trevon/brick
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
brick/docs/ROADMAP.md
trvon 7c664cb5a0
Some checks failed
build / build-linux (pull_request) Successful in 1m3s
Details
build / build-macos (pull_request) Has been cancelled
Details
build / release (pull_request) Has been cancelled
Details
docs: describe C/C++ and Rust resolution limits and next stages 
README limits now describe include-closure C++ lookup, declared-receiver
member resolution, Rust macro re-parsing, unresolved reasons, and
metrics.resolution, plus the remaining adapter gaps. The roadmap records
candidate edges, the Fold compiler call lane, and compile_commands/Cargo
module resolution as next stages; literature notes cite the call-graph and
name-resolution papers found through paperbridge.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-09-22 15:12:04 -06:00

3.1 KiB
Raw Permalink Blame History

Roadmap

Seed: v0.1

  • Versioned graph JSON.
  • Fast lexical multi-language preview.
  • Pinned default C++/JS/TS/Dart/Rust grammar extraction, source-relative imports, scoped heuristic resolution, and an exact labeled regression benchmark.
  • Dart classes, functions/methods, imports, inheritance, and local calls.
  • Focused P4_16 lexical structures, scoped transitions/apply/action calls, and positive plus clean-negative fixtures.
  • Qualified p4c v1.2.5.16 diagnostic-envelope import with Imported evidence and fail-closed symbolic-adapter status.
  • Explicit P4 parser-state/control-apply cycle, reachability, and unresolved dependency metrics with clean negatives.
  • Default-disabled Tree-sitter P4 pilot with pinned grammar/query identity, parallel Syntactic/Heuristic evidence, a versioned availability/setup probe, and a written non-promotion decision.
  • Deterministic schema-v2 structural diff over compatible schema-v4 buildings, with inspectable before/after facts and optional brick check count gates.
  • Schema-v1 repository-relative forbidden-dependency policy with exact scope, fail-closed wire values, and complete matched-edge evidence in brick check.
  • Read-only MCP 2025-06-18 stdio queries for dependency impact, SCC cycles, weak-graph bridges, and complete evidence over validated schema-v4 JSON.
  • Fold SARIF/native finding overlay.
  • Interactive Cytoscape building.
  • Transparent structural indicators and deterministic tests.

Next: compiler-grade relationships

  • Adapter contract for YAMS graph JSON and Fold ProjectCallGraph/fact IDs.
  • Qualification gate for real machine-readable P4 verifier contracts and separate p4c-graphs structural evidence.
  • Expand and qualify Tree-sitter/compiler frontends per language while keeping them separate evidence backends; do not promote the P4 pilot until macro and multi-project labeled-corpus gates pass.
  • Overload/alias/dynamic-dispatch ambiguity represented as candidate edges (schema 5); today ambiguity is an External target with reason ambiguous.
  • Fold ProjectCallGraph and fold-rust MIR call export, imported as a separate Imported call lane with per-file Heuristic/Imported agreement.
  • compile_commands.json include paths and Cargo workspace/mod-tree module resolution for the heuristic C++/Rust resolvers.
  • Namespace/package clustering and richer versioned policy rule kinds beyond the seed forbidden-dependency boundary.

Later: behavior and symbolic structure

  • Fold summary/counterexample overlays at function/block/operation levels.
  • Runtime trace heat maps kept separate from static reachability.
  • Rich diff visualization and rename-candidate analysis layered over the stable JSON diff without changing its identity semantics.
  • Richer MCP filters, pagination, unstable-wall interpretation, and remote transports layered over the seed read-only stdio query surface.

Quality campaign

  • Labeled multi-language relationship corpus with positive and clean negatives.
  • Precision/recall per edge kind and backend.
  • Layout usability tasks and graph-size/latency/RSS benchmarks.
  • Default changes only after repeated baseline/candidate artifacts and written promotion decisions.