rework GAT borrowck limitation error #145041
Conversation
rustbot
added
S-waiting-on-review
|
This PR modifies |
| && let ConstraintCategory::Predicate(span) = constraint.category | ||
| && let NllRegionVariableOrigin::Placeholder(placeholder) = origin.origin | ||
| && let Some(id) = placeholder.bound.kind.get_id() | ||
| && let Some(placeholder_id) = id.as_local() |
There was a problem hiding this comment.
This lint should ideally also work for placeholders from foreign crates by looking at the bound vars and span from the parent predicates_of.
Changing that is annoying and I did not do it in this PR.
There was a problem hiding this comment.
This logic will probably help a lot with various order-dependent things I introduce in my PRs, since selection of the correct placeholder is a lot less sensitive now! Thanks!
| note: due to current limitations in the borrow checker, this implies a `'static` lifetime | ||
| --> $DIR/implementation-not-general-enough-ice-133252.rs:16:18 | ||
| | | ||
| LL | fn force_send<T: Send>(_: T) {} |
There was a problem hiding this comment.
The fact that this note worked and was also applicable here feels somewhat incidental to me. It's caused by incorrectly handling the existential regions in opaque types. This is an instantiation of #110338 for opaque types instead of coroutines
lcnr
force-pushed
the
borrowck-limitations-error
branch
from
5a3ddbc to
ba8fa8f
Compare
lcnr
force-pushed
the
borrowck-limitations-error
branch
from
07a0562 to
da17a53
Compare
|
r? BoxyUwU |
|
r=me after comment or not |
BoxyUwU
added
S-waiting-on-author
lcnr
force-pushed
the
borrowck-limitations-error
branch
from
da17a53 to
3ebf611
Compare
The old one depends on the
ConstraintCategoryof the constraint which meant we did not emit this note if we had to prove the higher ranked trait bound due to e.g. normalization.This made it annoying brittle and caused MIR borrowck errors to be order dependent, fixes the issue in #140737 (comment).
r? types cc @amandasystems