Add inherent methods in libcore for [T], [u8], str, f32, and f64
# Background
Primitive types are defined by the language, they don’t have a type definition like `pub struct Foo { … }` in any crate. So they don’t “belong” to any crate as far as `impl` coherence is concerned, and on principle no crate would be able to define inherent methods for them, without a trait. Since we want these types to have inherent methods anyway, the standard library (with cooperation from the compiler) bends this rule with code like [`#[lang = "u8"] impl u8 { /*…*/ }`](https://github.com/rust-lang/rust/blob/1.25.0/src/libcore/num/mod.rs#L2244-L2245). The `#[lang]` attribute is permanently-unstable and never intended to be used outside of the standard library.
Each lang item can only be defined once. Before this PR there is one impl-coherence-rule-bending lang item per primitive type (plus one for `[u8]`, which overlaps with `[T]`). And so one `impl` block each. These blocks for `str`, `[T]` and `[u8]` are in liballoc rather than libcore because *some* of the methods (like `<[T]>::to_vec(&self) -> Vec<T> where T: Clone`) need a global memory allocator which we don’t want to make a requirement in libcore. Similarly, `impl f32` and `impl f64` are in libstd because some of the methods are based on FFI calls to C’s `libm` and we want, as much as possible, libcore not to require “runtime support”.
In libcore, the methods of `str` and `[T]` that don’t allocate are made available through two **unstable traits** `StrExt` and `SliceExt` (so the traits can’t be *named* by programs on the Stable release channel) that have **stable methods** and are re-exported in the libcore prelude (so that programs on Stable can *call* these methods anyway). Non-allocating `[u8]` methods are not available in libcore: https://github.com/rust-lang/rust/issues/45803. Some `f32` and `f64` methods are in an unstable `core::num::Float` trait with stable methods, but that one is **not in the libcore prelude**. (So as far as Stable programs are concerns it doesn’t exist, and I don’t know what the point was to mark these methods `#[stable]`.)
https://github.com/rust-lang/rust/issues/32110 is the tracking issue for these unstable traits.
# High-level proposal
Since the standard library is already bending the rules, why not bend them *a little more*? By defining a few additional lang items, the compiler can allow the standard library to have *two* `impl` blocks (in different crates) for some primitive types.
The `StrExt` and `SliceExt` traits still exist for now so that we can bootstrap from a previous-version compiler that doesn’t have these lang items yet, but they can be removed in next release cycle. (`Float` is used internally and needs to be public for libcore unit tests, but was already `#[doc(hidden)]`.) I don’t know if https://github.com/rust-lang/rust/issues/32110 should be closed by this PR, or only when the traits are entirely removed after we make a new bootstrap compiler.
# Float methods
Among the methods of the `core::num::Float` trait, three are based on LLVM intrinsics: `abs`, `signum`, and `powi`. PR https://github.com/rust-lang/rust/pull/27823 “Remove dependencies on libm functions from libcore” moved a bunch of `core::num::Float` methods back to libstd, but left these three behind. However they aren’t specifically discussed in the PR thread. The `compiler_builtins` crate defines `__powisf2` and `__powidf2` functions that look like implementations of `powi`, but I couldn’t find a connection with the `llvm.powi.f32` and `llvm.powi.f32` intrinsics by grepping through LLVM’s code.
In discussion starting at https://github.com/rust-lang/rust/issues/32110#issuecomment-370647922 Alex says that we do not want methods in libcore that require “runtime support”, but it’s not clear whether that applies to these `abs`, `signum`, or `powi`. In doubt, I’ve **removed** them for the trait and moved them to inherent methods in libstd for now. We can move them back later (or in this PR) if we decide that’s appropriate.
# Change details
For users on the Stable release channel:
* I believe this PR does not make any breaking change
* Some methods for `[u8]`, `f32`, and `f64` are newly available to `#![no_std]` users (fixes https://github.com/rust-lang/rust/issues/45803)
* There should be no visible change for `std` users in terms of what programs compile or what their behavior is. (Only in compiler error messages, possibly.)
For Nightly users, additionally:
* The unstable `StrExt` and `SliceExt` traits are gone
* Their methods are now inherent methods of `str` and `[T]` (so only code that explicitly named the traits should be affected, not "normal" method calls)
* The `abs`, `signum` and `powi` methods of the `Float` trait are gone
* The `Float` trait’s unstable feature name changed to `float_internals` with no associated tracking issue, to reflect it being a permanently unstable implementation detail rather than a public API on a path to stabilization.
* Its remaining methods are now inherent methods of `f32` and `f64`.
-----
CC @rust-lang/libs for the API changes, @rust-lang/compiler for the new lang items
Revert stabilization of never_type (!) et al
Fix#49691
I *think* this correctly adopts @nikomatsakis 's desired fix of:
* reverting stabilization of `!` and `TryFrom`, and
* returning to the previous fallback semantics (i.e. it is once again dependent on whether the crate has opted into `#[feature(never_type)]`,
* **without** attempting to put back in the previous future-proofing warnings regarding the change in fallback semantics.
(I'll be away from computers for a week starting now, so any updates to this PR should be either pushed into it, or someone else should adopt the task of polishing this fix and put up their own PR.)
rustc: Tweak custom attribute capabilities
This commit starts to lay some groundwork for the stabilization of custom
attribute invocations and general procedural macros. It applies a number of
changes discussed on [internals] as well as a [recent issue][issue], namely:
* The path used to specify a custom attribute must be of length one and cannot
be a global path. This'll help future-proof us against any ambiguities and
give us more time to settle the precise syntax. In the meantime though a bare
identifier can be used and imported to invoke a custom attribute macro. A new
feature gate, `proc_macro_path_invoc`, was added to gate multi-segment paths
and absolute paths.
* The set of items which can be annotated by a custom procedural attribute has
been restricted. Statements, expressions, and modules are disallowed behind
two new feature gates: `proc_macro_expr` and `proc_macro_mod`.
* The input to procedural macro attributes has been restricted and adjusted.
Today an invocation like `#[foo(bar)]` will receive `(bar)` as the input token
stream, but after this PR it will only receive `bar` (the delimiters were
removed). Invocations like `#[foo]` are still allowed and will be invoked in
the same way as `#[foo()]`. This is a **breaking change** for all nightly
users as the syntax coming in to procedural macros will be tweaked slightly.
* Procedural macros (`foo!()` style) can only be expanded to item-like items by
default. A separate feature gate, `proc_macro_non_items`, is required to
expand to items like expressions, statements, etc.
Closes#50038
[internals]: https://internals.rust-lang.org/t/help-stabilize-a-subset-of-macros-2-0/7252
[issue]: https://github.com/rust-lang/rust/issues/50038
This commit starts to lay some groundwork for the stabilization of custom
attribute invocations and general procedural macros. It applies a number of
changes discussed on [internals] as well as a [recent issue][issue], namely:
* The path used to specify a custom attribute must be of length one and cannot
be a global path. This'll help future-proof us against any ambiguities and
give us more time to settle the precise syntax. In the meantime though a bare
identifier can be used and imported to invoke a custom attribute macro. A new
feature gate, `proc_macro_path_invoc`, was added to gate multi-segment paths
and absolute paths.
* The set of items which can be annotated by a custom procedural attribute has
been restricted. Statements, expressions, and modules are disallowed behind
two new feature gates: `proc_macro_expr` and `proc_macro_mod`.
* The input to procedural macro attributes has been restricted and adjusted.
Today an invocation like `#[foo(bar)]` will receive `(bar)` as the input token
stream, but after this PR it will only receive `bar` (the delimiters were
removed). Invocations like `#[foo]` are still allowed and will be invoked in
the same way as `#[foo()]`. This is a **breaking change** for all nightly
users as the syntax coming in to procedural macros will be tweaked slightly.
* Procedural macros (`foo!()` style) can only be expanded to item-like items by
default. A separate feature gate, `proc_macro_non_items`, is required to
expand to items like expressions, statements, etc.
Closes#50038
[internals]: https://internals.rust-lang.org/t/help-stabilize-a-subset-of-macros-2-0/7252
[issue]: https://github.com/rust-lang/rust/issues/50038
Note that this commit, since it is trying to be minimal in order to
ease backporting to the beta and release channels, does *not* include
the old future-proofing warnings that we used to have associated with
such fallback to `()`; see discussion at this comment:
https://github.com/rust-lang/rust/issues/49691#issuecomment-381266730
This commit is just covering the feature gate itself and the tests
that made direct use of `!` and thus need to opt back into the
feature.
A follow on commit brings back the other change that motivates the
revert: Namely, going back to the old rules for falling back to `()`.
Add src/test/ui regression testing for NLL
This PR changes `x.py test` so that when you are running the `ui` test suite, it will also always run `compiletest` in the new `--compare-mode=nll`, which just double-checks that when running under the experimental NLL mode, the output matches the `<source-name>.nll.stderr` file, if present.
In order to reduce the chance of a developer revolt in response to this change, this PR also includes some changes to make the `--compare-mode=nll` more user-friendly:
1. It now generates nll-specific .stamp files, and uses them (so that repeated runs can reuse previously cached results).
2. Each line of terminal output distinguishes whether we are running under `--compare-mode=nll` by printing with the prefix `[ui (nll)]` instead of just the prefix `[ui]`.
Subtask of rust-lang/rust#48879
This allows easy revision of the update-references.sh script (included
here) so that it can update the expected output for nll rather than
stderr. It also reminds the rustc developer via the filename that they
are looking at output generated under comapre-mode=nll.
One could argue that there is still a problem with the strategy encoded here:
if we reach a scenario where a change to the compiler brings the output
under AST and NLL modes back into sync, this code will continue to still
generate output to distinct `foo.stderr` and `foo.nll.stderr` files, and
will continue to copy those two files back to corresponding distinct
files in the source tree, even if the *content* of the two files is now the
same.
* Arguably the "right thing" to do in that case is to remove the
`foo.nll.stderr` file entirely.
* However, I think the real answer is that we will probably want to
double-check such cases by hand anyway. We should be regularly
double-checking the diffs between `foo.stderr` and
`foo.nll.stderr`, and if we see a zero-diff case, then we should
evaluate whether that is correct, and if so, remove the file by
hand.)
* In any case, I think the default behavior encoded here (or at
least *intended* to be encoded here) is superior to the
alternative of *only* generating a `foo.nll.stderr` file if one
already existed in the source tree at the time that `compiletest`
was invoked (and otherwise unconditionally generating a
`foo.stderr` file, as was the behavior prior to this commit),
because that alternative is more likely to cause rustc developers
to overwrite a `foo.stderr` file with the stderr output from a
compare-mode=nll run, which will then break the *normal*
`compiletest` run and probably be much more confusing for the
average rustc developer.
This commit only applies the flag to the one test case,
ui/span/dropck_vec_cycle_checked.rs, that absolutely needs it. Without
the flag, that test takes an unknown amount of time (greater than 1
minute) to compile. But its possible that other tests would also
benefit from the flag, and we may want to make it the default (after
evaluating its impact on other tests).
In terms of its known impact on other tests, I have only evaluated the
ui tests, and the *only* ui test I have found that the flag impacts
(running under NLL mode, of course), is src/test/ui/nll/issue-31567.rs
In particular:
```
% ./build/x86_64-unknown-linux-gnu/stage1/bin/rustc ../src/test/ui/nll/issue-31567.rs
error[E0597]: `*v.0` does not live long enough
--> ../src/test/ui/nll/issue-31567.rs:22:26
|
22 | let s_inner: &'a S = &*v.0; //~ ERROR `*v.0` does not live long enough
| ^^^^^ borrowed value does not live long enough
23 | &s_inner.0
24 | }
| - borrowed value only lives until here
|
note: borrowed value must be valid for the lifetime 'a as defined on the function body at 21:1...
--> ../src/test/ui/nll/issue-31567.rs:21:1
|
21 | fn get_dangling<'a>(v: VecWrapper<'a>) -> &'a u32 {
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
error: aborting due to previous error
For more information about this error, try `rustc --explain E0597`.
% ./build/x86_64-unknown-linux-gnu/stage1/bin/rustc ../src/test/ui/nll/issue-31567.rs -Z nll-subminimal-causes
error[E0597]: `*v.0` does not live long enough
--> ../src/test/ui/nll/issue-31567.rs:22:26
|
22 | let s_inner: &'a S = &*v.0; //~ ERROR `*v.0` does not live long enough
| ^^^^^ borrowed value does not live long enough
23 | &s_inner.0
24 | }
| -
| |
| borrowed value only lives until here
| borrow later used here, when `v` is dropped
error: aborting due to previous error
For more information about this error, try `rustc --explain E0597`.
%
```
prep work for using timely dataflow with NLL
Two major changes:
**Two-phase borrows are overhauled.** We no longer have two bits per borrow. Instead, we track -- for each borrow -- an (optional) "activation point". Then, for each point P where the borrow is in scope, we check where P falls relative to the activation point. If P is between the reservation point and the activation point, then this is the "reservation" phase of the borrow, else the borrow is considered active. This is simpler and means that the dataflow doesn't have to care about 2-phase at all, at last not yet.
**We no longer support using the MIR borrow checker without NLL.** It is going to be increasingly untenable to support lexical mode as we go forward, I think, and also of increasingly little value. This also exposed a few bugs in NLL mode due to increased testing.
r? @pnkfelix
cc @bobtwinkles
Implement Chalk lowering rule Normalize-From-Impl
This extends the Chalk lowering pass with the "Normalize-From-Impl" rule for generating program clauses from a trait definition as part of #49177.
r? @nikomatsakis
Stabilize x86/x86_64 SIMD
This commit stabilizes the SIMD in Rust for the x86/x86_64 platforms. Notably
this commit is stabilizing:
* The `std::arch::{x86, x86_64}` modules and the intrinsics contained inside.
* The `is_x86_feature_detected!` macro in the standard library
* The `#[target_feature(enable = "...")]` attribute
* The `#[cfg(target_feature = "...")]` matcher
Stabilization of the module and intrinsics were primarily done in
rust-lang-nursery/stdsimd#414 and the two attribute stabilizations are done in
this commit. The standard library is also tweaked a bit with the new way that
stdsimd is integrated.
Note that other architectures like `std::arch::arm` are not stabilized as part
of this commit, they will likely stabilize in the future after they've been
implemented and fleshed out. Similarly the `std::simd` module is also not being
stabilized in this commit, only `std::arch`. Finally, nothing related to `__m64`
is stabilized in this commit either (MMX), only SSE and up types and intrinsics
are stabilized.
Closes#29717Closes#44839Closes#48556
This commit stabilizes the SIMD in Rust for the x86/x86_64 platforms. Notably
this commit is stabilizing:
* The `std::arch::{x86, x86_64}` modules and the intrinsics contained inside.
* The `is_x86_feature_detected!` macro in the standard library
* The `#[target_feature(enable = "...")]` attribute
* The `#[cfg(target_feature = "...")]` matcher
Stabilization of the module and intrinsics were primarily done in
rust-lang-nursery/stdsimd#414 and the two attribute stabilizations are done in
this commit. The standard library is also tweaked a bit with the new way that
stdsimd is integrated.
Note that other architectures like `std::arch::arm` are not stabilized as part
of this commit, they will likely stabilize in the future after they've been
implemented and fleshed out. Similarly the `std::simd` module is also not being
stabilized in this commit, only `std::arch`. Finally, nothing related to `__m64`
is stabilized in this commit either (MMX), only SSE and up types and intrinsics
are stabilized.
Closes#29717Closes#44839Closes#48556
Update `?` repetition disambiguation.
**Do not merge** (yet)
This is a test implementation of some ideas from discussion in https://github.com/rust-lang/rust/issues/48075 . This PR
- disallows `?` repetition from taking a separator, since the separator is never used.
- disallows the use of `?` as a separator. This allows patterns like `$(a)?+` to match `+` and `a+` rather than `a?a?a`. This is a _breaking change_, but maybe that's ok? Perhaps a crater run is the right approach?
cc @durka @alexreg @nikomatsakis
macros: Remove matching on "complex" nonterminals requiring AST comparisons
So, you can actually use nonterminals from outer macros in left hand side of nested macros and invocations of nested macros will try to match passed arguments to them.
```rust
macro outer($nt_item: item) {
macro inner($nt_item) {
struct S;
}
inner!($nt_item); // OK, `$nt_item` matches `$nt_item`
}
```
Why this is bad:
- We can't do this matching correctly. When two nonterminals are compared, the original tokens are lost and we have to compare AST fragments instead. Right now the comparison is done by `PartialEq` impls derived on AST structures.
- On one hand, AST loses information compared to original tokens (e.g. trailing separators and other simplifications done during parsing to AST), so we can produce matches that are not actually correct.
- On another hand derived `PartialEq` impls for AST structures don't make much sense in general and compare various auxiliary garbage like spans. For the argument nonterminal to match we should use literally the same token (possibly cloned) as was used in the macro LHS (as in the example above). So we can reject matches that are actually correct.
- Support for nonterminal matching is the only thing that forces us to derive `PartialEq` for all (!) AST structures. As I mentioned these impls are also mostly nonsensical.
This PR removes support for matching on all nonterminals except for "simple" ones like `ident`, `lifetime` and `tt` for which we have original tokens that can be compared.
After this is done I'll submit another PR removing huge number of `PartialEq` impls from AST and HIR structures.
This is an arcane feature and I don't personally know why would anyone use it, but the change should ideally go through crater.
We'll be able to support this feature again in the future when all nonterminals have original token streams attached to them in addition to (or instead of) AST fragments.
Hygiene 2.0: Avoid comparing fields by name
There are two separate commits here (not counting tests):
- The first one unifies named (`obj.name`) and numeric (`obj.0`) field access expressions in AST and HIR. Before field references in these expressions are resolved it doesn't matter whether the field is named or numeric (it's just a symbol) and 99% of code is common. After field references are resolved we work with
them by index for all fields (see the second commit), so it's again not important whether the field was named or numeric (this includes MIR where all fields were already by index).
(This refactoring actually fixed some bugs in HIR-based borrow checker where borrows through names (`S {
0: ref x }`) and indices (`&s.0`) weren't considered overlapping.)
- The second commit removes all by-name field comparison and instead resolves field references to their indices once, and then uses those resolutions. (There are still a few name comparisons in save-analysis, because save-analysis is weird, but they are made correctly hygienic).
Thus we are fixing a bunch of "secondary" field hygiene bugs (in borrow checker, lints).
Fixes https://github.com/rust-lang/rust/issues/46314
