-
Notifications
You must be signed in to change notification settings - Fork 12.7k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Eagerly normalize HIR paths in new trait solver #113473
Eagerly normalize HIR paths in new trait solver #113473
Conversation
I think always using Make the fields of
|
@lcnr: In that case, I think I'd just rather put structurally_resolved_ty calls in all of the places where we actually require a structurally resolved type in hir typeck, rather than adding a 3rd field to |
That's also fine 👍 |
Actually, changing all the places where I would try the 3 field option above, but literally every current usage of |
cc9bb78
to
4973567
Compare
The job Click to see the possible cause of the failure (guessed by this bot)
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
nits, then r=me
@@ -3142,7 +3144,8 @@ impl<'a, 'tcx> FnCtxt<'a, 'tcx> { | |||
fields: &[Ident], | |||
expr: &'tcx hir::Expr<'tcx>, | |||
) -> Ty<'tcx> { | |||
let container = self.to_ty(container).normalized; | |||
// We only need to normalize in the old solver | |||
let container = self.normalize(expr.span, self.to_ty(container)); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
do we not normalize in the structurally_resolve_type
inside of the loop?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
No, structurally resolve only processes obligations in the old solver
} | ||
QPath::LangItem(..) => { | ||
bug!("`resolve_ty_and_res_fully_qualified_call` called on `LangItem`") | ||
} | ||
}; | ||
|
||
// FIXME(-Ztrait-solver=next): Can use `try_structurally_resolve` after migration. | ||
let normalized = |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
let normalized = | |
let normalized_ty = |
self.handle_raw_ty(span, tcx.at(span).type_of(impl_def_id).instantiate_identity()); | ||
match ty.normalized.ty_adt_def() { | ||
let ty = tcx.at(span).type_of(impl_def_id).instantiate_identity(); | ||
let normalized = self.structurally_normalize_after_astconv(span, ty); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
let normalized = self.structurally_normalize_after_astconv(span, ty); | |
let normalized_ty = self.structurally_normalize_after_astconv(span, ty); |
@@ -1367,7 +1388,7 @@ impl<'a, 'tcx> FnCtxt<'a, 'tcx> { | |||
def_id, | |||
&[], | |||
has_self, | |||
self_ty.map(|s| s.raw), | |||
self_ty.map(|s| s), |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
self_ty.map(|s| s), | |
self_ty, |
?
@@ -1346,25 +1346,31 @@ impl<'a, 'tcx> FnCtxt<'a, 'tcx> { | |||
) -> Result<(&'tcx ty::VariantDef, Ty<'tcx>), ErrorGuaranteed> { | |||
let path_span = qpath.span(); | |||
let (def, ty) = self.finish_resolving_struct_path(qpath, path_span, hir_id); | |||
let normalized = self.structurally_normalize_after_astconv(path_span, ty); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
let normalized = self.structurally_normalize_after_astconv(path_span, ty); | |
let normalized_ty = self.structurally_normalize_after_astconv(path_span, ty); |
@@ -324,6 +324,26 @@ impl<'a, 'tcx> FnCtxt<'a, 'tcx> { | |||
) | |||
} | |||
|
|||
// FIXME(-Ztrait-solver=next): This could be replaced with `try_structurally_resolve` | |||
// calls after the new trait solver is stable. | |||
/// TODO: I don't know what to call this. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
why is that method needed 🤔 because we don't want to normalize with the old solver here?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
No, because we need to normalize with the old solver here. This one calls the version of normalize that does both.
☔ The latest upstream changes (presumably #113199) made this pull request unmergeable. Please resolve the merge conflicts. |
@compiler-errors FYI: when a PR is ready for review, send a message containing |
I really don't have the patience to rework this PR. The way that |
always normalize `LoweredTy` in the new solver I currently expect us to stop using alias bound candidates of normalizable aliases due to rust-lang/trait-system-refactor-initiative#77 by landing rust-lang#119744. At this point it mostly doesn't matter whether we eagerly normalize (and replace with infer vars in case of ambiguity). cc rust-lang#113473 previous attempt The infer var replacement for ambiguous projections can in very rare cases: - weaken inference rust-lang/trait-system-refactor-initiative#81 - strengthen inference rust-lang/trait-system-refactor-initiative#7 I do not expect this impact on inference to significantly affect real crates. r? `@compiler-errors`
always normalize `LoweredTy` in the new solver I currently expect us to stop using alias bound candidates of normalizable aliases due to rust-lang/trait-system-refactor-initiative#77 by landing rust-lang#119744. At this point it mostly doesn't matter whether we eagerly normalize (and replace with infer vars in case of ambiguity). cc rust-lang#113473 previous attempt The infer var replacement for ambiguous projections can in very rare cases: - weaken inference rust-lang/trait-system-refactor-initiative#81 - strengthen inference rust-lang/trait-system-refactor-initiative#7 I do not expect this impact on inference to significantly affect real crates. r? ``@compiler-errors``
Rollup merge of rust-lang#120378 - lcnr:normalize-ast, r=compiler-errors always normalize `LoweredTy` in the new solver I currently expect us to stop using alias bound candidates of normalizable aliases due to rust-lang/trait-system-refactor-initiative#77 by landing rust-lang#119744. At this point it mostly doesn't matter whether we eagerly normalize (and replace with infer vars in case of ambiguity). cc rust-lang#113473 previous attempt The infer var replacement for ambiguous projections can in very rare cases: - weaken inference rust-lang/trait-system-refactor-initiative#81 - strengthen inference rust-lang/trait-system-refactor-initiative#7 I do not expect this impact on inference to significantly affect real crates. r? ``@compiler-errors``
TODO: needs a new description now
r? @lcnr