OpenE2K/rust
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
126,628 Commits 1 Branch 0 Tags 636 MiB
cff5f56886
Commit Graph

206 Commits

Author SHA1 Message Date
Ralf Jung
cff5f56886 rename MaybeUninit slice methods 
first_ptr -> slice_as_ptr
first_ptr_mut -> slice_as_mut_ptr
slice_get_ref -> slice_assume_init_ref
slice_get_mut -> slice_assume_init_mut
 2020-09-05 17:24:22 +02:00
bors
70c5f6efc4 Auto merge of #75200 - ssomers:btree_valmut, r=Mark-Simulacrum 
BTreeMap: introduce marker::ValMut and reserve Mut for unique access

The mutable BTreeMap iterators (apart from `DrainFilter`) are double-ended, meaning they have to rely on a front and a back handle that each represent a reference into the tree. Reserve a type category `marker::ValMut` for them, so that we guarantee that they cannot reach operations on handles with borrow type `marker::Mut`and that these operations can assume unique access to the tree.

Including #75195, benchmarks report no genuine change:
```
benchcmp old new --threshold 5
 name                                 old ns/iter  new ns/iter  diff ns/iter   diff %  speedup
 btree::map::iter_100                 3,333        3,023                -310   -9.30%   x 1.10
 btree::map::range_unbounded_vs_iter  36,624       31,569             -5,055  -13.80%   x 1.16
```

r? @Mark-Simulacrum
 2020-09-04 23:16:23 +00:00
bors
ef55a0a92f Auto merge of #75207 - dylni:add-slice-check-range, r=KodrAus 
Add `slice::check_range`

This method is useful for [`RangeBounds`] parameters. It's even been [rewritten](22ee68dc58/src/librustc_data_structures/sorted_map.rs (L214)) [many](22ee68dc58/library/alloc/src/vec.rs (L1299)) [times](22ee68dc58/library/core/src/slice/mod.rs (L2441)) in the standard library, sometimes assuming that the bounds won't be [`usize::MAX`].

For example, [`Vec::drain`] creates an empty iterator when [`usize::MAX`] is used as an inclusive end bound:

```rust
assert!(vec![1].drain(..=usize::max_value()).eq(iter::empty()));
```

If this PR is merged, I'll create another to use it for those methods.

[`RangeBounds`]: https://doc.rust-lang.org/std/ops/trait.RangeBounds.html
[`usize::MAX`]: https://doc.rust-lang.org/std/primitive.usize.html#associatedconstant.MAX
[`Vec::drain`]: https://doc.rust-lang.org/std/vec/struct.Vec.html#method.drain
 2020-09-04 12:21:43 +00:00
The8472
2f23a0fcca fix debug assertion 
The InPlaceIterable debug assert checks that the write pointer
did not advance beyond the read pointer. But TrustedRandomAccess
never advances the read pointer, thus triggering the assert.
Skip the assert if the source pointer did not change during iteration.
 2020-09-03 22:15:47 +02:00
The8472
8e5fe5569b improve comments and naming 2020-09-03 20:59:37 +02:00
The8472
6464586542 add explanation to specialization marker 2020-09-03 20:59:36 +02:00
The8472
acdd441cc3 remove separate no-drop code path since it resulted in more LLVM IR 2020-09-03 20:59:36 +02:00
The8472
435219dd82 remove empty Vec extend optimization 
The optimization meant that every extend code path had to emit llvm
IR for from_iter and extend spec_extend, which likely impacts
compile times while only improving a few edge-cases
 2020-09-03 20:59:35 +02:00
The8472
9aeea00222 get things to work under min_specialization by leaning more heavily on #[rustc_unsafe_specialization_marker] 2020-09-03 20:59:34 +02:00
The8472
a62cd1b44c fix benchmark compile errors 2020-09-03 20:59:33 +02:00
The8472
bec9f9223c apply required min_specialization attributes 2020-09-03 20:59:32 +02:00
The8472
80638330f2 support in-place collect for MapWhile adapters 2020-09-03 20:59:32 +02:00
The8472
5530858a08 generalize in-place collect to types of same size and alignment 2020-09-03 20:59:31 +02:00
The8472
fa34b39cd6 increase comment verbosity 2020-09-03 20:59:30 +02:00
The8472
872ab780c0 work around compiler overhead around lambdas in generics by extracting them into free functions 2020-09-03 20:59:29 +02:00
The8472
771b8ecc83 extract IntoIter drop/forget used by specialization into separate methods 2020-09-03 20:59:29 +02:00
The8472
6ad133443a add benchmark to cover in-place extend 2020-09-03 20:59:28 +02:00
The8472
a7a8b52e91 remove redundant cast 2020-09-03 20:59:28 +02:00
The8472
470bf54f94 test drops during in-place iteration 2020-09-03 20:59:27 +02:00
The8472
fe350dd82d move unsafety into method, not relevant to caller 2020-09-03 20:59:27 +02:00
The8472
0d2d033415 replace unsafe ptr::write with deref-write, benchmarks show no difference 2020-09-03 20:59:26 +02:00
The8472
9596e5a2f2 pacify tidy 2020-09-03 20:59:26 +02:00
The8472
6ed05fd995 replace drop flag with ManuallyDrop 2020-09-03 20:59:25 +02:00
The8472
ab382b7661 mark as_inner as unsafe and update comments 2020-09-03 20:59:24 +02:00
The8472
2a51e579f5 avoid exposing that binary heap's IntoIter is backed by vec::IntoIter, use a private trait instead 2020-09-03 20:59:24 +02:00
The8472
c731648e77 fix: bench didn't black_box its results 2020-09-03 20:59:23 +02:00
The8472
0856771248 fix build issue due to stabilized feature 2020-09-03 20:59:23 +02:00
The8472
e85cfa4f22 impl TrustedRandomAccess for vec::IntoIter 2020-09-03 20:59:22 +02:00
The8472
e1151844fa bench larger allocations 2020-09-03 20:59:22 +02:00
The8472
fd16202e36 include in-place .zip() in test 2020-09-03 20:59:21 +02:00
The8472
fbb3371e5b remove unecessary feature flag 
# Conflicts:
#	library/alloc/src/lib.rs
 2020-09-03 20:59:21 +02:00
The8472
085eb20a61 move free-standing method into trait impl 2020-09-03 20:59:19 +02:00
The8472
3d5e9f1904 bench in-place zip 2020-09-03 20:59:18 +02:00
The8472
2b0b2ae9f6 additional specializations tests 2020-09-03 20:59:17 +02:00
The8472
00a32eb54f fix some in-place-collect edge-cases 
- it's an allocation optimization, so don't attempt to do it on ZSTs
- drop the tail of partially exhausted iters
 2020-09-03 20:59:17 +02:00
The8472
8c816b96dd remove redundant code 2020-09-03 20:59:16 +02:00
The8472
cc67c8eb91 improve comments 2020-09-03 20:59:16 +02:00
The8472
290fe895ba specialize creating a Vec from a slice iterator where T: Copy 
this was already implemented for Extend but not for FromIterator
 2020-09-03 20:59:15 +02:00
The8472
dac0edfaaa restore SpecFrom<T, TrustedLen<Item=T>> specialization by nesting 
specializations
 2020-09-03 20:59:15 +02:00
The8472
582fbb1d62 use From specializations on extend if extended Vec is empty 
this enables in-place iteration and allocation reuse in additional cases
 2020-09-03 20:59:14 +02:00
The8472
a596ff36b5 exercise more of the in-place pipeline in the bench 2020-09-03 20:59:14 +02:00
The8472
a9c78e371e bench in-place collect of droppables 2020-09-03 20:59:13 +02:00
The8472
8ac96e6a98 cyclic in-place reuse bench 2020-09-03 20:59:13 +02:00
The8472
bb4f888a59 return the things under test so they get black_box()'ed 2020-09-03 20:59:12 +02:00
The8472
2f700d085a add benches from bluss' gists 2020-09-03 20:59:12 +02:00
The8472
a4e385a0d0 use memmove instead of generic in-place iteration for IntoIter source 
this is the original SpecExtend<_, IntoIter> logic except generalizing
the fast-path to include a memmove
 2020-09-03 20:59:11 +02:00
The8472
631543dcb4 restore Vec::extend specialization for vec::IntoIter sources that 
was lost during refactoring
 2020-09-03 20:59:11 +02:00
The8472
07a8c1b95a hide binary_heap::IntoIter internals behind impl Trait 2020-09-03 20:59:10 +02:00
The8472
232065074d recover vectorization 
switch to try_fold and segregate the drop handling to keep
collect::<Vec<u8>>() and similar optimizer-friendly

It comes at the cost of less accurate debug_asserts and code complexity
 2020-09-03 20:59:09 +02:00
The8472
6c5c47b82b update benches 2020-09-03 20:59:09 +02:00