Prepare for lowering or-patterns
This is probably best reviewed commit-by-commit.
* Rustfmt `rustc_mir::build::{self, matches}`
* Remove `-Znll-dont-emit-read-for-match`
* Split `match_expr` into smaller functions
* Feature gate or-patterns in const contexts
cc https://github.com/rust-lang/rust/issues/54883
r? @Centril
Tweak errors for missing associated types and type parameters
* On `dyn Trait` missing associated types, provide a structured suggestion for them
* On missing type parameters, provide structured suggestion for them
* Point at trait definition when missing required type parameter
* Tweak output of E0658
* Tweak wording of E0719
* Account for `Trait1 + Trait2` case
Fix#66380, fix#60595. CC #63711.
Deprecate Error::description for real
`description` has been documented as soft-deprecated since 1.27.0 (17 months ago). There is no longer any reason to call it or implement it.
This PR:
- adds `#[rustc_deprecated(since = "1.41.0")]` to `Error::description`;
- moves `description` (and `cause`, which is also deprecated) below the `source` and `backtrace` methods in the Error trait;
- reduces documentation of `description` and `cause` to take up much less vertical real estate in rustdocs, while preserving the example that shows how to render errors without needing to call `description`;
- removes the description function of all *currently unstable* Error impls in the standard library;
- marks `#[allow(deprecated)]` the description function of all *stable* Error impls in the standard library;
- replaces miscellaneous uses of `description` in example code and the compiler.
---
Generalize `array_into_iter` lint to also lint for boxed arrays
`Box` is special in that a method call on a box can move the value out
of the box. Thus, the same backwards-compatibility problem can arise
for boxed arrays as for simple arrays.
---
CC #66145
r? @matthewjasper (as you reviewed the first PR)
`description` has been documented as soft-deprecated since 1.27.0 (17
months ago). There is no longer any reason to call it or implement it.
This commit:
- adds #[rustc_deprecated(since = "1.41.0")] to Error::description;
- moves description (and cause, which is also deprecated) below the
source and backtrace methods in the Error trait;
- reduces documentation of description and cause to take up much less
vertical real estate in rustdocs, while preserving the example that
shows how to render errors without needing to call description;
- removes the description function of all *currently unstable* Error
impls in the standard library;
- marks #[allow(deprecated)] the description function of all *stable*
Error impls in the standard library;
- replaces miscellaneous uses of description in example code and the
compiler.
This still doesn't handle the case entirely correctly, requiring a more
targeted approach with a better suggestion, but at least now the
suggested syntax makes *some* sense.
Differentiate todo! and unimplemented!
This updates the panic message and docs to make it clear that `todo!` is for unfinished code and `unimplemented!` is for partial trait or enum impls.
r? @Centril
Refactorings to borrowck region diagnostic reporting
This PR is a followup to #66886 and #67404
EDIT: updated
In this PR: Clean up how errors are collected from NLL: introduce `borrow_check::diagnostics::RegionErrors` to collect errors. This is later passed to `MirBorrowckCtx::report_region_errors` after the `MirBorrowckCtx` is created. This will allow us to refactor away some of the extra context structs floating around (but we don't do it in this PR). `borrow_check::region_infer` is now mostly free of diagnostic generating code. The signatures of most of the functions in `region_infer` lose somewhere between 4 and 7 arguments :)
Left for future (feedback appreciated):
- Merge `ErrorRegionCtx` with `MirBorrowckCtx`, as suggested by @matthewjasper in https://github.com/rust-lang/rust/pull/66886#issuecomment-559949499
- Maybe move the contents of `borrow_check::nll` into `borrow_check` and remove the `nll` submodule altogether.
- Find a way to make `borrow_check::diagnostics::region_errors` less of a strange appendage to `RegionInferenceContext`. I'm not really sure how to do this yet. Ideas welcome.
`Box` is special in that a method call on a box can move the value out
of the box. Thus, the same backwards-compatibility problem can arise
for boxed arrays as for simple arrays.
Add regression tests for fixed ICEs
Closes#61747 (fixed from 1.41.0-nightly (4007d4ef2 2019-12-01))
Closes#66205 (fixed from 1.41.0-nightly (4007d4ef2 2019-12-01))
Closes#66270 (fixed by #66246)
Closes#67424 (fixed by #67160)
Also picking a minor nit up from #67071 with 101dd7bad9
r? @Centril
Ensure that evaluating or validating a constant never reads from a static
r? @RalfJung
as per https://github.com/rust-lang/rust/pull/66302#issuecomment-554663387
This does not yet address the fact that evaluation of a constant can read from a static (under unleash-miri)
Initial implementation of `#![feature(bindings_after_at)]`
Following up on #16053, under the gate `#![feature(bindings_after_at)]`, `x @ Some(y)` is allowed subject to restrictions necessary for soundness.
The implementation and test suite should be fairly complete now.
One aspect that is not covered is the interaction with nested `#![feature(or_patterns)]`.
This is not possible to test at the moment in a good way because that feature has not progressed sufficiently and has fatal errors in MIR building. We should make sure to add such tests before we stabilize both features (but shipping one of them is fine).
r? @pnkfelix
cc @nikomatsakis @matthewjasper @pcwalton
cc https://github.com/rust-lang/rust/issues/65490
Improve diagnostics for invalid assignment
- Improve wording and span information for invalid assignment diagnostics.
- Link to https://github.com/rust-lang/rfcs/issues/372 when it appears the user is trying a destructuring assignment.
- Make the equality constraint in `where` clauses error consistent with the invalid assignment error.
Remove mem::uninitalized from tests
This purges uses of uninitialized where possible from test cases. Some
are merely moved over to the equally bad pattern of
MaybeUninit::uninit().assume_init() but with an annotation that this is
"the best we can do".
Fixes#62397
