improved unbound identifier detection #95
Conversation
| "false" => NixValue::Bool(false), | ||
| "null" => NixValue::Null, | ||
| // list obtained with `nix repl` tab completion | ||
| "abort" |
There was a problem hiding this comment.
Where do you recommend that we actually define these values?
There was a problem hiding this comment.
I don't know. Maybe we can ignore everything that starts with __, accept the builtins and hardcode the 3 or 4 remaining cases like true and false ?
src/eval.rs
Outdated
| /// whether the patter is incomplete (contains `...`) | ||
| ellipsis: bool, | ||
| /// the identifier bound by `@` | ||
| at: Option<String>, |
There was a problem hiding this comment.
| at: Option<String>, | |
| at: Option<ExprResultGc>, |
This would let us put the identifier in a Scope, and the Expression struct would also give us more info about where the identifier is located.
There was a problem hiding this comment.
I made it an ExprResultBox as there is no need for sharing as far as I can see
|
@symphorien Would you mind if I merge in the ExprSources alongside an rnix-parser update in a separate PR? I'd put all your edits in a commit, with you marked as the author and me as the committer, so you'd get credit in the Github UI. Updating rnix-parser would be easier with some of the definitions you added, like StringPartSource. |
|
Fine by me as long as it does not create extra merge conflicts, I have a branch with unused binding detection at https://github.com/symphorien/rnix-lsp/tree/unused and it's based on this branch. (will make a PR after some more testing) |
|
ping @aaronjanse |
followup to #88
parse all of rnix typed nodes so that unbound identifiers can be detected in much more cases
the evaluation machinery only returns unimplemented for the new nodes
at some point I ran rustfmt on the project by habit, I tried to undo some of the changes, but the diff is not minimal.