Slightly improve mismatched GAT where clause error #99353
Conversation
rustbot
added
the
T-compiler
|
r? @cjgillot (rust-highfive has picked a reviewer for you, use r? to override) |
rust-highfive
added
the
S-waiting-on-review
compiler-errors
force-pushed
the
gat-where-clause-mismatch
branch
from
134a487 to
3d3bf10
Compare
| @@ -1,20 +1,22 @@ | |||
| error: `impl` associated type signature for `A` doesn't match `trait` associated type signature | |||
| --> $DIR/impl_bounds.rs:15:5 | |||
| error[E0310]: the parameter type `T` may not live long enough | |||
There was a problem hiding this comment.
This is weird, lol, but seems more like a side-effect than due to these changes. I bet if we do the same setup with methods it would report the same error.
There was a problem hiding this comment.
Shouldn't we complain here that the bounds have changed? There should be something saying that the Self: 'static bound does not match the Self: 'a bound from the trait.
There was a problem hiding this comment.
It think the problem here is that we're not just comparing the bounds syntactically, but instead using the lifetime machinery.
So "given T: 'a, does T: 'static" (or vice versa) ends up erroring out in a harder to understand way than just a trait bounds mismatch... lifetime bounds mismatch sometimes can't even point to a bound at all when it's, for example, implied by the self type of the impl.
There was a problem hiding this comment.
I'm a bit suprised. In the case of a method, we get the expected "impl has stricter requirements than trait" error here too: https://play.rust-lang.org/?version=nightly&mode=debug&edition=2021&gist=8dada6c436de5aadf8d5a4d2f18cfd7d
There was a problem hiding this comment.
Fixed in 2bbcdc7, then I stacked another commit on top that combines the obligation cause codes so something like this is harder to miss in the future (and adjusts the wording a bit).
compiler-errors
force-pushed
the
gat-where-clause-mismatch
branch
from
3d3bf10 to
9734d79
Compare
compiler-errors
force-pushed
the
gat-where-clause-mismatch
branch
from
9734d79 to
3bbe95c
Compare
|
@bors r+ |
bors
added
S-waiting-on-bors
Rollup of 7 pull requests Successful merges: - rust-lang#98211 (Implement `fs::get_path` for FreeBSD.) - rust-lang#99353 (Slightly improve mismatched GAT where clause error) - rust-lang#99593 (Suggest removing the tuple struct field for the unwrapped value) - rust-lang#99615 (Remove some explicit `self.infcx` for `FnCtxt`, which already derefs into `InferCtxt`) - rust-lang#99711 (Remove reachable coverage without counters) - rust-lang#99718 (Avoid `&str`/`Symbol` to `String` conversions) - rust-lang#99720 (Sync rustc_codegen_cranelift) Failed merges: r? `@ghost` `@rustbot` modify labels: rollup
Rollup of 7 pull requests Successful merges: - rust-lang#98211 (Implement `fs::get_path` for FreeBSD.) - rust-lang#99353 (Slightly improve mismatched GAT where clause error) - rust-lang#99593 (Suggest removing the tuple struct field for the unwrapped value) - rust-lang#99615 (Remove some explicit `self.infcx` for `FnCtxt`, which already derefs into `InferCtxt`) - rust-lang#99711 (Remove reachable coverage without counters) - rust-lang#99718 (Avoid `&str`/`Symbol` to `String` conversions) - rust-lang#99720 (Sync rustc_codegen_cranelift) Failed merges: r? `@ghost` `@rustbot` modify labels: rollup
This makes the error reporting a bit more standardized between
whereon GATs and functions.cc #99206 (@BoxyUwU), don't want to mark this as as "fixed" because they're still not perfect, but this is still an improvement IMO so I want to land it incrementally.
regarding "consider adding where clause to trait definition", we don't actually do that for methods as far as i can tell? i could file an issue to look into that maybe.