Struct EntryFnIndex 

pub struct EntryFnIndex<'a> { /* private fields */ }

Bare-name → &FnDef index for the entry module only.

Why a typed wrapper instead of raw HashMap<String, &FnDef>?

Entry-only verify exception (epic #170 Phase 5 deferred follow-up): every helper-law-hint and contextual-helper-hint walker in this module operates exclusively on top-level entry fn defs, because verify <name> parses name as a single source Ident (see src/parser/blocks.rs::parse_verify). The parser does NOT accept dotted verify targets, so dep-module fns are unreachable from the verify-law surface today. Bare-name keying is therefore identity-safe by the parser invariant.

The typed wrapper exists so any future change that tries to thread dep-module fns through this surface trips the compiler: a HashMap<String, &FnDef> from a dep module won’t coerce into EntryFnIndex, forcing the contributor to confront the keying question explicitly. When module-scoped verify ships, the right migration is to drop this wrapper in favour of an FnId-keyed view sourced from CodegenContext.resolved_program.

temporary-migration-bridge: deferred per epic #170 audit (no live trigger today; documented in project_phase_e_scope_b_deferred memory).

Implementations

impl<'a> EntryFnIndex<'a>

pub fn from_entry_items(items: &'a [TopLevel]) -> Self

Build the index by walking entry-scope TopLevel items. Dep modules are intentionally NOT considered — see type doc.

pub fn get(&self, name: &str) -> Option<&&'a FnDef>

Bare-name lookup. Safe by parser invariant — see type doc.

pub fn contains_key(&self, name: &str) -> bool

Existence check. Same identity contract as Self::get.