feat: scope locals to their block in the type environment

Open
#258 0 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
4/5
Estimated time
3-5 days
Newbie friendliness
52/100
Issue type
Feature
Clarity
Clearly specified
Activity status
Active
Domain
tooling

Research direction

Start with LookupAt in internal/indexer/goextract/typeenv.go:89 and the test TestShadowedAliasResolvesByPosition in goextract/typeenv_test.go:92. Pick one of the name-only extractors, such as internal/indexer/mainstream/rustextract/typeenv.go:142, and add a shadow test where an inner-block local is followed by a call to the outer name. Done when that test passes and the outer name resolves after the block closes.

Written by the indexing model from the issue text.

Description

area:indexer enhancement priority:low
What exists today?

Each extractor's type environment finds a receiver's type by name. Four extractors also check the position of the call.

  • Go: internal/indexer/goextract/gotypeenv.go:146 calls LookupAt.
  • TypeScript: internal/indexer/tsextract/typeenv.go:139 calls LookupAt.
  • Java: internal/indexer/javaextract/typeenv.go:132 calls LookupAt.
  • C#: internal/indexer/csharpextract/typeenv.go:161 calls LookupAt.

LookupAt is defined at internal/indexer/goextract/typeenv.go:89. It returns the latest declaration whose scope holds the position.

Seven other lookups use the name-only Lookup, which ignores the position:

  • Python: internal/indexer/pyextract/typeenv.go:164
  • Ruby: internal/indexer/mainstream/rubyextract/typeenv.go:96
  • PHP: internal/indexer/mainstream/phpextract/typeenv.go:163
  • Kotlin: internal/indexer/mainstream/kotlinextract/typeenv.go:164
  • Swift: internal/indexer/mainstream/swiftextract/typeenv.go:140
  • Rust: internal/indexer/mainstream/rustextract/typeenv.go:142
  • C and C++: internal/indexer/mainstream/cextract/typeenv.go:175

A whole-body lookup is wrong only where the language ends a local at a block. Kotlin, Swift, Rust, C and C++ do. Ruby ends a local at a do or brace block, but not at if. Python and PHP have no block scope, so a local hides an outer name for the whole function. internal/indexer/goextract/scope.go:68 records the PHP rule.

No fixture covers a block-scoped shadow in these seven extractors yet.

How does it behave now?

Reading internal/indexer/mainstream/rustextract/typeenv.go:142, the lookup ignores the block. A local declared in an inner block still answers a call after that block closes. This is from reading the code. A run of the Rust extractor on a shadow fixture has not been done.

How should it behave?

Each extractor scopes a local by its language's rule:

  • Kotlin, Swift, Rust, C and C++: from the declaration to the end of the enclosing block.
  • Ruby: from the declaration to the end of the enclosing do or brace block.
  • Python and PHP: whole-function scope, which the language defines.

A call resolves to the declaration whose scope holds its position, as LookupAt does for Go. A call after an inner block resolves to the outer name again.

Add one shadowing test per extractor, modelled on TestShadowedAliasResolvesByPosition at internal/indexer/goextract/typeenv_test.go:92. Each test declares an inner-block local, then calls the outer name after the block. The test asserts the outer resolution. The Python and PHP tests assert whole-function scope, so the rule stays fixed.

The spec in openspec/specs/graph-index/spec.md already requires this. Its requirement is "A same-file reference resolves by lexical scope." The design of the per-body type environment is in openspec/changes/archive/2026-11-02-index-invariants/design.md, decision D17.

Would this change existing behavior?

Yes. Some calls that resolve today to an inner-block local will resolve to the outer name, or stay unresolved. The output shape does not change.

What kind of improvement?

Correctness in an edge case

Before submitting
  • I searched existing issues and this is not a duplicate
  • I checked that the current behavior is not a deliberate decision in openspec/specs/
Dominant language
Go
Stars
0
Forks
0
Avg merge
5h 16m
Merged PRs (30d)
107

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from seanb4t/codegraph-go

All issues in seanb4t/codegraph-go

Similar issues

More Go issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.