Repository navigation
refactor(parser): separate binding patterns from assignment targets 🧩 - #225
Merged
Merged
Conversation
timfennis
commented
Sep 9, 2026
timfennis
commented
Sep 9, 2026
timfennis
added a commit
that referenced
this pull request
Oct 13, 2026
The name resolver being built on `feat/new-analyser` will resolve type names through lexical scopes. Struct names are scoped like values, but they live in a separate type namespace. The resolver records its results in side tables keyed by `NodeId`, but type annotations had no node ids. Each name needs its own key: in `Map<P, Q>`, `P` and `Q` resolve separately. Parameter annotations can't borrow an owner's id either, because `FunctionParameter` has no node of its own. This follows the same approach as #225, which gave binding patterns and assignment targets their own ids. - `TypeExpr::Name` gets an `id: NodeId`. Both places that build one assign a fresh id: ordinary annotations, and the cast fast path for `x as Int < 5`. - The current analyser ignores the new field. - New parser tests check that every name in a nested annotation, and in the cast fast path, gets its own id. `TypeExpr::Tuple` gets no id, because tuple types have no name to resolve. Nothing in the workspace compares annotations for equality, so including the id in the derived `PartialEq` changes no behaviour. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Lvaluerepresented both declarations and writes, forcing the analyser and compiler to handle impossible index/member declarations. Split it intoBindingPatternfor declarations, parameters, and loop binders, andAssignmentTargetfor writes. Update parser, analyser, compiler, and LSP consumers around those roles.Both representations have location wrappers with a
NodeIdand complete source span, including nested destructuring elements. Consuming expression conversion preserves node identities; receivers and indices remain expressions. Retain identifier token spans separately so parenthesized targets still give precise definition locations. Existing resolution and inferred-type fields remain in the AST for the current analyser.Preserve the current accepted syntax, overload resolution, assignment evaluation order, and closure behavior. Function parameters still require identifiers, and member targets remain forbidden inside destructuring.
Validation:
cargo test --workspace --locked: 659 tests passed, including six new parser tests and two LSP definition tests.cargo build --no-default-features --locked: passed.cargo fmt --all --checkandgit diff --check: passed.AI disclosure: Codex implemented this refactor, added the tests, and wrote this PR description from a design discussed with the maintainer.