-
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
Ban multi-trait objects via trait aliases #59445
Ban multi-trait objects via trait aliases #59445
Conversation
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
ce751e4
to
e728b8d
Compare
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
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.
Implementation looks mostly good aside from comments below... The diagnostics look great.
However, I think some more tests are needed. In particular, it would be nice to exercise the recursive behavior here including things like dyn A + B
where trait A = C + D;
, trait B = E + F
, trait F = G + 'static;
and various other combinations including auto traits, region constraints, where clauses on the trait
alias, and so on.
47d6cd3
to
c0ef111
Compare
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
@@ -1175,14 +1198,13 @@ impl<'o, 'gcx: 'tcx, 'tcx> dyn AstConv<'gcx, 'tcx> + 'o { | |||
} else { | |||
self.re_infer(span, None).unwrap_or_else(|| { | |||
span_err!(tcx.sess, span, E0228, | |||
"the lifetime bound for this object type cannot be deduced \ | |||
from context; please supply an explicit bound"); | |||
"the lifetime bound for this object type cannot be deduced \ |
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.
"the lifetime bound for this object type cannot be deduced \ | |
"the lifetime bound for this object type cannot be inferred \ |
while reading this PR again, I noticed this... we use "infer" elsewhere and so we should prefer that everywhere instead of C++isms like "deduce".
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.
That should be a separate PR I think. You're welcome to make it though, and I can review. ;-)
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.
@alexreg Sure; leaving it here (unresolved) for visibility and as a reminder ;)
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.
@Centril With this PR merging now, just a note for you that this is unresolved, in case you want to create an issue or PR for it yourself.
a54c090
to
50871cb
Compare
This comment has been minimized.
This comment has been minimized.
50871cb
to
d2993e9
Compare
This comment has been minimized.
This comment has been minimized.
d2993e9
to
0f94b88
Compare
0f94b88
to
2590f72
Compare
This comment has been minimized.
This comment has been minimized.
2590f72
to
d408a1f
Compare
This comment has been minimized.
This comment has been minimized.
d408a1f
to
a42c23b
Compare
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
a42c23b
to
560a8de
Compare
is the beta cut already successfull even without #60999 having been merged? |
Not 100% sure, but @Centril and I thought that by the time this PR works its way up the queue now, the toolstate problem should have disappeared at least, so it should be no problem. |
⌛ Testing commit ce2ee30 with merge de414067da0f3f6c80530b9ba2a9f98e90c2f484... |
💔 Test failed - checks-travis |
The job Click to expand the log.
I'm a bot! I can only do what humans tell me to, so if this was not helpful or you have suggestions for improvements, please ping or otherwise contact |
#61027 already broke Clippy so for sure tools don't block it anymore. |
@bors retry |
…=oli-obk Ban multi-trait objects via trait aliases Obviously, multi-trait objects are not normally supported, so they should not be supported via trait aliases. This has been factored out from the previous PR #55994 (see point 1). r? @Centril CC @nikomatsakis ------------------ ### RELNOTES: We now allow `dyn Send + fmt::Debug` with equivalent semantics to `dyn fmt::Debug + Send`. That is, the order of the mentioned traits does not matter wrt. principal/not-principal traits. This is a small change that might deserve a mention in the blog post because it is a language change but most likely not. See https://github.com/rust-lang/rust/blob/ce2ee305f9165c037ecddddb5792588a15ff6c37/src/test/ui/traits/wf-trait-object-reverse-order.rs. // @Centril
☀️ Test successful - checks-travis, status-appveyor |
Pkgsrc changes: * NetBSD/sparc64 disabling of "packed" removed ("packed" removed upstream) * Adapt src_libstd_build.rs patch, update sed'ing of .cargo-checksum.json Build verified on NetBSD 8.0/amd64. Upstream changes: Version 1.36.0 (2019-07-04) ========================== Language -------- - [Non-Lexical Lifetimes are now enabled on the 2015 edition.][59114] - [The order of traits in trait objects no longer affects the semantics of that object.][59445] e.g. `dyn Send + fmt::Debug` is now equivalent to `dyn fmt::Debug + Send`, where this was previously not the case. Libraries --------- - [`HashMap`'s implementation has been replaced with `hashbrown::HashMap` implem entation.][58623] - [`TryFromSliceError` now implements `From<Infallible>`.][60318] - [`mem::needs_drop` is now available as a const fn.][60364] - [`alloc::Layout::from_size_align_unchecked` is now available as a const fn.][6 0370] - [`String` now implements `BorrowMut<str>`.][60404] - [`io::Cursor` now implements `Default`.][60234] - [Both `NonNull::{dangling, cast}` are now const fns.][60244] - [The `alloc` crate is now stable.][59675] `alloc` allows you to use a subset of `std` (e.g. `Vec`, `Box`, `Arc`) in `#![no_std]` environments if the environment has access to heap memory allocation. - [`String` now implements `From<&String>`.][59825] - [You can now pass multiple arguments to the `dbg!` macro.][59826] `dbg!` will return a tuple of each argument when there is multiple arguments. - [`Result::{is_err, is_ok}` are now `#[must_use]` and will produce a warning if not used.][59648] Stabilized APIs --------------- - [`VecDeque::rotate_left`] - [`VecDeque::rotate_right`] - [`Iterator::copied`] - [`io::IoSlice`] - [`io::IoSliceMut`] - [`Read::read_vectored`] - [`Write::write_vectored`] - [`str::as_mut_ptr`] - [`mem::MaybeUninit`] - [`pointer::align_offset`] - [`future::Future`] - [`task::Context`] - [`task::RawWaker`] - [`task::RawWakerVTable`] - [`task::Waker`] - [`task::Poll`] Cargo ----- - [Cargo will now produce an error if you attempt to use the name of a required dependency as a feature.][cargo/6860] - [You can now pass the `--offline` flag to run cargo without accessing the netw ork.][cargo/6934] You can find further change's in [Cargo's 1.36.0 release notes][cargo-1-36-0]. Clippy ------ There have been numerous additions and fixes to clippy, see [Clippy's 1.36.0 rel ease notes][clippy-1-36-0] for more details. Misc ---- Compatibility Notes ------------------- - [`std::arch::x86::_rdtsc` returns `u64` instead of `i64`][stdsimd/559] - [`std::arch::x86_64::_mm_shuffle_ps` takes an `i32` instead of `u32` for `mask `][stdsimd/522] - With the stabilisation of `mem::MaybeUninit`, `mem::uninitialized` use is no longer recommended, and will be deprecated in 1.38.0. [60318]: rust-lang/rust#60318 [60364]: rust-lang/rust#60364 [60370]: rust-lang/rust#60370 [60404]: rust-lang/rust#60404 [60234]: rust-lang/rust#60234 [60244]: rust-lang/rust#60244 [58623]: rust-lang/rust#58623 [59648]: rust-lang/rust#59648 [59675]: rust-lang/rust#59675 [59825]: rust-lang/rust#59825 [59826]: rust-lang/rust#59826 [59445]: rust-lang/rust#59445 [59114]: rust-lang/rust#59114 [cargo/6860]: rust-lang/cargo#6860 [cargo/6934]: rust-lang/cargo#6934 [`VecDeque::rotate_left`]: https://doc.rust-lang.org/std/collections/struct.VecDeque.html#method.rotate_left [`VecDeque::rotate_right`]: https://doc.rust-lang.org/std/collections/struct.VecDeque.html#method.rotate_right [`Iterator::copied`]: https://doc.rust-lang.org/std/iter/trait.Iterator.html#tymethod.copied [`io::IoSlice`]: https://doc.rust-lang.org/std/io/struct.IoSlice.html [`io::IoSliceMut`]: https://doc.rust-lang.org/std/io/struct.IoSliceMut.html [`Read::read_vectored`]: https://doc.rust-lang.org/std/io/trait.Read.html#method.read_vectored [`Write::write_vectored`]: https://doc.rust-lang.org/std/io/trait.Write.html#method.write_vectored [`str::as_mut_ptr`]: https://doc.rust-lang.org/std/primitive.str.html#method.as_mut_ptr [`mem::MaybeUninit`]: https://doc.rust-lang.org/std/mem/union.MaybeUninit.html [`pointer::align_offset`]: https://doc.rust-lang.org/std/primitive.pointer.html#method.align_offset [`future::Future`]: https://doc.rust-lang.org/std/future/trait.Future.html [`task::Context`]: https://doc.rust-lang.org/beta/std/task/struct.Context.html [`task::RawWaker`]: https://doc.rust-lang.org/beta/std/task/struct.RawWaker.html [`task::RawWakerVTable`]: https://doc.rust-lang.org/beta/std/task/struct.RawWakerVTable.html [`task::Waker`]: https://doc.rust-lang.org/beta/std/task/struct.Waker.html [`task::Poll`]: https://doc.rust-lang.org/beta/std/task/enum.Poll.html [clippy-1-36-0]: https://github.com/rust-lang/rust-clippy/blob/master/CHANGELOG.md#rust-136 [cargo-1-36-0]: https://github.com/rust-lang/cargo/blob/master/CHANGELOG.md#cargo-136-2019-07-04 [stdsimd/522]: rust-lang/stdarch#522 [stdsimd/559]: rust-lang/stdarch#559
Obviously, multi-trait objects are not normally supported, so they should not be supported via trait aliases.
This has been factored out from the previous PR #55994 (see point 1).
r? @Centril
CC @nikomatsakis
RELNOTES:
We now allow
dyn Send + fmt::Debug
with equivalent semantics todyn fmt::Debug + Send
.That is, the order of the mentioned traits does not matter wrt. principal/not-principal traits.
This is a small change that might deserve a mention in the blog post because it is a language change but most likely not.
See https://github.com/rust-lang/rust/blob/ce2ee305f9165c037ecddddb5792588a15ff6c37/src/test/ui/traits/wf-trait-object-reverse-order.rs.
// @Centril