Conversation
…ating so PRs only run relevant checks (Aaron 2026-05-01; third B-0177 audit hit) Aaron 2026-05-01 named "B-0157 detect-changes pattern" in the prefer-mechanical-external-anchors memo + provided full pattern source via direct sibling-repo inspection. ID was reserved 2026-05-01; per-row file was never filed. Otto 2026-05-03 audit pass found B-0157 still missing — third concrete hit for B-0177's audit hypothesis (B-0141, B-0142, B-0157 all from same-period memos, all reserved-but-never-filed). Pattern from sibling repo (`../no-copy-only-learning-agents-insight`): detect-changes.yaml runs first, emits per-change-class outputs (dotnet/fsharp/tools-ts/docs/etc.); downstream workflows gate on those outputs. PRs only run workflows their changes need. M-effort: detect-changes.yaml (~50-100 LOC) + per-workflow gating (~15-42 workflows) + test fixtures + required-checks audit. Composes with: B-0125 (specific instance), B-0153 (lint suite), B-0156 (TS port), B-0177 (audit hypothesis empirical hit #3).
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 231e72bed7
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
Pull request overview
This PR files backlog row B-0157 and adds it to the generated backlog index. It documents a proposed CI pattern where a first-pass workflow classifies changed files so downstream workflows only run when their change class is relevant.
Changes:
- Adds a new P2 backlog row for the detect-changes workflow-gating pattern, including origin, scope, and composition links.
- Updates
docs/BACKLOG.mdto include the new B-0157 entry in the P2 index.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 6 comments.
| File | Description |
|---|---|
docs/backlog/P2/B-0157-detect-changes-pattern-parallel-pr-friendly-aaron-2026-05-01.md |
New backlog row describing the detect-changes CI-gating proposal and its lineage/composition. |
docs/BACKLOG.md |
Adds the B-0157 index entry so the backlog index reflects the new per-row file. |
…n + partial pre-existence acknowledged (#1357) Three substantive review findings absorbed: 1. Cross-workflow `needs:` is invalid — `needs:` works only within single workflow file. Architecture corrected to specify three options (Architecture A: reusable workflow via workflow_call; Architecture B: per-workflow detect-job; Architecture C: single consolidated workflow). Architecture A recommended (matches sibling-repo pattern + preserves separation). 2. .yaml → .yml extension throughout — repo's existing workflows consistently use .yml; one-off .yaml suffix would create drift. 3. Problem-statement acknowledged partial pre-existence — gate.yml already skips on docs-only via path-gate; codeql.yml short-circuits pure docs; backlog-only changes have backlog-index-integrity. Row generalizes the partial pattern into uniform mechanism, NOT from-zero gating. The B-0142-stale-cross-references in some threads were merge-ordering false positives (B-0142 not present at PR-eval time; merged later as #1349) — handled in resolve-thread comments.
Summary
Files `B-0157` — third concrete hit for B-0177's audit hypothesis. Aaron 2026-05-01 reserved the ID in the prefer-mechanical-external-anchors memo + provided full pattern source via sibling-repo direct inspection. The row never had a per-row file until now.
Audit pattern emerging
All three from 2026-05-01 memos; all three reserved-but-never-filed; all three surfaced by Otto 2026-05-03 audit pass post-Aaron's "a lot of backlog lost in our memories" observation. B-0177's hypothesis is empirically validated three-fold.
Row scope
The sibling-repo pattern from `../no-copy-only-learning-agents-insight`:
Effect: PRs only run workflows their changes need. CI throughput scales with relevance, not with PR count.
Composes with
Test plan