Add benchmark and fast path for BufReader::read_exact
At work, we have a wrapper type that implements this optimization. It would be nice if the standard library were faster.
Before:
```
test io::buffered::tests::bench_buffered_reader_small_reads ... bench: 7,670 ns/iter (+/- 45)
```
After:
```
test io::buffered::tests::bench_buffered_reader_small_reads ... bench: 4,457 ns/iter (+/- 41)
```
resolve: Reject ambiguity built-in attr vs different built-in attr
Fixes https://github.com/rust-lang/rust/issues/79798.
Resolution ensures that inert attributes cannot be used through imports like this, but built-in attributes don't go through initial resolution (only through resolution validation), so we have to keep some extra data (the built-in attribute name) to prevent it from happening.
correctly deal with late-bound lifetimes in anon consts
adds support for using late bound lifetimes of the parent context in anon consts.
```rust
#![feature(const_generics)]
const fn inner<'a>() -> usize where &'a (): Sized { 3 }
fn test<'a>() {
let _: [u8; inner::<'a>()];
}
```
The lifetime `'a` is late bound in `test` so it's not included in its generics but is instead dealt with separately in borrowck.
This didn't previously work for anon consts as they have to use the late bound lifetimes of their parent which has
to be explicitly handled.
r? ```@matthewjasper``` cc ```@varkor``` ```@eddyb```
BTreeMap: expose new_internal function and sanitize from_new_internal
`new_internal` is the functional core of the imperative `push_internal_level`, and `from_new_internal` can easily do a proper job instead of returning a half-baked node.
r? `@Mark-Simulacrum`
Add `as_rchunks` (and friends) to slices
`@est31` mentioned (https://github.com/rust-lang/rust/issues/76354#issuecomment-717027175) that, for completeness, there needed to be an `as_chunks`-like method that chunks from the end (with the remainder at the beginning) like `rchunks` does.
So here's a PR for `as_rchunks: &[T] -> (&[T], &[[T; N]])` and `as_rchunks_mut: &mut [T] -> (&mut [T], &mut [[T; N]])`.
But as I was doing this and copy-pasting `from_raw_parts` calls, I thought that I should extract that into an unsafe method. It started out a private helper, but it seemed like `as_chunks_unchecked` could be reasonable as a "real" method, so I added docs and made it public. Let me know if you think it doesn't pull its weight.
Move some tests to more reasonable directories - 2
All tests with a score equal or greater than 1.0 were moved to their respective directories by issuing
```bash
cat FILE | tr -s " " | tr -d '():' | sort -k3 | awk '$3 >= 1' | cut -d " " -f1-2 | sed 's;\\;/;g' | xargs -n2 git mv
```
**Observation**: The first column values is the only column with results greater zero
To attest the confidentiality of the model, some manual revision of at least of tests is needed and this process will be tracked in the following list:
* `src/test/ui/abi/issue-28676.rs` OK #28676
* `src/test/ui/array-slice-vec/issue-15730.rs` OK
* `src/test/ui/associated-types/issue-24338.rs` OK #54823
* `src/test/ui/associated-types/issue-48551.rs` Looks OK #48551
* `src/test/ui/associated-types/issue-50301.rs` Looks OK #63577
...
cc #73494
r? `@petrochenkov`
This fixed things the last time I had a problem like this. And plausibly will here too -- the check it's failing on is for the high bit being set in the length of the slice, which is a check that's only in a debug_assert.
Rollup of 17 pull requests
Successful merges:
- #78455 (Introduce {Ref, RefMut}::try_map for optional projections in RefCell)
- #80144 (Remove giant badge in README)
- #80614 (Explain why borrows can't be held across yield point in async blocks)
- #80670 (TrustedRandomAaccess specialization composes incorrectly for nested iter::Zips)
- #80681 (Clarify what the effects of a 'logic error' are)
- #80764 (Re-stabilize Weak::as_ptr and friends for unsized T)
- #80901 (Make `x.py --color always` apply to logging too)
- #80902 (Add a regression test for #76281)
- #80941 (Do not suggest invalid code in pattern with loop)
- #80968 (Stabilize the poll_map feature)
- #80971 (Put all feature gate tests under `feature-gates/`)
- #81021 (Remove doctree::Import)
- #81040 (doctest: Reset errors before dropping the parse session)
- #81060 (Add a regression test for #50041)
- #81065 (codegen_cranelift: Fix redundant semicolon warn)
- #81069 (Add sample code for Rc::new_cyclic)
- #81081 (Add test for #34792)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
doctest: Reset errors before dropping the parse session
The first parse is to collect whether the code contains macros, has
`main`, and uses other crates. In that pass we ignore errors as those
will be reported when the test file is actually built.
For that we need to reset errors in the `Diagnostic` otherwise when
dropping it unhandled errors will be reported as compiler bugs.
Fixes#80992
Put all feature gate tests under `feature-gates/`
There was one directory that had only a single test and there was also a
test in the top-level directory. This moves both of them to
`feature-gates/`.
Add a regression test for #76281
This has been fixed between 1.47.0-nightly (663d2f5cd 2020-08-22) and 1.47.0-nightly (5180f3da5 2020-08-23). Maybe fixed by #73526?
Created `wasm` dir, it currently has only one test but I'll move some wasm-related tests there as a follow-up.
Closes#76281
Re-stabilize Weak::as_ptr and friends for unsized T
As per [T-lang consensus](https://hackmd.io/7r3_is6uTz-163fsOV8Vfg), this uses a branch to handle the dangling case. The discussed optimization of only doing the branch in the T: ?Sized case is left for a followup patch, as doing so is not trivial (as it requires specialization) and not _obviously_ better (as it requires using `wrapping_offset` rather than `offset` more).
<details><summary>Basically said optimization</summary>
Specialize on `T: Sized`:
```rust
fn as_ptr(&self) -> *const T {
if [ T is Sized ] || !is_dangling(ptr) {
(ptr as *mut T).set_ptr_value( (ptr as *mut u8).wrapping_offset(data_offset) )
} else {
ptr::null()
}
}
fn from_raw(*const T) -> Self {
if [ T is Sized ] || !ptr.is_null() {
let ptr = (ptr as *mut RcBox).set_ptr_value( (ptr as *mut u8).wrapping_offset(-data_offset) );
Weak { ptr }
} else {
Weak::new()
}
}
```
(but with more `set_ptr_value` to avoid `Sized` restrictions and maintain metadata.)
Written in this fashion, this is not a correctness-critical specialization (i.e. so long as `[ T is Sized ]` is false for unsized `T`, it can be `rand()` for sized `T` without breaking correctness), but it's still touchy, so I'd rather do it in another PR with separate review.
---
</details>
This effectively reverts #80422 and re-establishes #74160. T-libs [previously signed off](https://github.com/rust-lang/rust/pull/74160#issuecomment-660539373) on this stable API change in #74160.
Clarify what the effects of a 'logic error' are
This clarifies what a 'logic error' is (which is a term used to describe what happens if you put things in a hash table or btree and then use something like a refcell to break the internal ordering). This tries to be as vague as possible, as we don't really want to promise what happens, except "bad things, but not UB". This was discussed in #80657
TrustedRandomAaccess specialization composes incorrectly for nested iter::Zips
I found this while working on improvements for TRA.
After partially consuming a Zip adapter and then wrapping it into another Zip where the adapters use their `TrustedRandomAccess` specializations leads to the outer adapter returning elements which should have already been consumed.
If the optimizer gets tripped up by the addition this might affect performance for chained `zip()` iterators even when the inner one is not partially advanced but it would require more extensive fixes to `TrustedRandomAccess` to communicate those offsets earlier.
Included test fails on nightly, [playground link](https://play.rust-lang.org/?version=nightly&mode=debug&edition=2018&gist=24fa1edf8a104ff31f5a24830593b01f)
Introduce {Ref, RefMut}::try_map for optional projections in RefCell
This fills a usability gap of `RefCell` I've personally encountered to perform optional projections, mostly into collections such as `RefCell<Vec<T>>` or `RefCell<HashMap<U, T>>`:
> This kind of API was briefly featured under Open questions in #10514 back in 2013 (!)
```rust
let values = RefCell::new(vec![1, 2, 3, 4]);
let b = Ref::opt_map(values.borrow(), |vec| vec.get(2));
```
It primarily avoids this alternative approach to accomplish the same kind of projection which is both rather noisy and panicky:
```rust
let values = RefCell::new(vec![1, 2, 3, 4]);
let b = if values.get(2).is_some() {
Some(Ref::map(values.borrow(), |vec| vec.get(2).unwrap()))
} else {
None
};
```
### Open questions
The naming `opt_map` is preliminary. I'm not aware of prior art in std to lean on here, but this name should probably be improved if this functionality is desirable.
Since `opt_map` consumes the guard, and alternative syntax might be more appropriate which instead *tries* to perform the projection, allowing the original borrow to be recovered in case it fails:
```rust
pub fn try_map<U: ?Sized, F>(orig: Ref<'b, T>, f: F) -> Result<Ref<'b, U>, Self>
where
F: FnOnce(&T) -> Option<&U>;
```
This would be more in line with the `try_map` method [provided by parking lot](https://docs.rs/lock_api/0/lock_api/struct.RwLockWriteGuard.html#method.try_map).
implement ptr::write without dedicated intrinsic
This makes `ptr::write` more consistent with `ptr::write_unaligned`, `ptr::read`, `ptr::read_unaligned`, all of which are implemented in terms of `copy_nonoverlapping`.
This means we can also remove `move_val_init` implementations in codegen and Miri, and its special handling in the borrow checker.
Also see [this Zulip discussion](https://rust-lang.zulipchat.com/#narrow/stream/131828-t-compiler/topic/ptr.3A.3Aread.20vs.20ptr.3A.3Awrite).
The first parse is to collect whether the code contains macros, has
`main`, and uses other crates. In that pass we ignore errors as those
will be reported when the test file is actually built.
For that we need to reset errors in the `Diagnostic` otherwise when
dropping it unhandled errors will be reported as compiler bugs.
Fixes#80992
