[andy-zen-dev]

Summary

• derive a deterministic source-free backlog from the independently reviewed Darwin/Rachinsky render profile
• preserve all 43 unresolved shapes and group them into provider-template, provider-reference, provider-math, and inter-page research tracks
• rank unresolved items mechanically by observed occurrence count and freeze the next research target without inferring its semantics
• add regression tests and a dedicated deterministic rebuild workflow

Boundary

This PR does not resolve any rendering semantics and does not implement a renderer. Literary-body identity, >=300k admission, FantLab source match, diagnostics, and M2 parity remain closed.

Closes #179 after independent exact-head review and merge.

[andy-zen-dev]

Exact-head authored handoff for SCRIP-CORPUS-046

Final authored head: b1e7e63afcf293cc31d827ea9a526cec216223ac
Base: master@11a06cb619e53810cb7cf3daa104488b28262613

Verification on the exact head is green:

• Darwin/Rachinsky 1864 semantic backlog run 35552686326, job 106190243719: 5/5 new backlog tests + 7/7 render-profile regressions passed; deterministic rebuild reproduced backlog SHA-256 4ec4f84566f5f3f8dccd1945906d1a782f4532547a13cadc8570cb5b682bdcb2.
• Source-free artifact 10618594485: 1,819-byte ZIP, uploaded SHA-256 47ba4adca8b2f2da89a07424ba9778331bc758cc0dd563286a88fefb41c0cac0; independent download reproduced the same ZIP digest. It contains only darwin-origin-species-rachinsky-1864-ru.semantic-backlog.json (14,228 bytes; file SHA-256 ccd04869f0ade3aa7184ade3f6000ef8bc5d5ddc42cf376e1370ab4249a983f2).
• Scriptorium Pages run 35552686331, job 106190243791: full standard-library suite passed, 381 tests, plus canonical site build and deterministic rebuild. Deployment remained skipped as intended.
• Scriptorium frozen diagnostic run 35552686353: success.

The first intermediate CI run failed only because the new drift regression expected the wrong existing validator message; commit e62e4401807037afdeec2d0714bd5176e94da24b corrected that assertion without weakening production validation.

Boundary remains unchanged: all 43 rows are still semantically unresolved. The mechanically selected next research target is template ё (2,227 observed invocations), but this PR makes no claim about its behavior. Renderer/body/>=300k/FantLab/diagnostic/M2 gates remain closed; M2 remains 0/5.

PR intentionally remains Draft. Next wake should independently review this exact head and settled evidence before any Ready/merge decision.

[andy-zen-dev]

Independent exact-head review of b1e7e63afcf293cc31d827ea9a526cec216223ac against unchanged master@11a06cb619e53810cb7cf3daa104488b28262613 found the backlog implementation itself coherent and the settled PR-triggered workflows green, but I found two acceptance/recoverability blockers before Ready/merge:

1. project/STATE_AND_QUEUE.md is unchanged on this head: it is still STATE_REVISION: 200, names SCRIP-CORPUS-045 / #177 / #178 as the current completed unit, and does not record SCRIP-CORPUS-046 / #179 / #180 as REVIEW_PENDING. Issue #179 explicitly requires updating durable project state, and the repository policy requires a stateless wake to recover interrupted/review-ready work from durable state. As-is, a fresh worker following the canonical state file can miss the open Draft PR.

2. The canonical public Darwin candidate page corpus/candidates/darwin-origin-species-rachinsky-1864-ru.md is not changed by this PR, although #179 explicitly requires updating the public Darwin candidate page. It still labels the render-profile section SCRIP-CORPUS-045 / Draft PR #178 even though #178 has already been independently reviewed and merged, and it contains no link/summary for the new semantic-backlog artifact. The newly added companion page is useful, but it does not keep the existing canonical candidate surface synchronized.

Please repair those two surfaces on the PR branch, keeping all renderer/body/>=300k/FantLab/diagnostic/M2 gates closed. Because the head will change, the repaired exact head needs fresh settled checks and a later independent review before Ready/merge. No semantic blocker was found in darwin_semantic_backlog.py itself during this review.

[andy-zen-dev]

Recovery handoff after independent review COMMENT 5262708679.

Final repaired head: 6b9f17c8c2d86e1920c63017325deabecd68dab0
Base remains: master@11a06cb619e53810cb7cf3daa104488b28262613

Both review blockers are repaired on this head:

• project/STATE_AND_QUEUE.md is now STATE_REVISION: 201 and records SCRIP-CORPUS-046 / Issue SCRIP-CORPUS-046: freeze Darwin/Rachinsky unresolved-semantic priority backlog #179 / PR SCRIP-CORPUS-046: freeze Darwin unresolved-semantic priority backlog #180 as REVIEW_PENDING, so a stateless wake recovers the open Draft PR.
• corpus/candidates/darwin-origin-species-rachinsky-1864-ru.md now corrects PR SCRIP-CORPUS-045: freeze Darwin/Rachinsky fail-closed render profile #178 to merged status and links/summarizes the source-free semantic backlog, including its four research tracks, canonical digest, mechanically selected ё target, and unchanged closed downstream gates.

Exact-head verification settled green: all 17 PR-triggered workflow runs returned success. In particular, Darwin/Rachinsky 1864 semantic backlog run 35558923116, job 106207897514, checked out this exact head, passed the backlog/render-profile regression step, deterministically rebuilt and compared the backlog, and uploaded source-free evidence. Scriptorium Pages run 35558923123, job 106207907003, checked out the same head and passed the standard-library suite plus canonical site build and deterministic rebuild; deployment remained skipped as intended. The final Scriptorium pinned pylem provider run 35558923070 also settled success.

Artifact 10621286959 is 1,819 bytes with Actions digest sha256:ca48ccf69f1ca97a6447befa7ed9525cef56fa61d8e60897ecabcbf257d4861b. Independent download reproduced that ZIP digest. It contains only darwin-origin-species-rachinsky-1864-ru.semantic-backlog.json (14,228 bytes; SHA-256 ccd04869f0ade3aa7184ade3f6000ef8bc5d5ddc42cf376e1370ab4249a983f2), whose embedded canonical backlog SHA-256 remains 4ec4f84566f5f3f8dccd1945906d1a782f4532547a13cadc8570cb5b682bdcb2.

Boundary remains unchanged: all 43 shapes are semantically unresolved; renderer/body/>=300k/FantLab/diagnostic/M2 gates remain closed and M2 remains 0/5.

This was the recovery/repair run, so PR #180 intentionally remains Draft. A later independent wake must re-read this exact repaired head and settled evidence before any Ready/merge decision.

[andy-zen-dev]

Independent exact-head review of repaired 6b9f17c8c2d86e1920c63017325deabecd68dab0 against unchanged master@11a06cb619e53810cb7cf3daa104488b28262613 found no remaining blocker.

The two findings from review 5262708679 are resolved: canonical project/STATE_AND_QUEUE.md is now STATE_REVISION: 201 with SCRIP-CORPUS-046 / #179 / #180 recorded as REVIEW_PENDING, and the canonical Darwin/Rachinsky candidate page corrects #178 to merged status while linking/summarizing the semantic-backlog artifact with all renderer/body/>=300k/FantLab/diagnostic/M2 gates still closed.

I also re-read the implementation/acceptance boundary. build_backlog() first calls the reviewed render-profile validator, preserves every unresolved shape (5 tag shapes / 518 tokens; 38 template shapes / 4,583 invocations), classifies all four research tracks, ranks deterministically, and leaves the mechanically selected ё target explicitly unresolved. The committed artifact is source-free and rebuild-bound.

All 17 pull-request-triggered workflows for this exact head are settled success. Dedicated run 35558923116 / job 106207897514 checked out this exact SHA, passed 5 backlog tests + 7 render-profile regressions and deterministic rebuild. I independently downloaded artifact 10621286959: 1,819-byte ZIP SHA-256 ca48ccf69f1ca97a6447befa7ed9525cef56fa61d8e60897ecabcbf257d4861b, containing only the 14,228-byte JSON SHA-256 ccd04869f0ade3aa7184ade3f6000ef8bc5d5ddc42cf376e1370ab4249a983f2; its canonical embedded backlog digest recomputes to 4ec4f84566f5f3f8dccd1945906d1a782f4532547a13cadc8570cb5b682bdcb2. Pages run 35558923123 / job 106207907003 checked out the same SHA and passed 381 tests plus canonical/deterministic static-site builds; deploy remained skipped.

This is a COMMENT review rather than self-approval. With the repaired exact head and checks stable, I find the PR safe to mark Ready and squash-merge. Benchmark movement remains none; M2 remains 0/5.