Rust: Type inference for ? expressions

[hvitved]

DCA looks great; we gain more call edges without affecting performance.

[Copilot]

Pull Request Overview

This PR adds support for inferring types of Rust ? (try) expressions and includes test coverage for Result-based ? patterns.

• Introduce inferTryExprType in the inference engine and include it in cached inference.
• Add a new try_expressions test module exercising ? on various Result scenarios.
• Update expected outputs for the new tests and extend the Postgres path‐resolution consistency expectations.

Reviewed Changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated no comments.

File	Description
rust/ql/lib/codeql/rust/internal/TypeInference.qll	Added inferTryExprType, an import, and integrated it into cached inference
rust/ql/test/library-tests/type-inference/main.rs	Added try_expressions module with ?-operator tests for Result
rust/ql/test/library-tests/type-inference/type-inference.expected	Updated expected outputs to include ?-expression inference entries
rust/ql/test/library-tests/frameworks/postgres/CONSISTENCY/PathResolutionConsistency.expected	Added duplicate method-call entries for consistency verification

Comments suppressed due to low confidence (3)

rust/ql/test/library-tests/type-inference/main.rs:922

• The new inferTryExprType also handles Option-based ? expressions, but the tests only cover Result. Consider adding a test function that uses ? on an Option to validate inference for Option<T>.

mod try_expressions {

rust/ql/lib/codeql/rust/internal/TypeInference.qll:9

• The import codeql.rust.frameworks.stdlib.Stdlib is not used by inferTryExprType or elsewhere in this file. Consider removing it to avoid a redundant dependency.

+private import codeql.rust.frameworks.stdlib.Stdlib

rust/ql/test/library-tests/type-inference/main.rs:945

• In try_chained, the first ? is applied to x: Result<Result<S1, S1>, S1>, but the function signature returns Result<S1, S2>. Without converting S1 to S2 (e.g. via map_err), this will fail to compile. Align the error type or add an explicit conversion.

fn try_chained() -> Result<S1, S2> {

[geoffw0]

Code changes LGTM, and DCA LGTM as well.

[geoffw0]

How much of a concern are the new inconsistencies?

[hvitved]

These consistencies are not really new, they are just surfaced because we can now infer a type for conn at

codeql/rust/ql/test/library-tests/frameworks/postgres/main.rs

Line 9 in bab84d0

let mut conn = postgres::Client::connect("host=localhost user=postgres", postgres::NoTls)?;

.