Conversation
rustbot
added
S-waiting-on-author
This comment has been minimized.
This comment has been minimized.
lcnr
force-pushed
the
find-param-in-clause-ambig
branch
from
3564a3f to
755953f
Compare
This comment has been minimized.
This comment has been minimized.
see the added test
lcnr
force-pushed
the
find-param-in-clause-ambig
branch
from
755953f to
a4564e7
Compare
lcnr
changed the title
FindParamInClause global on overflowFindParamInClause handle edge-cases
|
|
rustbot
added
S-waiting-on-review
|
r? BoxyUwU |
|
Requested reviewer is already assigned to this pull request. Please choose another assignee. |
| TypeFlags::HAS_PLACEHOLDER | TypeFlags::HAS_INFER | TypeFlags::HAS_ALIAS, | ||
| ) => | ||
| { | ||
| ct.super_visit_with(self) |
There was a problem hiding this comment.
why the not doing recursion depth stuff here 🤔
There was a problem hiding this comment.
existing xd :3 because the only way constants recurse is through types atm
There was a problem hiding this comment.
I guess this has to change with mgca
There was a problem hiding this comment.
up to u if u want to handle it in this PR, it'd be a fairly reasonable FIXME(mgca) to get some new-ish contributor to work on as onboarding
| })); | ||
| ty::Infer(_) => ControlFlow::Break(Ok(Certainty::AMBIGUOUS)), | ||
| _ if ty.has_type_flags( | ||
| TypeFlags::HAS_PLACEHOLDER | TypeFlags::HAS_INFER | TypeFlags::HAS_ALIAS, |
There was a problem hiding this comment.
so there are two meaningful changes here:
- we now recurse if there are aliases
- we now recurse if there are non-region-infers
recursing if there are aliases makes sense (that's certainly the point of this PR) but I don't fully understand why we care about infer vars or not.
There was a problem hiding this comment.
to treat ambiguous normalization as Global. You could imagine normalizing T::Assoc to Vec<T::AmbiguousAlias> which ends up as Vec<?infer>
I guess we could always force GLobal if there are infer vars instead of recursing
There was a problem hiding this comment.
do we have a test with ambiguous aliases? I guess you could only really get ambiguity here via overflow (are opaque types relevant?)?
4 participants
If normalization is ambiguous, we want to treat the where-bound as non-global. The affected code should pretty much always have other errors in that case. It does make reasoning about trait solver cycles and rerunning more confusing and just feels wrong ™️ I encountered this while looking into
ml-kem(rust-lang/trait-system-refactor-initiative#246). I don't know whether this matters in practice and don't care to look into it too deeply.We also want to properly handle concrete aliases which mention to something mentioning a generic parameter due to another where-bound.
r? @BoxyUwU