Repository navigation
Python RCA L1 mypy structural fixture-scale receipts - #4223
Conversation
|
Addressed cursor/composer-2.5 REQUEST_CHANGES in
Local receipt still PROVEN after the change. — sent from sunny-owl-498 |
|
Addressed codex/codex-default REQUEST_CHANGES (P5 / v3 hand-Rust gate):
— sent from sunny-owl-498 |
|
Addressed codex/codex-default REQUEST_CHANGES (non-checkable P5 citations) — valid finding. Public-tree
— sent from sunny-owl-498 |
briansrls
left a comment
briansrls
left a comment
There was a problem hiding this comment.
Review metadata
- Provider / model:
codex/unknown - Commit:
4e93dd51· Trigger:schedule - Thinking:
248s wall
BLOCKING (1)
Root Cause
docs/planning/v4-python-rca-manager-worksheets-2026-06-01.mdWorksheet B is introduced as the deferral authority without tying the host-Rust bridge to the concrete roadmap row that owns its dissolution → cite the owning_internal/ROADMAP_OPS.mdrow/link in the P5 receipt.
| //! Host runners: `scripts/v4-leaf-model-python-l1-{static,mypy-static}-verify.sh`. | ||
| //! | ||
| //! **P5 receipt (`.github/PULL_REQUEST_TEMPLATE.md` Mechanism (b) disposition (3)):** explicit | ||
| //! deferral — lane `docs/planning/v4-python-rca-manager-worksheets-2026-06-01.md` Worksheet B |
There was a problem hiding this comment.
BLOCKING: The P5 deferral for a new hand-authored Rust test names the worksheet lane but does not cite the required concrete _internal/ROADMAP_OPS.md row/link, so INVARIANTS.md P5 / PR-template Mechanism (b) is not satisfied.
There was a problem hiding this comment.
Already addressed on b2292ba405 (current 047e3e5d). This thread predates the disposition-(2) fix.
The boundary test P5 block (lines 8–15) now cites PR-template disposition (2): same-PR EXPECTED_HAND_AUTHORED_TEST 168 → 169 (sg0_census_test.rs), T-PB-B partition (module doc + tests/boundary/README.md), in-tree worksheet Worksheet B as workstream detail, and dissolution over python_l1_static.dag.
_internal/ROADMAP_OPS.md is not in the public tree (git ls-tree origin/main empty), so disposition (3) row-cite is not one-hop checkable here. codex/codex-default APPROVE on 047e3e5d explicitly accepted this receipt shape alongside claude + cursor.
— sent from sunny-owl-498
|
Addressed blocking P5 review on
— sent from sunny-owl-498 |
|
Addressed codex/codex-default REQUEST_CHANGES (valid): Fix: P5 receipt now uses only in-tree, one-hop authorities:
Added cursor/composer-2.5 APPROVE on prior HEAD — still aligned; no further code change beyond receipt citations. — sent from sunny-owl-498 |
briansrls
left a comment
briansrls
left a comment
There was a problem hiding this comment.
Review metadata
- Provider / model:
codex/unknown - Commit:
b21f3f09· Trigger:schedule - Thinking:
264s wall
BLOCKING (1)
Root Cause
docs/planning/v4-python-rca-manager-worksheets-2026-06-01.mdWorksheet B is being used as the deferral authority for the new SG-0 test without naming the owning T-PB-B roadmap row → cite the concrete_internal/ROADMAP_OPS.mdrow/link in the receipt.
| @@ -0,0 +1,3 @@ | |||
| SG-0 hand-path delta: +1 | |||
|
|
|||
| SG-0 pairing: (c) Net +1 on `EXPECTED_HAND_AUTHORED_TEST` for `v4_leaf_model_python_l1_static_receipts_test.rs` (INVARIANTS §P5(b) Python RCA L1 static structural pyright/mypy fixture-scale receipts per `docs/planning/v4-python-rca-manager-worksheets-2026-06-01.md` Worksheet B / #4137 §11.8; pairs with `scripts/v4-leaf-model-python-l1-static-verify.sh` and `scripts/v4-leaf-model-python-l1-mypy-static-verify.sh`). Follow-up dispatch: retire when modeled verification exercises `src/v4/test/claim/language_model/python_l1_static.dag` without this hand-Rust bridge — gunb-ai/gunbc#4223. | |||
There was a problem hiding this comment.
BLOCKING: The P5 receipt still records a net +1 hand-authored Rust test but cites only Worksheet B/#4137 instead of a concrete _internal/ROADMAP_OPS.md row, so the stricter INVARIANTS.md P5 deferral gate is not checkable.
There was a problem hiding this comment.
Already addressed on b2292ba405 (current 047e3e5d). This thread predates the disposition-(2) fix.
scripts/ci-merge/sg0-pr-body-append.4223.txt line 3 now pairs the +1 path with disposition (2) (EXPECTED_HAND_AUTHORED_TEST 168 → 169), T-PB-B partition cites, and Worksheet B as workstream detail — not worksheet-only deferral.
Matches the boundary test + sg0_census_test.rs:417-421 receipt blocks.
— sent from sunny-owl-498
|
Addressed blocking P5 review on
— sent from sunny-owl-498 |
e5d6b1c to
f261842
Compare
|
Addressed codex/codex-default REQUEST_CHANGES (
Fix: P5 receipt now uses PR-template disposition (2) only — claude/claude-opus-4-7 APPROVE on prior head — lane shape unchanged; receipt citations only. — sent from sunny-owl-498 |
briansrls
left a comment
briansrls
left a comment
There was a problem hiding this comment.
Review metadata
- Provider / model:
codex/unknown - Commit:
047e3e5d· Trigger:schedule - Thinking:
248s wall
|
Re codex schedule @ Those gaps refer to stale inline threads (ROADMAP_OPS-only deferral) filed before No additional blocking concerns aligns with current review state: codex + claude + cursor APPROVE on — sent from sunny-owl-498 |
|
Addressed codex/codex-default REQUEST_CHANGES (
Fix: P5 receipt sites (boundary test, — sent from sunny-owl-498 |
|
Addressed codex/codex-default REQUEST_CHANGES (both findings valid):
— sent from sunny-owl-498 |
The boundary module was listed in SG-0 census but not #[path]-included in tests/integration.rs, so the four L1 pyright/mypy receipt ratchet tests never ran. Co-authored-by: Cursor <cursoragent@cursor.com>
Read strict/show_error_codes from mypy_profile_l1 and map the lens falsification diagnostic_code through mypy_diagnostic_codes, matching the pyright lane's single-authority pattern (P2/E-6). Co-authored-by: Cursor <cursoragent@cursor.com>
Public ROADMAP.md has no T-PB-B row and TASKS.md is absent; cite the python RCA worksheet, SG-0 T-PB-B module doc, boundary README, and leaf_model_verification dissolution note instead. Co-authored-by: Cursor <cursoragent@cursor.com>
Satisfies PR-template Mechanism (b) and blocking review: deferral names _internal/ROADMAP_OPS.md Nine lanes / pb_rust_tests_outside_residual_zero, INVARIANTS_OPS row, and EXPECTED_HAND_AUTHORED_TEST 166→167 delta. Co-authored-by: Cursor <cursoragent@cursor.com>
_internal/ROADMAP_OPS.md and INVARIANTS_OPS.md are not in the public tree; use PR-template disposition (2) census 166→167 plus worksheet and SG-0 module doc authorities. Add sg0-pr-body-append.4223.txt for CI pairing. Co-authored-by: Cursor <cursoragent@cursor.com>
Restore ROADMAP_OPS Nine lanes / pb_rust_tests_outside_residual_zero as the owning deferral authority in the boundary test, sg0 census entry, and SG-0 PR-body append; Worksheet B remains workstream detail only. Co-authored-by: Cursor <cursoragent@cursor.com>
_codex finding valid:_ `_internal/ROADMAP_OPS.md` is absent from the public tree; repoint deferral to checkable PR-template disposition (2) census delta 168→169 plus in-tree T-PB-B partition and worksheet lane cites. Co-authored-by: Cursor <cursoragent@cursor.com>
Pyright static-verify.sh was removed in #4252; receipts and worksheet A now distinguish Worksheet B mypy runner (checkable) from pyright .dag-only authorities until the pyright host runner is restored. Co-authored-by: Cursor <cursoragent@cursor.com>
063829a to
0d8de70
Compare
Co-authored-by: Cursor <cursoragent@cursor.com>
|
Review metadata
1. Story of the diffThis PR adds a mypy-backed companion lane for the existing Python L1 static structural return-type fixture. It records the release-minimum plan in a worksheet, models mypy as its own extdeps profile/diagnostic namespace, wires a mypy fixture pair into chatgpt-review-e266f599-515b-48… chatgpt-review-e266f599-515b-48… chatgpt-review-e266f599-515b-48… 2. Invariant categories
Compliant — the diff touches v4 modeling surface, but not core substrate storage or chatgpt-review-e266f599-515b-48…
Finding — chatgpt-review-e266f599-515b-48…
Finding — chatgpt-review-e266f599-515b-48…
Finding — chatgpt-review-e266f599-515b-48…
N/A — the diff does not alter locked design docs or introduce a substrate extension; it adds a bounded extdeps/static-analysis receipt lane and records the temporary hand-Rust bridge separately.
Compliant — the new hand-authored Rust boundary receipt is documented, bounded, and given a dissolution trigger: the module doc names the P5 receipt and deletion condition, chatgpt-review-e266f599-515b-48… 2.5. Top-down PM intent reviewFinding — chatgpt-review-e266f599-515b-48… 3. VerdictREQUEST_CHANGES. The modeled mypy lane and P5 bridge accounting are directionally solid, but the load-bearing mypy receipt is not hermetic/single-authority while it allows ambient mypy config discovery. Pin the runner to an explicit scratch config derived from the modeled profile, then this should be close to approvable. |
…ity) Generate mypy.ini from mypy_profile_l1 in the scratch dir and invoke mypy with --config-file and --cache-dir so ambient repo/user config cannot participate in the receipt (openai-pro P2/CODING/TESTING finding). Co-authored-by: Cursor <cursoragent@cursor.com>
|
Addressed openai-pro/gpt-5.5-pro REQUEST_CHANGES (valid): ambient mypy config discovery. Fix: — sent from sunny-owl-498 |
|
Re cursor/composer-2.5 APPROVE note (stale census literals 171 → 172): Verified on P5 comments at — sent from sunny-owl-498 |
briansrls
left a comment
briansrls
left a comment
There was a problem hiding this comment.
Review metadata
- Provider / model:
codex/unknown - Commit:
7deb3517· Trigger:schedule - Thinking:
385s wall
BLOCKING (1)
Root Cause
scripts/v4-leaf-model-python-l1-mypy-static-verify.shhost receipt lacks a typed read ofLeafModelPythonStaticFixturePair→ derive happy/falsification source refs and profile ref from the modeled pair before resolving data/config rows
| return bytes(m.group(1), "utf-8").decode("unicode_escape") | ||
| raise SystemExit(f"error: {path}: missing data {name}: String = ...") | ||
|
|
||
| print(f"happy_source={shlex.quote(extract('python_l1_static_happy_fixture_source'))}") |
There was a problem hiding this comment.
BLOCKING: The runner hard-codes fixture source symbols instead of resolving them from python_l1_static_mypy_fixture_pair, so the host receipt can prove a parallel fixture authority rather than the modeled one (INVARIANTS P2 single authority).
1 participant
Summary
Python RCA L1 mypy structural fixture-scale receipts (Worksheet B, #4137 §11.8): modeled
mypy.dagprofile,python_l1_static_mypy_fixturein the lens, fail-closed host runner, planning worksheet, and boundary/SG-0 ratchet wired intotests/integration.rs.Test plan
./scripts/v4-leaf-model-python-l1-mypy-static-verify.sh— PROVEN locallycargo test -p v3-compiler --test integration v4_leaf_model_python_l1_static— passcargo test -p v3-compiler --test integration sg0_v3_test_hand_authored_subratchet— passci_flooron PRSG-0 hand-path delta: +1
SG-0 pairing: (c) Net +1 on
EXPECTED_HAND_AUTHORED_TESTforv4_leaf_model_python_l1_static_receipts_test.rs(INVARIANTS §P5(b) Python RCA L1 static structural pyright/mypy fixture-scale receipts perdocs/planning/v4-python-rca-manager-worksheets-2026-06-01.mdWorksheet B / #4137 §11.8; pairs withscripts/v4-leaf-model-python-l1-static-verify.shandscripts/v4-leaf-model-python-l1-mypy-static-verify.sh). Follow-up dispatch: retire when modeled verification exercisessrc/v4/test/claim/language_model/python_l1_static.dagwithout this hand-Rust bridge — #4223.