries the reply-only fallback behind an unverified cause list". Lead the warning with the effect (resolution returned 404 at the endpoint, this pass posts replies only), then list route availability, a deleted comment, and token access as things to check, not as "the cause". - `src/reconcile/dispositions.ts:130`, "Superseded-thread comment incorrectly says labelled open threads get no write". Reword the comment: an open, unlabelled thread gets one label, any open thread is resolved when the API is available, and closed threads get no writes. - `src/reconcile/inline.ts:49`, "Leading-marker comment claims authorship that the code never verifies". Replace the comment with "Treat only a leading marker as an idempotency marker; a quoted marker is text. This check does not verify the author." - `src/reconcile/dispositions.test.ts:281`, "404 degradation test misses repeated failed resolution attempts". Record resolution attempts or assert the mock and the warning are called exactly once, keeping the second conversation so the latch is exercised. - `src/reconcile/dispositions.test.ts:385`, "Supersession tests bypass the marker check with resolved anchors". Give at least one already-marked anchor `resolver: null` and assert it gets neither a second reply nor a resolution. - `src/reconcile/inline.test.ts:272`, "Head-file fetch test cannot distinguish a path-anchor fetch". Give the path and snippet anchors distinct paths, or add a path-only case that asserts zero `getRawFile` calls.
Sign in to join this conversation.
No labels
No milestone
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
j4k-oss/review-wrapper#24
No description provided.