Skip to content

Reconcile upstream aliases and declarations through #2192 - #348

Merged
bompus merged 6 commits into
fork/consolidatedfrom
reconcile/upstream-2192
Oct 1, 2026
Merged

bompus merged 6 commits into
fork/consolidatedfrom
reconcile/upstream-2192

Conversation

@bompus

@bompus bompus commented Oct 1, 2026

Copy link
Copy Markdown
Owner

Merge upstream through 7c1c9a36 (colbymchenry#2192), preserving upstream ancestry. This brings in colbymchenry#2186, colbymchenry#2187, colbymchenry#2188, colbymchenry#2189 and colbymchenry#2192: C++ declarations around macros and conditionals, React Router config helpers, Lua local aliases, CommonJS require calls and namespace re-exports.

Port import/export and alias resolution to the fork's native kernel. Returned Lua module tables define public members; lexical shadows, overwritten tables and forwarding cycles remain unresolved. CommonJS uses explicit forwarded exports, and namespace exports retain their member boundary. Router helpers require a literal object and a direct literal/template result. C++ alias and complex-receiver lookup uses declaration scope plus the caller's include closure; ambiguous owners and unsubstituted template parameters remain unresolved.

Validation: native build and clippy, 52 focused regression controls, 6,868 passing suite tests (34 existing skips), eight unchanged golden dumps, five precision gates, and the TypeScript/UI build all pass. Existing test assertions are unchanged. Twelve pinned real-repository comparisons cover Slim, Vite, Javalin, Petclinic, Exposed, fmt, ripgrep, Express, Kong, bulletproof-react, Zod and nlohmann/json. The six unchanged controls retain identical call edges. Bulletproof-react gains eleven source-checked navigation edges; Zod gains 184 call edges. Source audits verify restored serializer, JSON pointer and Lua forward-function calls, and correct the newly exposed receiver targets. Edge-diff totals include qualified-name/provenance changes and are not counts of unique semantic improvements.

Kong indexing was 9.8 seconds in the final observed run versus 5.5 seconds at baseline, after removing repeated declaration scans from an earlier 28.9-second snapshot. These are single-run observations, not a controlled performance benchmark. Re-index affected projects after upgrading.

README checked: upstream merge point, Parsing, resolution and languages (Parser, Files with syntax errors, Name resolution), native extraction/resolution comparisons, routing additions and measured-results rows. Updated the merge point and current capability descriptions, plus matching site pages and framework coverage. Historical measurements remain pinned to their stated revisions; no fresh performance claim was substituted.

Loading
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants