Recover from more const arguments that are not wrapped in curly braces #107190
Conversation
rustbot
added
S-waiting-on-review
| @@ -5,6 +5,22 @@ LL | fn f1<'a>(arg : Box<dyn X<Y = B = &'a ()>>) {} | |||
| | - ^ expected one of 7 possible tokens | |||
| | | | |||
| | maybe try to close unmatched angle bracket | |||
| | | |||
| help: expressions must be enclosed in braces to be used as const generic arguments | |||
There was a problem hiding this comment.
Hmmm... this seems pretty strange
There was a problem hiding this comment.
Well, I'm pretty sure dyn is parsed as a normal identifier here and not as the edition-2018-specific keyword dyn since the test doesn't specify any edition thereby defaulting to 2015. Hence we suggest {dyn} and dyn> (where dyn is interpreted as the name of a type).
There was a problem hiding this comment.
Not saying that this isn't something I should fix. I will definitely try to improve this output if possible.
There was a problem hiding this comment.
Is the suggestion better on edition 2021?
There was a problem hiding this comment.
Yeah, it is (1 error instead of 2 & no (weird) suggestions):
error: expected one of `!`, `(`, `+`, `,`, `::`, `<`, or `>`, found `=`
--> tests/ui/generic-associated-types/parse/trait-path-expected-token.rs:7:33
|
7 | fn f1<'a>(arg : Box<dyn X<Y = B = &'a ()>>) {}
| - ^ expected one of 7 possible tokens
| |
| maybe try to close unmatched angle bracket
error: aborting due to previous error
There was a problem hiding this comment.
It may be worthwhile teaching this recovery to ignore dyn identifiers (or perhaps all keyword identifiers), even if valid, just for better aesthetics.
There was a problem hiding this comment.
Okay, I've gone for a much simpler solution: I will only emit the suggestion if , or > succeeds the unbraced expression to avoid suggesting garbage if the input is garbage. I've added a small comment for that, too. Now there are no diagnostic regressions at all anymore.
|
Cool, this seems valid. @bors r+ |
bors
added
S-waiting-on-bors
…iaskrgr Rollup of 8 pull requests Successful merges: - rust-lang#107022 (Implement `SpecOptionPartialEq` for `cmp::Ordering`) - rust-lang#107100 (Use proper `InferCtxt` when probing for associated types in astconv) - rust-lang#107103 (Use new solver in `evaluate_obligation` query (when new solver is enabled)) - rust-lang#107190 (Recover from more const arguments that are not wrapped in curly braces) - rust-lang#107306 (Correct suggestions for closure arguments that need a borrow) - rust-lang#107339 (internally change regions to be covariant) - rust-lang#107344 (Minor tweaks in the new solver) - rust-lang#107373 (Don't merge vtables when full debuginfo is enabled.) Failed merges: r? `@ghost` `@rustbot` modify labels: rollup
Recover from some array, borrow, tuple & arithmetic expressions in const argument positions that lack curly braces and provide a suggestion to fix the issue continuing where #92884 left off. Examples of such expressions:
[],[0],[1, 2],[0; 0xff],&9,("", 0)and(1 + 2) * 3(we previously did not recover from them).I am not entirely happy with my current solution because the code that recovers from
[0](coinciding with a malformed slice type) and[0; 0](coinciding with a malformed array type) is quite fragile as the aforementioned snippets are actually successfully parsed as types byparse_tysince it itself already recovers from them (returning[⟨error⟩]and[⟨error⟩; 0]respectively) meaning I have to manually look forTyKind::Errs and construct a separate diagnostic for the suggestion to attach to (thereby emitting two diagnostics in total).Fixes #81698.
@rustbot label A-diagnostics
r? diagnostics