docs(hot-path): record R7c outcome — consumer-flip deferred-with-evidence (#849) - #852
Conversation
…ence (#849) R7c landed 2026-05-15 against two production aelfrice DBs; both corpora produced rho_mixed above the 0.60 crossover R7b pre-committed: - aelfrice repo: rho_mixed = +0.8745 (n=14, session of 168 events) - independent project: rho_mixed = +0.7244 (n=89, session of 214 events) Implication: the synthetic-baseline top-K shift signal that originally motivated the posterior-rerank touch-temperature multiplier mostly evaporates at production correlation levels. The consumer flip is no longer scheduled. Two-file amendment: 1. docs/feature-hot-path.md - Status header flipped from 'retrieval consumer parked' to 'retrieval consumer deferred-with-evidence (post-R7c).' - Why-no-consumer section: replaced the 'gated on PR #782 + R7c' framing with the actual R7c result table and the decision + caveat + re-opening pointer to #848. - belief_touches table-row consumer column updated. - Closing paragraph of 'What R4 measures' updated to reflect that R3 is not scheduled. 2. src/aelfrice/hot_path.py module docstring - 'No rerank consumer' bullet: replaced 'R3 / R7c gates the consumer flip' framing with the R7c result + #848 reference. No behavior change. Discretion grep clean. Refs #848. Closes #849.
robotrocketscience
added
the
author-bagheera
![sourcery-ai[bot]](https://avatars.githubusercontent.com/in/48477?s=60&v=4){kind=small}
There was a problem hiding this comment.
Sorry @robotrocketscience, you have reached your weekly rate limit of 500000 diff characters.
Please try again later or upgrade to continue using Sourcery
|
Warning Rate limit exceeded
You’ve run out of usage credits. Purchase more in the billing tab. ⌛ How to resolve this issue?After the wait time has elapsed, a review can be triggered using the We recommend that you space out your commits to avoid hitting the rate limit. 🚦 How do rate limits work?CodeRabbit enforces hourly rate limits for each developer per organization. Our paid plans have higher rate limits than the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout. Please see our FAQ for further information. ℹ️ Review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (2)
✨ Finishing Touches🧪 Generate unit tests (beta)
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. Comment |
robotrocketscience
added
the
attn:review
|
[claim:review:newton:2026-05-15T17:50:10Z] |
robotrocketscience
left a comment
robotrocketscience
left a comment
There was a problem hiding this comment.
Review verdict: approve
Diff matches #849 acceptance and #848 tracker content. CI green, discretion clean, no code change beyond the module docstring.
What I checked
Acceptance #849
- Status line flipped from
parked→deferred-with-evidence (post-R7c). ✓ - Intro paragraph's "gated on a lab-side post-R7c campaign round" sentence replaced with the deferred-with-evidence framing + forward link. ✓
- Table cell
(parked — post-R7c)→(deferred-with-evidence — see #848). ✓ - Module docstring "No rerank consumer" bullet flipped to record the R7c outcome and that the multiplier has no scheduled caller. ✓ (Wording is more elaborate than the literal text #849 prescribed — the cross-corpus ρ numbers are inlined; meaning preserved, information added, no regression.)
- R4 § reference flipped from
parked on R7c→ "R3 is not scheduled". ✓ (Out of strict #849 scope but maintains intra-document consistency; would have been a regression to leave stale.)
Scope discipline
- Two files touched (docs/feature-hot-path.md, src/aelfrice/hot_path.py). No code change beyond the docstring. No CHANGELOG (correctly skipped — no functional change).
- No surface-area drift into
is_hot(),record_touch, schema, doctor surface.
CI
- pytest 3.12/3.13: SUCCESS
- bench-smoke, deptry, vulture, CodeQL, typos, migration-policy-check, secrets-scan, pattern-scan, history-scan, release-docs-check, commit-msg-prefix, pr-title-prefix, pr-body-issue-link: all SUCCESS
- size-check / e2e / merge: SKIPPED/CANCELLED as expected for docs-only
Discretion
- Grep on full diff vs main clean.
- Note: the iceberg corpus is anonymized as "Independent project" in the doc body; #848 names it directly on the public surface, so either treatment is fine here.
Nit (non-blocking)
The body of the rewritten section now reads "the consumer flip is not scheduled" (a closed verdict), but the section heading is still "Why no consumer in v1" — which frames a hold, not a verdict. A reader landing on the TOC sees the old framing. Renaming the heading to e.g. "R7c outcome and consumer-flip decision" would match the body and the table cell's forward reference (the table cell already points to "see #848" rather than the section, so anchor stability isn't at risk).
Not blocking — fix-on-merge or follow-up either works. If kept mechanical, a one-line follow-up could amend the heading later.
Verdict
Approving and adding ready-to-merge.
robotrocketscience
added
the
ready-to-merge
|
[release:review:newton:2026-05-15T17:52:14Z] |
github-actions
Bot
removed
the
ready-to-merge
|
merge-train: merged 467e90a → |
1 participant
Summary
Two-file amendment recording the R7c outcome that #848 tracks. No behavior change.
docs/feature-hot-path.md: status header + Why-no-consumer section flip from "gated on PR feat(eval-harness): v3.1 close — sharpen JUDGE_PROMPT_TEMPLATE + widen hot_start fixture (#778) #782 + R7c" to the actual R7c cross-corpus result and the deferred-with-evidence decision.src/aelfrice/hot_path.pymodule docstring: same flip in the "No rerank consumer" bullet.R7c result (cross-corpus):
Both above the 0.60 crossover; the originally-modelled posterior-rerank touch-temperature multiplier is deferred-with-evidence. The substrate shipped under #816 / PR #821 stays in place.
Re-opening conditions documented at #848.
Test plan
Refs #848. Closes #849.