Skip to content

AGENTS.md

Latest commit

 

History

History
67 lines (51 loc) · 3.12 KB

AGENTS.md

File metadata and controls

67 lines (51 loc) · 3.12 KB
Raw

AGENTS.md

Project

  • Rust Language Server Protocol (LSP) server for OpenEdge ABL.
  • Uses tower-lsp for protocol plumbing and tree-sitter-abl for parsing.

Key paths

  • src/main.rs: process entrypoint, parser initialization, and LSP service bootstrap.
  • src/backend.rs: LanguageServer implementation and advertised capabilities.
  • src/handlers/*: LSP request/notification entrypoints. Keep only protocol orchestration, backend calls, and response shaping; avoid heavy parsing/analysis logic here.
  • src/analysis/*: Reusable, mostly pure language-analysis logic over text/tree/schema (collectors, resolvers, symbol/type/definition helpers). Shared across handlers.
  • src/utils/*: Low-level generic utilities (position/offset math, path helpers, text-sync helpers, tree-sitter traversal helpers) that are not domain-feature specific.
  • tree-sitter-abl: dev symlink to tree-sitter-abl repository.
  • tree-sitter-df: dev symlink to tree-sitter-df repository.
  • playground/*: manual testing suite.

Workflow

  • Format code:
    • cargo fmt
  • Lint (if available locally):
    • cargo clippy --all-targets --all-features -- -D warnings
  • Run tests:
    • cargo test
  • Type-check/Linting:
    • cargo check
    • cargo clippy --fix
  • Build server:
    • cargo build
  • Run the server (stdio transport):
    • cargo run

If cargo commands fail with crate download errors, the environment has no network access to crates.io; rerun in a network-enabled environment.

Parser repo integration

  • tree-sitter-abl and tree-sitter-df are git dependencies, so parser changes are not picked up here until dependency update/lockfile refresh.
  • You can access source code of parsers through a symlink that is present in the repository at ./tree-sitter-abl and ./tree-sitter-df to look up syntax details.
  • Use parser repo commands when grammar/syntax behavior is involved:
    • cd /home/jk/tree-sitter-abl && bun run test
    • cd /home/jk/tree-sitter-abl && bun run parse example.p
    • cd /home/jk/tree-sitter-abl && bun run parse:snippet '<snippet>'
    • cd /home/jk/tree-sitter-abl && bun run reference '<query>'
  • If any change is needed in parser / would be help out please tell, we will implement it in parser by other agent.

Conventions

  • Prefer small, focused handler logic and move reusable analysis into src/analysis/* and src/utils/*.
  • Do not silently degrade diagnostics/completion behavior when adding new features; keep existing flows intact unless change is intentional and documented.
  • When the user provides a code sample to reproduce a bug, do not copy that sample 1:1 into tests by default; prefer a minimal synthetic regression case that reproduces the same failure.

Required verification sequence (after every code change)

Run in this exact order:

  1. cargo test
  2. cargo clippy --fix --allow-dirty
  3. cargo fmt
  4. cargo build -r (manual changes)

Notes:

  • Do not skip clippy --fix --allow-dirty unless it fails due to an external/tooling issue.

Notes

  • This repository may not always include extensive tests; when adding behavior-heavy logic, add tests where practical or document manual verification steps.