Repository navigation
Code Analysis Review Map
Current cumulative Labs review (9 November 2026). Draft #1301 compares the v14 product tree directly against Labs at bf0a29e76. It preserves the earlier contributors' commits. #1283 remains an overlapping earlier proposal with downstream branches. The integration replay and evidence index covers six source/package smoke cases and links the original research records. Review has been requested. The table below records the earlier review map.
Snapshot: 6 November 2026. These links organize review questions and evidence; they do not establish approval or readiness. All listed author PRs are open Drafts. Check each current head and discussion before review. Labs base at this snapshot is bf0a29e7.
| Concern | PR and dependency | Evidence | Review decision |
|---|---|---|---|
| Stable finding IDs | #1212, Labs | PR reproduction | Is precedence for a 12-digit stable hash over an ambiguous old counter acceptable? |
| Conditional cycles | #1213, Labs | Validation · optional chains | Are guarded loads distinguished from expressions eagerly evaluated to select them? |
| Relative CommonJS runtime files | #1214, based on #1213 | Validation | Does captured runtime-file selection preserve unresolved cases and match the scoped Node controls? |
| Python packages and relative namespaces | #1216, based on #1213 | Python imports · relative namespaces | Are source-root order, parent paths and namespace portions modeled conservatively for the captured scope? |
| Deep Python expressions | #1217, Labs | SymPy · PR | Does the bounded change preserve evidence and failure reporting for deep expressions? |
| Java Unicode in comments and literals | #1218, Labs | Java Unicode | Is the safe escape boundary and disclosure for unsupported cases sufficient? |
| JS/TS package resolution | #1219 → #1220, Labs | Node conditions · conditional null | Are loading mode, captured implementation choices, matched-null blocking and array fallbacks justified by the scoped controls and remaining gaps? |
| Lexical calls and parameter scopes | #1225, on #1220 | Call evidence · parameter defaults | Is conservative binding and explicit unresolved coverage worth the raw-output cost, including the default-parameter guard? |
| Single-entry orientation | #1226, on #1225 | Question follow-up | Are freshness, unavailable-versus-zero coverage, and the 40-call presentation limit clear enough for source investigation? |
| Relative default calls | #1227, on #1226 | Default calls · parameter defaults | Are the direct-declaration boundary and v1/v2 compatibility acceptable? |
| Paired source questions | #1229, on #1227 | Change questions · parameter defaults | Are definition pairing, configuration coverage and explicitly unresolved dispatch sufficient for orientation without implying complete impact? |
| Saved Python/Java source questions | #1234, on #1229 | Saved language questions | Are captured source/function/file hints useful with calls explicitly unavailable and overload ambiguity retained? |
| Java static import namespaces | #1231, Labs | Java static imports | Does separating static value imports from type lookup correct the tested missing dependencies without asserting unsupported resolution? |
The native stack #1221 contains #1219 → #1220 → #1225 → #1226 → #1227 → #1229 → #1234. #1214 and #1216 both target #1213; their common base is not evidence that they belong to stack #1221. The other rows target Labs independently. Recheck actual base and stack membership before any integration decision.
Adjacent teammate work remains separately owned: #1222 addresses invalid complexity thresholds, #1223 parse/unanalyzed-language disclosure, and #1224 baseline accept/revoke. At the 7 November refresh, #1222 and #1223 are non-Draft, while #1224 is Draft pending scope agreement; all remain open against Labs; coordinate overlapping validation and generated artifacts with their authors rather than treating them as dependencies or as part of this stack.
Native stack #1237 contains #1235 (combined baseline), #1236 (Python call evidence), and #1243 (Python/Java definition comparisons). All are unmerged Drafts; the original contributions above remain open and attributed. Fixed evaluation, replay and limitations. Review exact source/header ranges, conservative targets, source freshness, explicit partial coverage and the bounded grading denominator.
Java calls #1239 is a separately developed slice based on #1235; its own evidence is not proof of the combined paired-definition composition.
See the collaboration overview for original contributor attribution, A1/A2 packaging updates, #1224 scope deferral, Atlas and source-review work, the #1244 composition and separate green-but-unmerged CI repair #1242.
Draft #1250 now composes #1222/#1223 onto #1244, preserving original commit ancestry and attribution. See the team overview for exact heads, scope and the reproduced baseline test failure. Original PRs remain open; no merge or approval occurred.
Draft #1284, head 59f1a80058c5e5fd8859be960b436a05a799d8e7, starts at #1283/0c81af0ea and targets the #1283 source branch, not Labs. Fixed Echo/client_golang questions, gopls comparison and reproduction. Six and nine interface-mediated references now reach the captured declaration with no wrong targets; the changed concrete implementation still reports only direct callers. Review: is a declaration anchor with explicitly unknown runtime dispatch useful without implying implementation impact?
Draft #1285, head 209f4857c4b746a4f0ac6fd88bdd449ae2d2e080, starts at #1284's source branch codex/code-analysis-go-gopls-eval from 59f1a8005 and targets that branch, not Labs. Fixed Echo/client_golang/afero questions, method-set rules, gopls comparison and reproduction. The change question now separates direct calls from interface candidates: 3 direct + 6 candidates, 47 direct + 9 candidates, 0 direct + 38 candidates, wrong = 0 and missed = 0. Every displayed direct call and all 15 previous-round candidates also match the source repository version, target file, declaration and owning type; a same-named interface-method swap and four field controls are rejected. T/*T, full method sets, promoted wrappers and old captures without evidence are explicit; runtimeTarget is never asserted. Afero was the uninvolved repository (HttpFs excluded as a same-name non-implementation). Remote CI succeeded at the final head. Review: does the per-candidate chain justify each interface position without being read as runtime dispatch, and are the incomplete-interface/generic/alias unknowns declared clearly enough?