maskhjarna / tree-sitter-purescript Goto Github PK

View Code? Open in Web Editor NEW

6.0 6.0 3.0 34.9 MB

License: MIT License

Python 0.18% JavaScript 20.05% Scheme 1.77% C 76.27% Shell 0.08% Makefile 1.66%

tree-sitter-purescript's People

Contributors

Stargazers

Watchers

Forkers

fredthedino tautologico blinxen

tree-sitter-purescript's Issues

No rows syntax support

Row types don't parse.

A potential implementation should also unify row types and record syntax under the same machinery.

Type functions broken

module Test where

type A :: forall k. k -> k
type A x = x

Derive keyword breaks highlighting

What the title says

A wildcard as the name of a top-level declaration is considered legal

_ = 1

does not error out, whereas it should. Using the wildcard as the whole LHS of = (or <-) is only legal in where, do and let blocks.

Record updates break highlighting

It gets reset in some conditions, in inconsistent ways. Some examples:
1.

2.

3.

4.

Foreign import with type signature ending with a higher kinded type broken

module Test where

foreign import f :: String -> Maybe Int
foreign import f2 :: String -> String

Extended latin characters broken

module Test where

foo :: Int -> Int
foo a = 2 * a

föö :: Int -> Int
föö a = 2 * a

Neither the JS grammar nor the C scanner currently have a notion of documentation comments.
A practical consequence of this is that currently it is not possible to implement markdown injection queries so that doc comments (and maybe, regular comments too?) could be formatted by editors. This isn't hugely important in PureScript because of the guarantees of the type system, but still it would be nice to have markdown support in docs.

`?-` Holes

At the value level they get parsed as implicit arguments (a Haskell legacy).
At the type level they get parsed as ERROR + implicit arguments.

Qualified imports with dots break highlighting

Everything after this statement has broken highlighting:

Comma-separated constraints are broken

Neither of these parse correctly:

class (C1 a, C2 a) ⇐ C3 a
instance (C1 a, C2 a) ⇒ C3 a

It didn't affect #16 because in this PR I couldn't grok what relates to constraints in the codebase, it seemed a bit complicated and related to Haskell more than to PureScript, so there I just re-implemented the thing. A future implementation should start with rewriting the constraints and providing utils to class, instance and derive (newtype) instance declarations.

Record field names surrounded with quotes not valid

Doc-comments are not recognized as comments