Worked issue #10446 hide "obvious" inlays #10864
Conversation
The suggested method in the issue didn't work because that code path appears not to be hit during the generation of these types. Instead I introduced a new concept of unambigious types as part of let statements and filtered out cases that the issue seemed to be pointing at. There were two tests that seemed to depend on the old behavior that I cleaned up in order to bring into alignment with the new behavior, this
This cleans up several if liet statements but...
| let test = Some(Test { a: Some(3), b: 1 }); | ||
| //^^^^ Option<Test> |
There was a problem hiding this comment.
These changes are undesirable, the idea behind this "hiding" is to only hide info if it is redundant. In this case here specifically, only Some is mentioned, so the fact that this local is of the type Option<Test> is not redundant and as such should always be shown.
|
call point is: if config.hide_named_constructor_hints
&& is_named_constructor(sema, pat, &ty_name).is_some()hide_named_constructor_hints is false, and so we never evaluate the is_named_constructor due to short circuiting. |
|
Ah you need to enable the config for your test cases specifically as done here https://github.com/rust-analyzer/rust-analyzer/blob/9f447ad522a2e7464d827d5b51cdeb53a90c91f5/crates/ide/src/inlay_hints.rs#L1340-L1386 |
|
@emeryc any updates here? |
|
Actual solution is going to be different enough for this PR to be not worth it. Sorry I should have just closed this sooner. |
The suggested method in the issue didn't work because that code path appears not to be hit during the generation of these types. Instead I introduced a new concept of unambigious types as part of let statements and filtered out cases that the issue seemed to be pointing at.
There were two tests that seemed to depend on the old behavior that I cleaned up in order to bring into alignment with the new behavior, this
This should solve #10446