type alias impl trait: add tests showing that hidden type only outlives lifetimes that occur in bounds #103876
Conversation
|
Thanks for the pull request, and welcome! The Rust team is excited to review your changes, and you should hear from @lcnr (or someone else) soon. Please see the contribution instructions for more information. |
rustbot
added
S-waiting-on-review
oli-obk
changed the title
|
We also need a similar fix for outlive components. For example this should also pass: // ... same test as #102417
fn want_static<T: 'static>(_: T) {}
fn test<'a>() {
want_static(<&'a () as Callable>::call());
}Now with all these non-local workarounds I wonder if the generics should instead be modified during HIR lowering or in the query
lol why rustbot assumes you are a new contributor? |
yea, that's what I looked at first, but couldn't actually hit it. Thanks for building a repro |
oh definitely. That's actually what I tried first in #102417, but that is very hard at present. We basically need #96840 first in order to be able to do this nicely. |
| let z = y.foo(); | ||
| let _a = &x; // invalidate the `&'a mut`in `y` | ||
| //~^ ERROR cannot borrow `x` | ||
| let _b = z; // this should *not* check that `'a` in the type `Foo<'a>::foo::opaque` is live |
There was a problem hiding this comment.
that comment seems wrong? If you have FooX<'a> with an explicit 'a the error feels correct? 🤔
can you add a comment to the top of the file explaining what we're testing here?
There was a problem hiding this comment.
that comment seems wrong? If you have
FooX<'a>with an explicit'athe error feels correct?
that's the thing... lifetimes on type aliases should not matter for anything. Only lifetimes in the aliased type. Once/if we avoid adding uncaptured lifetimes to opaque types, this would automatically end up compiling. It may end up working with #103491 or with that plus some small changes.
|
☔ The latest upstream changes (presumably #103491) made this pull request unmergeable. Please resolve the merge conflicts. |
oli-obk
force-pushed
the
tait_implications
branch
from
a08d7b7
to
ca57832
Compare
oli-obk
changed the title
|
This entire PR has become unnecessary with #103491, so all it does now is add tests |
|
@bors r+ rollup |
bors
added
S-waiting-on-bors
…iaskrgr Rollup of 14 pull requests Successful merges: - rust-lang#103876 (type alias impl trait: add tests showing that hidden type only outlives lifetimes that occur in bounds) - rust-lang#104427 (Explain why `rematch_impl` fails to be infallible) - rust-lang#104436 (Add slice to the stack allocated string comment) - rust-lang#104523 (Don't use periods in target names) - rust-lang#104627 (Print all features with --print target-features) - rust-lang#104911 (Make inferred_outlives_crate return Clause) - rust-lang#105002 (Add `PathBuf::as_mut_os_string` and `Path::as_mut_os_str`) - rust-lang#105023 (Statics used in reachable function's inline asm are reachable) - rust-lang#105045 (`rustc_ast_{passes,pretty}`: remove `ref` patterns) - rust-lang#105049 (Hermit: Minor build fixes) - rust-lang#105051 (Replace a macro with a function) - rust-lang#105062 (rustdoc: use shorthand background for rustdoc toggle CSS) - rust-lang#105066 (move `candidate_from_obligation` out of assembly) - rust-lang#105068 (Run patchelf also on rust-analyzer-proc-macro-srv.) Failed merges: - rust-lang#105050 (Remove useless borrows and derefs) r? `@ghost` `@rustbot` modify labels: rollup
fixes #103642
#102417 only made sure that hidden types cannot outlive lifetimes other than the ones mentioned on bounds, but didn't allow us to actually infer anything from that.
cc @aliemjay