Add new alias kind Ambiguous and renormalize instantiated binders#156453
Draft
adwinwhite wants to merge 13 commits into
Draft
Add new alias kind Ambiguous and renormalize instantiated binders#156453adwinwhite wants to merge 13 commits into
Ambiguous and renormalize instantiated binders#156453adwinwhite wants to merge 13 commits into
Conversation
rustbot
added
S-waiting-on-author
Contributor
Author
|
@bors try @rust-timer queue |
Collaborator
|
Awaiting bors try build completion. @rustbot label: +S-waiting-on-perf |
rustbot
added
the
S-waiting-on-perf
This comment has been minimized.
This comment has been minimized.
rust-bors Bot
pushed a commit
that referenced
this pull request
May 11, 2026
Add new alias kind `Ambiguous` and renormalize instantiated binders
Collaborator
|
The job Click to see the possible cause of the failure (guessed by this bot) |
Contributor
Author
|
@bors try cancel |
Contributor
|
Try build cancelled. Cancelled workflows: |
Contributor
Author
|
Ignore tools failures for now. @bors try @rust-timer queue |
Collaborator
|
This pull request is already queued and waiting for a try build to finish. |
This comment has been minimized.
This comment has been minimized.
rust-bors Bot
pushed a commit
that referenced
this pull request
May 11, 2026
Add new alias kind `Ambiguous` and renormalize instantiated binders
Contributor
|
💔 Test for 1c015e6 failed: CI. Failed job:
|
Collaborator
|
The job Click to see the possible cause of the failure (guessed by this bot) |
lcnr
reviewed
May 11, 2026
|
|
||
| // Placeholders (all printed as `_` to uniformize them). | ||
| ty::Param(_) | ty::Bound(..) | ty::Placeholder(_) | ty::Infer(_) | ty::Error(_) => { | ||
| ty::Alias(ty::AliasTy { kind: ty::Ambiguous { .. }, .. }) |
Contributor
There was a problem hiding this comment.
unreachable
lcnr
reviewed
May 11, 2026
| Unnormalized::new_wip(unnormalized_external_impl_sig), | ||
| ); | ||
| let external_impl_sig = | ||
| ocx.normalize(&norm_cause, param_env, Unnormalized::new(unnormalized_external_impl_sig)); |
Contributor
There was a problem hiding this comment.
Suggested change
| ocx.normalize(&norm_cause, param_env, Unnormalized::new(unnormalized_external_impl_sig)); | |
| ocx.normalize(&norm_cause, param_env, Unnormalized::new_wip(unnormalized_external_impl_sig)); |
ideally this should be a map_unnormalized call in which we instantiate the binder
lcnr
reviewed
May 11, 2026
| ty::Alias(ty::AliasTy { kind: ty::Ambiguous { .. }, .. }) => { | ||
| let sp = tcx.def_span(impl_def_id); | ||
| span_bug!(sp, "weird self type for autotrait impl") | ||
| } |
Contributor
There was a problem hiding this comment.
redundant?
lcnr
reviewed
May 11, 2026
|
|
||
| // Placeholders (all printed as `_` to uniformize them). | ||
| ty::Param(_) | ty::Bound(..) | ty::Placeholder(_) | ty::Infer(_) | ty::Error(_) => { | ||
| ty::Alias(ty::AliasTy { kind: ty::Ambiguous { .. }, .. }) |
Contributor
There was a problem hiding this comment.
unreachable
Contributor
There was a problem hiding this comment.
can you also make ty::Infer unreachable?
lcnr
reviewed
May 11, 2026
Comment on lines
+49
to
+50
| // Now that we perform eager normalization inside the solver in some places, | ||
| // The aliases might be normalized away. |
Contributor
There was a problem hiding this comment.
when do we do eager norm for the inputs of alias relate goals? 🤔
lcnr
reviewed
May 11, 2026
| /// next solver. See `NormalizationFolder`. | ||
| /// | ||
| /// The def_id and args are the same as the original alias. | ||
| AmbiguousTy { def_id: I::DefId }, |
Contributor
There was a problem hiding this comment.
why does this contain the def_id
lcnr
reviewed
May 11, 2026
| } | ||
| } | ||
|
|
||
| pub fn instantiate_binder_with_fresh_vars_and_normalize_with<T, F>( |
Contributor
There was a problem hiding this comment.
hmm, this is different from what I'd imagine 🤔
I would expect us to have
fn instantiate_binder_no_ambig_aliases(x: Binder<T>) -> T {
instantiate_binder_renormalize_ambig_aliases(x, |_| unreachable!())
}
fn instantiate_binder_renormalize_ambig_aliases(x: Binder<T>, normalize: impl FnMut(AliasTerm<'tcx>) -> Term<'tcx>) -> T {
instantiate binder
walk with renormalize ambig aliases folder
}I'd then keep the borrowck code etc the same as rn while using instantiate_binder_no_ambig_aliases and use instantiate_binder_renormalize_ambig_aliases only in the places that can encounter ambig aliases
Contributor
There was a problem hiding this comment.
where all the old solver exclusive code can just use instantiate_binder_no_ambig_aliases
Contributor
There was a problem hiding this comment.
and the new solver places that currently can't easily use a normalization routine can just recreate the ty::Alias or ConstKind::Alias with a FIXME that they should instead normalize
lcnr
reviewed
May 11, 2026
| self.print_def_path(def_id, args)?; | ||
| ty::Alias(ref data @ ty::AliasTy { kind, args, .. }) => match kind | ||
| .reveal_ambiguous(self.tcx()) | ||
| { |
Contributor
There was a problem hiding this comment.
hmm, instead of reeal_ambiguous Ambiguous aliases can just forward to their contained alias?
lcnr
reviewed
May 11, 2026
Comment on lines
+660
to
+667
| // Convert `AmbiguousTy` into its original kind. | ||
| pub fn reveal_ambiguous(self, interner: I) -> Self { | ||
| if let AliasTermKind::AmbiguousTy { def_id } = self { | ||
| interner.alias_term_kind_from_def_id(def_id) | ||
| } else { | ||
| self | ||
| } | ||
| } |
Contributor
There was a problem hiding this comment.
ah, so what you do is ty::Alias(Projection { def_id }, [Self, T]) to ty::Alias(Ambiguous { def_id }, [Self, T]) instead of
ty::Alias(Projection { def_id }, [Self, T]) to ty::Alias(Ambiguous, ty::Alias(Projection { def_id }, [Self, T])]) which is what I would have expected 🤔
lcnr
reviewed
May 11, 2026
lcnr
reviewed
May 11, 2026
| ) -> InferOk<'tcx, T> { | ||
| if self.infcx.next_trait_solver() { | ||
| let Normalized { value, obligations } = | ||
| crate::solve::normalize(*self, value, NormalizationScope::AmbiguousAlias); |
Contributor
There was a problem hiding this comment.
i would not add support for this to the old trait solver at all 🤔 i would only ever use ambiguous aliases in the new solver, so here ambiguous aliases should be unreachable
lcnr
reviewed
May 11, 2026
| &self, | ||
| value: ty::Binder<Self::Interner, T>, | ||
| f: impl FnOnce(T) -> U, | ||
| f: impl FnOnce(ty::Unnormalized<Self::Interner, T>) -> U, |
Contributor
There was a problem hiding this comment.
Suggested change
| f: impl FnOnce(ty::Unnormalized<Self::Interner, T>) -> U, | |
| normalize_ambiguous_aliases: impl FnMut(ALiasTerm<'tcx>) -> Term<'tcx> | |
| f: impl FnOnce(T) -> U, |
is what I would have liked🤔 I personally feel like Unnormalized for both being fully unnormalized and only having to renormalize ambiguous aliases feels suboptimal to me
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Still WIP.
- fix rustdoc
- also normalize for diagnostics and assert the remaining instantiation doesn't need renormalization.
- remove the accidental staged
cargosubmoduleBut maybe you wanna have a quick look first.
r? lcnr