Rust: Unify type inference for tuple indexing expressions #20182
Conversation
hvitved
force-pushed
the
rust/type-inference-tuple-types-follow-up
branch
from
0d7a2f6 to
454ab4d
Compare
There was a problem hiding this comment.
Pull Request Overview
This PR unifies type inference for tuple indexing expressions by treating all field expressions (x.i) uniformly, regardless of whether the type of x is a tuple or a tuple struct/enum. This aligns with the comment in the GitHub issue to standardize the handling of tuple field access.
- Introduces a new
TupleTypeParameterDecldeclaration type within the field expression matching framework - Removes separate handling functions for tuple indexing expressions in favor of the unified field expression approach
- Adds test case to validate tuple field access through references
Reviewed Changes
Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
rust/ql/lib/codeql/rust/internal/TypeInference.qll |
Main implementation: adds TupleTypeParameterDecl to the field expression matching system and removes old tuple-specific inference functions |
rust/ql/lib/codeql/rust/internal/Type.qll |
Minor fix: removes redundant TTuple constraint from UnitType class |
rust/ql/test/library-tests/type-inference/main.rs |
Adds test case for tuple field access through references |
rust/ql/test/library-tests/type-inference/type-inference.expected |
Updates expected test output to reflect the unified type inference approach |
| abstract string toString(); | ||
|
|
||
| abstract Location getLocation(); | ||
| } |
There was a problem hiding this comment.
The abstract Declaration class has a concrete implementation of getTypeParameter that returns none() for all subclasses. This suggests that either this method should be abstract (forcing subclasses to implement it) or it should have a more meaningful default implementation. Consider making this method abstract if different declaration types should handle type parameters differently.
Copilot uses AI. Check for mistakes.
This PR adresses the comment https://github.com/github/codeql/pull/20041/files#r2207752592, i.e. type inference for all field expressions,
x.i, are treated uniformly, regardless of whether the type ofxis a tuple or a tuple struct/enum.