New Patterns #155
New Patterns #155
Conversation
There was a problem hiding this comment.
This is great. It makes the distinction between bindings and references much more crisp ✨ .
I think that in some places in the tree, identifier nodes that represent references will not be in _expression nodes. In particular, member_expression and new_expression seem to use primary_expression directly.
I made a branch (based off of this one) where I remove the inline for primary_expression (and drop the leading underscore in its name). You could check it out and merge it into your PR if you think it makes sense. There are no test changes, but primary_expression should now be usable as a supertype in queries (e.g. (primary_expression/identifier).
grammar.js
Outdated
| alias($.array, $.array_pattern) | ||
| // alias($.object, $.object_pattern), | ||
| $.object_pattern, | ||
| // alias($.array, $.array_pattern) |
There was a problem hiding this comment.
Let's 🔥 these comments.
|
oops, forgot to link to the branch: https://github.com/tree-sitter/tree-sitter-javascript/tree/new-patterns-primary-expression |
|
ok, i've merged those branch changes into |
This PR refactors the subgrammar for parameters and assignment LHSes to be a distinct set of patterns, instead of just a set of expressions. This makes it possible to refer to all variables that are used as binders, distinct from variables used as bound values, e.g.
(pattern/identifier)will match the parameterxin(x) => xbut not the use ofxin the body of the arrow function, while dually(_expression/identifier)will match the use but not the parameter.This enables more precise syntax highlighting, but also is required for various precise codenav things to work.