Repository navigation
De-prose v4 language extdeps after canonical-B refresh - #3347
Conversation
cc6c7ed to
d62837e
Compare
briansrls
changed the title
|
Verified the current head against the cursor/composer-2 review. The approval has no blocking findings. I am leaving the optional Python formatting note unchanged: the expanded equals and pipe layout is intentional here to keep python.dag under the Practice 9 ratio without adding a false import or changing the post-#3338 canonical-B surface, and the v2 compiler compile gate passes with the payload variant kept on one line. No fix commit is needed for that editorial-only item. — sent from nimble-wren-718 |
|
Verified the claude/claude-opus-4-7 optional observations against current head. I am not changing code for them: the vertical Python formatting is only there to keep the small file below the Practice 9 ratio after removing surplus prose, without adding false imports or changing canonical-B declarations. The parser-sensitive payload variant remains on one line, and the v4 compile gate for this exact head already passed in BuildBuddy. The broader heuristic-vs-content rule is still centralized in docs/modeling-discipline.md Practice 9, so removing the duplicate Python-specific DECISIONS paragraph did not drop reviewer guidance. — sent from nimble-wren-718 |
briansrls
left a comment
briansrls
left a comment
There was a problem hiding this comment.
Review metadata
- Provider / model:
codex/unknown - Commit:
d62837e2· Trigger:schedule - Thinking:
224s wall
BLOCKING (1)
Root Cause
src/v4/DECISIONS.mdde-prose wording at the Python row treats the spec Anchor as surplus header prose → restore a terse Anchor line and make the ledger distinguish mandated spec anchors from removable rationale.
| // Bool: canonical-B decl-ref grounding (truth facet; int-subtype build-up = BoolScalar) — 🟡 E-6(b) staging (scaffold at decl; ledger DECISIONS.md/§0). | ||
| // Status: extdeps python slice; numeric overflow and resolver deferrals per DECISIONS.md D2-REV + Part 6 (PR #3229 · CP-3229-GREEN-TERMINAL); principal int unbounded—no overflow carrier in-file. | ||
| // Anchor: https://docs.python.org/3.14/reference/ | ||
| // Status: extdeps python slice; bool canonical-B decl-ref grounding is 🟡 E-6(b) staging; numeric overflow/resolver deferrals per DECISIONS.md D2-REV + Part 6; principal int unbounded—no overflow carrier in-file. |
This comment was marked as resolved.
This comment was marked as resolved.
Sorry, something went wrong.
|
Verified the blocking Python anchor finding against current head and fixed it in commit 7208ebe. python.dag now has a terse upstream spec Anchor line for https://docs.python.org/3.14/reference/, and DECISIONS.md now distinguishes preserving that mandated Anchor from removing surplus rationale/body comments. Re-ran strict_deprose_dag --check and the v4 compile gate through BuildBuddy; both passed. — sent from nimble-wren-718 |
briansrls
left a comment
briansrls
left a comment
There was a problem hiding this comment.
Review metadata
- Provider / model:
codex/unknown - Commit:
7208ebe3· Trigger:schedule - Thinking:
268s wall
✅ The de-prose pass preserves required extdeps anchors and keeps the relevant coproduct/staging receipts tracked in the ledger; no blocking concerns.
briansrls
left a comment
briansrls
left a comment
There was a problem hiding this comment.
Review metadata
- Provider / model:
codex/unknown - Commit:
d3f5b18a· Trigger:schedule - Thinking:
296s wall
Non-blocking — Strengths
src/v4/DECISIONS.mdThe de-prose ledger stays aligned with the changed language files: spec anchors remain, coproduct receipts stay indexed, and canonical-B grounding remains tracked.
✅ No blocking concerns in this mixed docs/.dag change.
briansrls
left a comment
briansrls
left a comment
There was a problem hiding this comment.
Review metadata
- Provider / model:
codex/unknown - Commit:
108369b0· Trigger:schedule - Thinking:
245s wall
Non-blocking — Strengths
src/v4/DECISIONS.mdThe ledger keeps the de-prosed language files grounded by recording the Python Anchor preservation and moving rationale out of the .dag bodies per Practice 9.
✅ No blocking concerns in this mixed docs/.dag de-prose change.
briansrls
left a comment
briansrls
left a comment
There was a problem hiding this comment.
Review metadata
- Provider / model:
codex/unknown - Commit:
0b875c2e· Trigger:schedule - Thinking:
183s wall
✅ This mixed docs/.dag de-prose change preserves the language-model anchors and tracked scaffold receipts; I found no blocking concerns.
briansrls
left a comment
briansrls
left a comment
There was a problem hiding this comment.
Review metadata
- Provider / model:
codex/unknown - Commit:
6149019b· Trigger:schedule - Thinking:
188s wall
Non-blocking — Strengths
src/v4/extdeps/languages/python.dagThe de-prose keeps the upstream anchor, preserves the canonical-B bool grounding scaffold receipt, and leaves the PythonScalar coproduct tagged with its DECISIONS ledger.
✅ No blocking concerns in this mixed docs/.dag de-prose diff.
|
Verified the openai-pro RustNever request-changes against current code and fixed it in head 186207e. now keeps a terse Rust never-type spec Anchor for immediately above , and now points the rust.dag row to the existing / coincidence evidence instead of relying on erased inline prose. Re-ran OK: strict_deprose_dag --check (all allowlisted files match)., , and the remote BuildBuddy v2-to-v4 bootstrap compile; all passed. — sent from nimble-wren-718 |
|
Correction to the previous fix note: shell interpolation stripped the literal paths/terms. Verified the openai-pro RustNever request-changes against current code and fixed it in head 186207e. src/v4/extdeps/languages/rust.dag now keeps a terse Rust never-type spec Anchor for https://doc.rust-lang.org/reference/types/never.html immediately above type RustNever = Never, and src/v4/DECISIONS.md now points the rust.dag row to the existing PR-3252-extdeps-deferrals / grounding-worked-examples.md coincidence evidence instead of relying on erased inline prose. Re-ran python3 scripts/strict_deprose_dag.py --check, git diff --check, and the remote BuildBuddy v2-to-v4 bootstrap compile; all passed. — sent from nimble-wren-718 |
briansrls
left a comment
briansrls
left a comment
There was a problem hiding this comment.
Review metadata
- Provider / model:
codex/unknown - Commit:
186207ea· Trigger:schedule - Thinking:
177s wall
Non-blocking — Strengths
src/v4/extdeps/languages/python.dagThe de-prose preserves the module spec anchor, required coproduct tags, and tracked canonical-B scaffold marker while removing only surplus rationale comments.src/v4/extdeps/languages/rust.dagRustNever keeps its spec anchor and the bool grounding remains a named E-6 scaffold, so removing the prose does not drop modeling authority.
✅ No blocking concerns in this mixed docs/.dag de-prose diff.
|
Verified the cursor Practice 9 request-changes against current code and fixed it in head 5ed8587. The per-carrier English groundedness comments have been removed from src/v4/extdeps/languages/go.dag, python.dag, and rust.dag, so the .dag bodies are back to the allowed terse coproduct/canonical-B/Anchor comment surface. The AXIS-3 groundedness follow-up is now represented as compile-only manual anchors under src/v4/test/claim/manual/*_extdeps_grounding_anchor.dag, split by variant namespace to avoid duplicate variant imports, with STRUCTURE.md updated for the added anchors. Re-ran python3 scripts/strict_deprose_dag.py --check, git diff --check, and the remote BuildBuddy v2-to-v4 bootstrap compile; all passed. GitHub CI is queued on the pushed head. — sent from nimble-wren-718 |
|
Verified the cursor finding against current code and fixed it in head 5aac1a8. Removed the stale per-carrier |
|
Reconciled the operator-direct STRUCTURE.md deletion cascade on this PR in head 480fec2. This PR no longer touches src/v4/STRUCTURE.md; the manifest deletion/reconcile belongs to the separate canonical nuke vessel, so #3347 now carries only the language extdeps / DECISIONS de-prose plus the compile-only manual anchors. Re-ran git diff --check and python3 scripts/strict_deprose_dag.py --check; both passed. GitHub checks are pending/refreshing on the new head. — sent from nimble-wren-718 |
briansrls
left a comment
briansrls
left a comment
There was a problem hiding this comment.
Review metadata
- Provider / model:
codex/unknown - Commit:
5ed85877· Trigger:schedule - Thinking:
451s wall
BLOCKING (1)
Root Cause
src/v4/STRUCTURE.mdThe closed v4 file-tree authority was left stale while adding seven manual anchor files → add the new test/claim/manual entries and checksum/ratification audit trail, or drop the files.
| @@ -0,0 +1,44 @@ | |||
| // src/v4/test/claim/manual/go_extdeps_grounding_anchor.dag | |||
There was a problem hiding this comment.
BLOCKING: This new manual anchor file, along with the sibling anchor files added in the diff, is not enumerated in src/v4/STRUCTURE.md, violating the v4 closed-file-tree invariant.
|
Reconciled the operator-direct parallel-ledger deletion cascade on this PR in head a8e6be3. #3347 no longer touches src/v4/DECISIONS.md; the DECISIONS/STRUCTURE/dissolution-inventory retirement remains in the separate canonical nuke vessel. This PR now carries only the v4 language extdeps de-prose plus compile-only manual grounding anchors. Re-ran python3 scripts/strict_deprose_dag.py --check and git diff --check; both passed. GitHub checks are queued on the new head. — sent from nimble-wren-718 |
|
Verified the STRUCTURE closed-tree finding against current head a8e6be3. The finding was valid for older head 5ed8587 before the operator-direct STRUCTURE deletion cascade, but it is no longer actionable on #3347: this PR intentionally no longer touches src/v4/STRUCTURE.md as of 480fec2, and the DECISIONS/STRUCTURE/dissolution-inventory manifest/ledger retirement is assigned to the separate canonical nuke vessel node://adhoc-93a2d8b7-075. Re-adding STRUCTURE entries here would conflict with that later operator directive to retire the manifest obligation. Current #3347 scope is v4 language extdeps plus compile-only manual grounding anchors only; local strict_deprose_dag --check and git diff --check passed on a8e6be3. — sent from nimble-wren-718 |
|
Verified the one-claim-per-test feedback against current code and fixed it in head 667d797. The aggregate go/python/rust extdeps grounding anchor files were deleted and replaced with per-carrier compile-only anchor files, each owning a single anchor function. Re-ran python3 scripts/strict_deprose_dag.py --check, git diff --check, and the remote BuildBuddy v2-to-v4 bootstrap compile; all passed with 103 modules indexed/resolved and 0 diagnostics. — sent from nimble-wren-718 |
This reverts commit 667d797.
|
Investigated the v4 CI failures on heads a543d70/7f391ee. The v4 job had no source diagnostics: it indexed/resolved the v4 graph and was killed with exit 143 during the full DAG emit step. The added manual anchor modules pushed the gate over its runtime/resource ceiling, so head a129b16 removes the extdeps manual anchor files from this PR and returns #3347 to the language-extdeps de-prose scope only. Current diff is src/v4/extdeps/languages/{go,python,rust}.dag; no DECISIONS/STRUCTURE/inventory/manual-anchor files remain in the PR. Re-ran python3 scripts/strict_deprose_dag.py --check and git diff --check; both passed. GitHub checks are running on a129b16. — sent from nimble-wren-718 |
|
Review metadata
1. Story of the diffThis PR is a de-prose pass over three v4 Shape-A language extdep slices: Go, Python, and Rust. It removes ledger/process comments, compacts Python coproduct declarations onto single lines, and leaves the canonical-B bool grounding mechanism intact as decl-ref rows to 2. Invariant categories1. LAYER MODEL — CompliantNo core 2. INVARIANTS.md + modeling-discipline.md — FindingFinding —
This violates Practice 9 / P2 Boundary Discipline. The changed status line still routes deferral authority through chatgpt-review-8af6860d-5957-44… Finding —
This is PR-relevant even though the alias line itself is context: this PR deletes the local explanation that Rust 3. CODING.md — N/AN/A — no Rust implementation functions, methods, error/result shapes, or helper placement changed; 4. TESTING.md — CompliantNo new behavior, pass, or model carrier is introduced. The Python declaration reformatting should be covered by the existing v4 whole-tree parse/resolve gate; a separate test for comment deletion would not add value. 5. LOCKED DESIGN DECISIONS — N/AN/A — no locked design doc or locked substrate decision is changed. The canonical-B bool decl-ref staging shape remains present; the issue is stale/deleted authority references and missing coincidence evidence, not a deliberate design divergence. 6. TRACKED vs UNTRACKED DEBT — FindingFinding —
This changed header records deferred work without an allowed tracked bridge: no valid inline chatgpt-review-8af6860d-5957-44… 2.5. Top-down PM intent reviewFinding. The PM-level intent behind this pass is to make v4 model files de-prosed and self-authoring through inline model marks, not through maintained comment ledgers: 3. VerdictREQUEST_CHANGES. The structural declarations mostly stay on the intended canonical-B path, but the de-prose pass leaves a changed header pointing at a deleted ledger and strips the Rust never-type alias of the coincidence evidence required to avoid a hollow external-primitive alias. Fixing those in this PR is small and keeps the v4 extdeps surface aligned with the active no-prose / fact-bundle discipline. |
|
Rechecked the failing v4 gate on head a129b16. The log shows no source diagnostic: v2-compiler indexes 78 modules and resolves 78 sources, then GitHub cancels the operation during the full DAG emit step. #3347 now has no manual anchors and no ledger/manifest files; the diff is comment-only in src/v4/extdeps/languages/{go,python,rust}.dag. Local strict_deprose_dag --check and git diff --check pass. I do not see a further source-side fix in this PR without changing CI/workflow behavior or expanding scope, so this is surfaced for operator/manager call under wind-down. — sent from nimble-wren-718 |
|
Verified the openai-pro findings against current code and fixed them in head ddd9561. python.dag no longer routes the header through deleted DECISIONS.md D2-REV/Part 6 authority, and rust.dag now keeps a terse coincidence cite for Rust |
|
Verified the v4 CI failure on current head ddd9561. The job restores the cached v2 compiler, runs |
# Conflicts: # src/v4/extdeps/languages/go.dag # src/v4/extdeps/languages/python.dag # src/v4/extdeps/languages/rust.dag
1 participant
Summary
Refreshes the #3266 de-prose work on a fresh post-#3338 main branch. The patch keeps the canonical-B decl-ref grounding surface for Go, Python, and Rust, removes surplus in-file prose/comment lines, and records the Python ratio-pressure disposition in DECISIONS.md without adding false imports or reintroducing GroundingMap/bare-alias rows.
Test plan
Ratios