Optimize bidi character detection.
Should fix most of the performance regression of the bidi character detection (#90514), to be confirmed with a perf run.
Initialize LLVM time trace profiler on each code generation thread
In https://reviews.llvm.org/D71059 LLVM 11, the time trace profiler was
extended to support multiple threads.
`timeTraceProfilerInitialize` creates a thread local profiler instance.
When a thread finishes `timeTraceProfilerFinishThread` moves a thread
local instance into a global collection of instances. Finally when all
codegen work is complete `timeTraceProfilerWrite` writes data from the
current thread local instance and the instances in global collection
of instances.
Previously, the profiler was intialized on a single thread only. Since
this thread performs no code generation on its own, the resulting
profile was empty.
Update LLVM codegen to initialize & finish time trace profiler on each
code generation thread.
cc `@tmandry`
r? `@wesleywiser`
Append .0 to unsuffixed float if it would otherwise become int token
Previously the unsuffixed f32/f64 constructors of `proc_macro::Literal` would create literal tokens that are definitely not a float:
```rust
Literal::f32_unsuffixed(10.0) // 10
Literal::f32_suffixed(10.0) // 10f32
Literal::f64_unsuffixed(10.0) // 10
Literal::f64_suffixed(10.0) // 10f64
```
Notice that the `10` are actually integer tokens if you were to reparse them, not float tokens.
This diff updates `Literal::f32_unsuffixed` and `Literal::f64_unsuffixed` to produce tokens that unambiguously parse as a float. This matches longstanding behavior of the proc-macro2 crate's implementation of these APIs dating back at least 3.5 years, so it's likely an unobjectionable behavior.
```rust
Literal::f32_unsuffixed(10.0) // 10.0
Literal::f32_suffixed(10.0) // 10f32
Literal::f64_unsuffixed(10.0) // 10.0
Literal::f64_suffixed(10.0) // 10f64
```
Fixes https://github.com/dtolnay/syn/issues/1085.
Implementation of GATs outlives lint
See #87479 for background. Closes#87479
The basic premise of this lint/error is to require the user to write where clauses on a GAT when those bounds can be implied or proven from any function on the trait returning that GAT.
## Intuitive Explanation (Attempt) ##
Let's take this trait definition as an example:
```rust
trait Iterable {
type Item<'x>;
fn iter<'a>(&'a self) -> Self::Item<'a>;
}
```
Let's focus on the `iter` function. The first thing to realize is that we know that `Self: 'a` because of `&'a self`. If an impl wants `Self::Item` to contain any data with references, then those references must be derived from `&'a self`. Thus, they must live only as long as `'a`. Furthermore, because of the `Self: 'a` implied bound, they must live only as long as `Self`. Since it's `'a` is used in place of `'x`, it is reasonable to assume that any value of `Self::Item<'x>`, and thus `'x`, will only be able to live as long as `Self`. Therefore, we require this bound on `Item` in the trait.
As another example:
```rust
trait Deserializer<T> {
type Out<'x>;
fn deserialize<'a>(&self, input: &'a T) -> Self::Out<'a>;
}
```
The intuition is similar here, except rather than a `Self: 'a` implied bound, we have a `T: 'a` implied bound. Thus, the data on `Self::Out<'a>` is derived from `&'a T`, and thus it is reasonable to expect that the lifetime `'x` will always be less than `T`.
## Implementation Algorithm ##
* Given a GAT `<P0 as Trait<P1..Pi>>::G<Pi...Pn>` declared as `trait T<A1..Ai> for A0 { type G<Ai...An>; }` used in return type of one associated function `F`
* Given env `E` (including implied bounds) for `F`
* For each lifetime parameter `'a` in `P0...Pn`:
* For each other type parameter `Pi != 'a` in `P0...Pn`: // FIXME: this include of lifetime parameters too
* If `E => (P: 'a)`:
* Require where clause `Ai: 'a`
## Follow-up questions ##
* What should we do when we don't pass params exactly?
For this example:
```rust
trait Des {
type Out<'x, D>;
fn des<'z, T>(&self, data: &'z Wrap<T>) -> Self::Out<'z, Wrap<T>>;
}
```
Should we be requiring a `D: 'x` clause? We pass `Wrap<T>` as `D` and `'z` as `'x`, and should be able to prove that `Wrap<T>: 'z`.
r? `@nikomatsakis`
Rollup of 5 pull requests
Successful merges:
- #89942 (Reorder `widening_impl`s to make the doc clearer)
- #90569 (Fix tests using `only-i686` to use the correct `only-x86` directive)
- #90597 (Warn for variables that are no longer captured)
- #90623 (Remove more checks for LLVM < 12)
- #90626 (Properly register text_direction_codepoint_in_comment lint.)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
This reverts commit 12fbabd27f700a59d0e7031f0839b220c3514bcb.
It was only needed because of using raw `clone3` instead of `fork`, but
we only do that now when a pidfd is requested.
In #89522 we learned that `clone3` is interacting poorly with Gentoo's
`sandbox` tool. We only need that for the unstable pidfd extensions, so
otherwise avoid that and use a normal `fork`.
`Candidate` enum has only a single `Ref` variant. Refactor it into a
struct and reduce overall indentation of the code by two levels.
No functional changes.
Properly register text_direction_codepoint_in_comment lint.
This makes it known to the compiler so it can be configured like with `#![allow(text_direction_codepoint_in_comment)]`.
Fixes#90614.
Fix ICE when rustdoc is scraping examples inside of a proc macro
This PR provides a clearer semantics for how --scrape-examples interacts with macros. If an expression's span AND it's enclosing item's span both are not `from_expansion`, then the example will be scraped. The added test case `rustdoc-scrape-examples-macros` shows a variety of situations.
* A macro-rules macro that takes a function call as input: good
* A macro-rules macro that generates a function call as output: bad
* A proc-macro that generates a function call as output: bad
* An attribute macro that generates a function call as output: bad
* An attribute macro that takes a function call as input: good, if the proc macro is designed to propagate the input spans
I ran this updated rustdoc on pyo3 and confirmed that it successfully scrapes examples from inside a proc macro, eg
<img width="1013" alt="Screen Shot 2021-11-04 at 1 11 28 PM" src="https://user-images.githubusercontent.com/663326/140412691-81a3bb6b-a448-4a1b-a293-f7a795553634.png">
(cc `@mejrs)`
Additionally, this PR fixes an ordering bug in the highlighting logic.
Fixes https://github.com/rust-lang/rust/issues/90567.
r? `@jyn514`
In https://reviews.llvm.org/D71059 LLVM 11, the time trace profiler was
extended to support multiple threads.
`timeTraceProfilerInitialize` creates a thread local profiler instance.
When a thread finishes `timeTraceProfilerFinishThread` moves a thread
local instance into a global collection of instances. Finally when all
codegen work is complete `timeTraceProfilerWrite` writes data from the
current thread local instance and the instances in global collection
of instances.
Previously, the profiler was intialized on a single thread only. Since
this thread performs no code generation on its own, the resulting
profile was empty.
Update LLVM codegen to initialize & finish time trace profiler on each
code generation thread.
Re-export some iterators from `core` in `std`
These iterators seem to have been forgotten to be re-exported from `std` (through `alloc`)
These are stable:
`core::slice::{SplitInclusive, SplitInclusiveMut}`
This one is still unstable:
`core::slice::EscapeAscii` (cc #77174)
