Rust: parsing-layer coverage gaps (9 findings)

[magyargergo]

Parent epic: #1919
Enabled by: #1920 (node-type validator) — closes F67, F74 · #1921 (heritage helper) — closes F69

9 verified findings (1 HIGH). Each box is a discrete fix unit. Legend: ✅ confirmed · 🟡 partial.

• ✅ F66 [HIGH] let mut x = ... bindings dropped by scope-resolution type-binding/declaration queries
  • grammar: node-types.json: let_declaration.pattern type is _pattern (required)
  • source: query.ts: every let_declaration rule uses pattern:(identifier) at lines 43-44,63-65,68-71,74-77,80-84,87-89 (verbatim)
• ✅ F67 [MEDIUM] Dead method_call_expression branch — Rust method-call binding inference never fires
• ✅ F68 [MEDIUM] Variable extractor drops let mut x declarations (name resolution fails)
• ✅ F69 [MEDIUM] Heritage queries miss impl ScopedTrait for Type (scoped trait paths)
• ✅ F70 [MEDIUM] Struct-literal construction calls miss scoped and turbofish struct names
• ✅ F71 [LOW] union_item definitions are not extracted anywhere
• ✅ F72 [LOW] Function-like macro invocations are not captured as calls
• ✅ F73 [LOW] Variadic parameters skipped in method/parameter extraction
• ✅ F74 [LOW] Dead slice_type branch in for-loop element-type extraction

Verified against main @ a93ecee0. Finding IDs use an F prefix to avoid colliding with unrelated issue/PR numbers.

[magyargergo]

Partial overlap with PR #1956 (fix #1951)

Finding	#1956 impact
F69 [MEDIUM] impl ScopedTrait for Type	Solved (U1) — bareTypeIdentifier resolves scoped/generic trait/type tails; legacy @heritage impl_item arms widened to one real-parse-verified alternation. Fixture: rust-qualified-trait.
Cross-module trait-impl collision	Solved (review P1) — emitRustTraitImplEdges now scope-aware via finalized indexes + import disambiguation; refuses on ambiguity. Fixture: rust-cross-module-collision.
Heritage worker-path regression	Solved — trait-impl edges via emitHeritageEdges hook; worker-forced regression included.
F66–F68, F70–F74	Not addressed by #1956 (F67/F74 dead-branch class was #1920 / PR #1937)

Why: Rust IMPLEMENTS edges from impl blocks were missing in worker mode (#1951) and scoped trait paths (F69) would have broken parity once scope-resolution owned impl heritage.

[magyargergo]

Coverage vs origin/main @ 7691abf7

3/9 done · 0 in flight · 6 open

Finding	Severity	Status	Evidence
F66	HIGH	Open	rust/query.ts — let_declaration uses pattern:(identifier) only; let mut x dropped
F67	MEDIUM	Done	#1937 — dead method_call_expression branch removed
F68	MEDIUM	Open	Variable extractor still fails let mut x name resolution
F69	MEDIUM	Done	#1940 — scoped trait impl (impl ns::Trait for T)
F70	MEDIUM	Open	Struct-literal scoped/turbofish names not captured
F71	LOW	Open	union_item not extracted
F72	LOW	Open	Macro invocations not captured as calls
F73	LOW	Open	Variadic parameters skipped in extraction
F74	LOW	Done	#1937 — dead slice_type branch removed

[prajapatisparsh]

hey @magyargergo i wish to work on this issue

[magyargergo]

Overlap scan vs #1978 / #1982

Finding	#1978/#1982 impact
F69 [MEDIUM] Heritage: impl ScopedTrait for Type (scoped trait paths)	Partial — heritage / IMPLEMENTS capture was fixed in merged #1940 / #1956. #1978 still covers the struct-node side: impl Trait for a::Inner needs a qualified struct node for method ownership (HAS_METHOD) when the struct path is scoped — deferred from #1977.
(not an F-id) Mod-nested same-tail inherent impl collision (mod a { struct Inner; impl Inner {…} } + sibling in mod b)	Tracked in #1982 §2 — findEnclosingClassInfo inherent-impl branch returns early keyed by raw impl target text; needs mod-scope qualification + registry-primary bridge alignment. Skipped test: rust.test.ts → describe.skip('Rust inline mod-nested same-tail collision … (issue #1978)').
F66–F68, F70–F74	Not addressed by #1978/#1982.

Takeaway: F69’s heritage half is done; remaining Rust same-tail / scoped-path work for graph node identity and inherent-impl method ownership lives under #1978 / #1982, not this ticket’s F-list.