• R/O
  • HTTP
  • SSH
  • HTTPS

List of commits

Tags
No Tags

Frequently used words (click to add to your profile)

javac++androidlinuxc#windowsobjective-ccocoa誰得qtpythonphprubygameguibathyscaphec計画中(planning stage)翻訳omegatframeworktwitterdomtestvb.netdirectxゲームエンジンbtronarduinopreviewer

rayonパッケージ


RSS
Rev. Time Author
b216c3e 2019-08-16 07:09:45 Josh Stone

Update bare traits to `dyn Trait`

This syntax is available since Rust 1.27.

58b0275 2019-08-16 07:08:50 Josh Stone

Once:new() is const-stable since Rust 1.24

... so stop using `ONCE_INIT` which is deprecated in 1.37.

83b67e2 2019-08-16 01:02:24 bors[bot]

Merge #673

673: Reduce the genericity of many closures r=nikomatsakis a=cuviper

The `take_while` closure only needs to be generic in `T`, along with a
captured `&AtomicBool`. When we write the closure directly, it carries
all the type baggage of `WhileSomeFolder<'f, C>::consumer_iter<I>`,
where the monomorphized type may explode. Instead, we can move this
closure to a standalone function for reduced genericity.

Co-authored-by: Josh Stone <cuviper@gmail.com>

dbc114f 2019-08-15 05:32:08 Josh Stone

Avoid closures in the Copied implementation

97cc314 2019-08-15 05:29:27 Josh Stone

Reduce genericity in more iterator closures

ba41dab 2019-08-15 05:29:27 Josh Stone

Reduce genericity in join wrappers

249fbb9 2019-08-15 05:27:54 Josh Stone

Reduce the genericity of extend closures

Most of these closures only need to be generic in the item type, but
when created in the context of some `I` generic iterator, they inherit
that genericity. This affects both type length and code duplication.

d2c1424 2019-08-15 05:27:54 Josh Stone

Reduce genericity in FromParallelIterator for Result

224287b 2019-08-15 05:27:54 Josh Stone

Reduce the genericity of WhileSomeFolder::consume_iter's closure

The `take_while` closure only needs to be generic in `T`, along with a
captured `&AtomicBool`. When we write the closure directly, it carries
all the type baggage of `WhileSomeFolder<'f, C>::consumer_iter<I>`,
where the monomorphized type may explode. Instead, we can move this
closure to a standalone function for reduced genericity.

8d2e048 2019-08-09 01:37:30 bors[bot]

Merge #675

675: Update crossbeam-deque to 0.7 r=cuviper a=Aaron1011

The only substantive change is to 'take_local_job'.
'Worker::pop` now returns a plain 'Option<T>', so there's
no need to loop anymore

Co-authored-by: Aaron Hill <aa1ronham@gmail.com>
Co-authored-by: Josh Stone <cuviper@gmail.com>

99706a4 2019-08-09 01:32:35 Josh Stone

Bump minor versions for the increase to Rust 1.28

022006d 2019-08-09 01:28:19 Josh Stone

use Worker directly to create deques

678ffbb 2019-07-23 02:40:14 bors[bot]

Merge #668

668: Reusing LockLatch to reduce allocation r=cuviper a=ralfbiedert

This addresses the first half of #666.

I kept the `LockLatch` and added a `reset()` method because that had overall fewer lines of code. If you want I can still change it to a separate `ReusableLockLatch` type.


Co-authored-by: Ralf Biedert <rb@xr.io>
Co-authored-by: Josh Stone <cuviper@gmail.com>

9619a16 2019-07-23 02:01:49 Josh Stone

in_worker_cross doesn't need an indirect latch

20c526e 2019-07-19 17:43:09 Ralf Biedert

Removed RefCell, refs, moved thread_local, applied cargo fmt.

45a1c33 2019-07-19 06:06:12 bors[bot]

Merge #674

674: Make test 'iter_panic_fuse' deterministic r=cuviper a=Aaron1011

Fixes #667

'panic_fuse' will only stop other threads 'as soon as possible' -
if the other threads are fast enough, they might end up processing
the entire rest of the iterator.

This commit changes the test 'iter_panic_fuse' to properly take this
into account, by creating a custom threadpool with only 1 thread.
This makes the test deterministic - with only one thread, the panic
is guaranmteed to be observed when the next item is processed, causing
the desired early exit.

Co-authored-by: Aaron Hill <aa1ronham@gmail.com>
Co-authored-by: Josh Stone <cuviper@gmail.com>

2a2c5a8 2019-07-19 06:03:31 Josh Stone

Use a single thread to test cmp short circuit

67a9aa2 2019-07-19 05:58:14 Josh Stone

Split the pool in iter_panic_fuse for formatting's sake

d9bc1fc 2019-07-13 20:44:35 Ralf Biedert

We don't need these.

5045a24 2019-07-13 20:41:48 Ralf Biedert

Actually impl `LatchProbe` and `Latch` for &L where L: Latch.

073b79b 2019-07-13 20:29:42 Ralf Biedert

Impl `LatchProbe` and `Latch` for &L where L: Latch.

42d1d4f 2019-07-13 08:28:33 Aaron Hill

Revert 'more-asserts' dependency addition

2b5227c 2019-07-13 03:12:18 bors[bot]

Merge #665

665: Add ParallelIterator::copied() r=nikomatsakis a=cuviper



Co-authored-by: Josh Stone <cuviper@gmail.com>

e11aad9 2019-07-08 03:47:34 Aaron Hill

Update crossbeam-deque to 0.7

The only substantive change is to 'take_local_job'.
'Worker::pop` now returns a plain 'Option<T>', so there's
no need to loop anymore

9532eb5 2019-07-08 03:36:27 Aaron Hill

Make test 'iter_panic_fuse' deterministic

Fixes #667

'panic_fuse' will only stop other threads 'as soon as possible' -
if the other threads are fast enough, they might end up processing
the entire rest of the iterator.

This commit changes the test 'iter_panic_fuse' to properly take this
into account, by creating a custom threadpool with only 1 thread.
This makes the test deterministic - with only one thread, the panic
is guaranmteed to be observed when the next item is processed, causing
the desired early exit.

I've also added the 'more-asserts' crate as a dev dependency, so
that we can print out a more informative error message in
'iter_panic_fuse'

228dbf7 2019-07-02 18:17:30 Ralf Biedert

Added lifetime bound.

bfc072f 2019-07-02 16:00:52 Ralf Biedert

Added TLS LockLatch with a reset() method.

55f235f 2019-07-02 15:28:48 Ralf Biedert

Make StackJob take a &mut latch.

481939b 2019-07-02 15:26:19 Ralf Biedert

Ignore CLion files.

0d17931 2019-06-26 00:50:54 Josh Stone

Add ParallelIterator::copied()