Conversation
Codex P2 + Copilot P1 both flagged that `_patterns.md` was classified as "current-state operational substrate" while sitting in `docs/pr-preservation/**` (which the file itself classifies as history-class). Internal contradiction. Resolved by introducing a third named surface class: **synthesis-over-history** — current-state-tracking abstractions over history-class corpora. Co-located with the corpus they abstract over (so `_patterns.md` stays in `docs/pr-preservation/` alongside the per-log records). The underscore prefix is named as a candidate convention for "synthesis-indices over history-class corpora." Updated three sections: 1. "Surface-class taxonomy" — added "Synthesis-over-history surfaces" section with `_patterns.md` as the canonical example. 2. "How to update this file" — replaced the "current-state operational substrate" claim with the synthesis-over-history distinction + co-location rationale. 3. Closing line — clarified Otto-279 carve-out applies to the per-log records (history-class); synthesis-over-history is the third class distinct from both pure history-class and pure current-state. Composes with the future BP-NN promotion path: the four-stable- outcome-classes + history-class surfaces + synthesis-over-history third class are mature enough to codify together in `docs/AGENT-BEST-PRACTICES.md` as a single stable rule. Other #448 reviewer findings (cited drain-logs not in current corpus) are stale-resolved-by-reality — drain-logs are landing across PRs #437-#465; the corpus is converging.
There was a problem hiding this comment.
Pull request overview
Resolves an internal surface-class taxonomy contradiction in docs/pr-preservation/_patterns.md by introducing a third surface class (“synthesis-over-history”) to distinguish synthesis indices from the underlying history-class drain-log corpus.
Changes:
- Adds “synthesis-over-history surfaces” as an explicit third surface class in the taxonomy.
- Updates the “How to update this file” guidance to clarify co-location and the per-file carve-out.
Merged
4 tasks
Codex/Copilot caught 4 findings on PR #466's surface-class clarification: - Three hyphen-line-wrap renderings (`synthesis-` / `synthesis-over-` / `synthesis-` at end-of-line render with extra space when Markdown treats newline as whitespace). Reflowed all three to keep the term contiguous on one line. - Surface-class taxonomy still listed `docs/pr-preservation/**` as history-class which would include `_patterns.md`. Tightened to the per-log glob `docs/pr-preservation/*-drain-log.md` with explicit pointer to the synthesis-over-history exception below. Same shape as the inline-code-span line-wrap class documented in `_patterns.md` (Class A in PR #465 BACKLOG row) but for prose-line hyphenation rather than code-span line-wrap. Worth noting in a future expansion of the doc-lint suite — extend the regex to catch `\w-$` end-of-line patterns.
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: cb62527eb2
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".
| - `docs/ROUND-HISTORY.md` | ||
| - `docs/aurora/**` | ||
| - `docs/pr-preservation/**` | ||
| - `docs/pr-preservation/*-drain-log.md` (per-log records; |
There was a problem hiding this comment.
Narrowing docs/pr-preservation/** to docs/pr-preservation/*-drain-log.md drops existing provenance docs like docs/pr-preservation/141-ci-fix-log.md out of history-class, so that file now implicitly falls under current-state rules even though it is the same point-in-time record type (names, checks, and audit trail). This reintroduces taxonomy drift for review tooling and attribution guidance; the history-class selector should include all preservation logs in this directory (or explicitly include the CI-fix log shape).
Useful? React with 👍 / 👎.
AceHack
added a commit
that referenced
this pull request
Apr 25, 2026
…iated shape divergence (#467) * hygiene(_patterns.md): document Otto-250-canonical vs Otto-268-abbreviated drain-log shape divergence Multiple post-merge reviewer threads on the Otto-268-wave drain-logs (#437-#465) flagged that the abbreviated inline shape doesn't match the canonical Otto-250 multi-section format used in older drain-logs (#108, #247, etc.). The findings are correct: the shapes ARE different, and the abbreviated shape compresses verbatim reviewer/reply text into short prose. Documenting both shapes here in `_patterns.md` as a known divergence + three maintainer-decision options: (a) Rewrite — high churn, highest faithfulness. (b) Accept divergence — document both as valid. (c) Hybrid — canonical for substantive logs (math / crypto / algorithms / ADRs); abbreviated for low-substance. Until maintainer decides, drain-runners writing future logs default to canonical; existing Otto-268-wave logs stay abbreviated with this known-divergence pointer. This is the right place for the documentation per the synthesis- over-history surface-class declaration: `_patterns.md` reflects current corpus state including known shape divergences, separate from the per-log records. Composes with the per-PR drain-log threads currently flagging the shape-conformance gap on #437/#441/#442/#444/#445/#446/#447/#449/ #460/#461/#464/#466 — those threads can now reply with a one-line pointer to this section instead of repeating the maintainer- decision-pending framing. * hygiene(#467): fix 4 Copilot findings on shape-divergence section Four Copilot catches converging on the same root cause: the shape-divergence section was citing drain-log PR-numbers (#437-#465) when readers will look for #437-drain-log.md etc. and not find them — drain-log FILE numbers reference the PRESERVED PR (e.g., #421/#422/#423), not the drain-log PR itself. The current file corpus ends at #434. Fixes: - :374 + :394 — drop the misleading "#437-#465" PR-range citation; cite the actual in-repo abbreviated-shape examples (#421/#422/#423) by file path; drop the unsupported "22+" numeric estimate in favour of "the 2026-04-25 backfill wave". - :380 — abbreviated template snippet now matches what the in-repo abbreviated logs actually use: `Finding:` bullet included, `Thread ID:` and `:LINE` placeholders dropped (those are canonical-shape fields, not abbreviated-shape). Reader can now copy/paste the template accurately. - :388 — soften the "Substance is preserved" overstatement to the objective claim: key metadata (reviewer/severity/outcome/ commit) is preserved; verbatim original-comment + verbatim reply text are NOT preserved; multi-section structure + file:line locator + Thread ID are also typically omitted. The fixes are themselves Class A (count-vs-list cardinality on "#437-#465" range) + truth-drift on a freshly-landed doc — appropriate self-application of the discipline this section documents.
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
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.
Summary
Codex P2 + Copilot P1 both flagged a real internal contradiction on #448's
_patterns.md: the file declareddocs/pr-preservation/**as history-class (per the surface-class taxonomy section), then later claimed_patterns.mditself is "current-state operational substrate" (in the "How to update this file" section). Both can't be true — directory-level surface-class declaration conflicts with file-level reclassification.Resolved by introducing a third named surface class: synthesis-over-history — current-state-tracking abstractions over history-class corpora.
Coverage
docs/pr-preservation/*-drain-log.mdstays history-class (per-log records preserve provenance, Otto-279 carve-out applies).docs/pr-preservation/_patterns.md(this file) is synthesis-over-history — a third surface class distinct from both pure history-class and pure current-state._patterns.mdstays indocs/pr-preservation/); the underscore prefix is named as a candidate convention.Three sections updated
_patterns.mdas the canonical example.Stale-resolved-by-reality (other #448 reviewer findings)
The remaining #448 reviewer findings cite drain-log PRs that don't yet have logs in current main (#191, #219, #231, #377, #135, #85, #195, #206, #427, #428, #430, #425, #426, #432, #435). Those drain-logs are landing across PRs #437-#465 (Otto-268 backfill is the in-flight wave). The corpus is converging — those findings will be stale-resolved-by-reality as the cited PRs land in main.
Test plan
🤖 Generated with Claude Code