Skip to content

docs(diary): record the PR-pipeline incidents behind the merged stack - #77

Merged
ManSio merged 1 commit into
mainfrom
docs/pr-pipeline-posts
Oct 10, 2026
Merged

ManSio merged 1 commit into
mainfrom
docs/pr-pipeline-posts

Conversation

@ManSio

@ManSio ManSio commented Oct 10, 2026

Copy link
Copy Markdown
Owner

What

Diary record of the three problems behind the merged migration stack — none of which were code defects:

  1. feat(index): E26 file-level chunk augmentation behind flag (default OFF) #73 turned CONFLICTING after fix(search): graceful fast-mode fallback when embedder is down #72 landed on main. The forward merge resolved AGENT_DIARY.md as a union automatically; the engine.py auto-merge was read by hand and correctly kept both the E24 fast-fallback try/except and the new MSCODEBASE_DENSE_OFF condition.
  2. "No CI" was a trigger bug, not a code bug. pull_request: branches: [main] meant stacked PRs produced zero runs, so phases 2 and 3 shipped with a local clean-state run as their only evidence. Fixed in ci: run CI for stacked PRs whose base is a feature branch #76.
  3. fix(search): module-doc chunker + P3 anchor + reranker scale + measurement hygiene (P3 gate GREEN) #58 sat CONFLICTING for 11 days with 1282 lines of code and 3 test files never reaching main. Revived with a forward merge keeping both sides of all three conflicts; 2097 tests green; merged.

Two near-misses also recorded

  • A git diff hunk made it look like the P3 anchor call sat after return pool (dead code). The diff had glued two hunks together; the file was correct. An AST check before editing prevented "fixing" a bug that did not exist.
  • The portfolio purity guard caught three paraphrase collisions caused by the new lab corpus (retrieval/score/results, vector/matching, produced/written). Fixed by rewording the experiment text, not the test — the 0/8 false-paraphrase baseline is preserved.

Verification

verify_diary.py: gate-zero 2096 passed, 10 skipped, 34 entries, 0 failures. Ruff clean, all pre-commit hooks pass.

Three problems that had nothing to do with the code being green:
- #73 turned CONFLICTING after #72 landed on main; the engine.py auto-merge was verified by hand (E24 fallback + MSCODEBASE_DENSE_OFF both survived).
- The missing CI was a trigger bug, not a code bug: pull_request only matched main, so stacked PRs produced no run at all. Fixed in PR #76.
- #58 sat CONFLICTING for 11 days with 1282 lines of code and 3 test files never reaching main; revived by a forward merge that keeps both sides of all three conflicts (2097 tests green, PR merged).

Also records two near-misses worth keeping: the diff that looked like dead code was two hunks glued together (AST check before editing prevented a fix to a non-bug), and the portfolio purity guard caught three paraphrase collisions caused by the new corpus - reworded the text, never the test.
@coderabbitai

coderabbitai Bot commented Oct 10, 2026

Copy link
Copy Markdown

Important

  • 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration
  • Configuration used: Organization UI
  • Review profile: CHILL
  • Plan: Advanced
  • Run ID: 4e8fec2b-ccf9-4ede-ae4b-1c6bfd3dd429

  • Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

Comment @coderabbitai help to get the list of available commands.

@ManSio
ManSio merged commit 1557e53 into main Oct 10, 2026
12 of 13 checks passed
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