bors
c102c5cfc6
Auto merge of #96020 - martingms:optimize-relate_substs, r=nnethercote
...
Micro-optimize `ty::relate::relate_substs` by avoiding `match`
Was a top-20 hot function in a callgrind profile of compiling `bitmaps-3.1.0`.
Yields some small speedups on that crate and some others according to local benching:
Benchmark | Profile | Scenario | % Change | Significance Factor?
-- | -- | -- | -- | --
bitmaps-3.1.0 | check | full | -1.88% | 9.42x
bitmaps-3.1.0 | debug | full | -1.80% | 8.99x
bitmaps-3.1.0 | opt | full | -1.70% | 8.49x
bitmaps-3.1.0 | check | incr-full | -1.54% | 7.68x
deep-vector | debug | full | 1.52% | 7.61x
bitmaps-3.1.0 | debug | incr-full | -1.45% | 7.26x
bitmaps-3.1.0 | opt | incr-full | -1.39% | 6.95x
nalgebra-0.30.1 | check | full | -0.68% | 3.42x
nalgebra-0.30.1 | debug | full | -0.64% | 3.22x
nalgebra-0.30.1 | opt | full | -0.64% | 3.20x
projection-caching | check | full | -0.61% | 3.05x
nalgebra-0.30.1 | check | incr-full | -0.56% | 2.78x
nalgebra-0.30.1 | opt | incr-full | -0.54% | 2.72x
nalgebra-0.30.1 | debug | incr-full | -0.54% | 2.69x
projection-caching | check | incr-full | -0.50% | 2.51x
tt-muncher | opt | full | -0.48% | 2.42x
projection-caching | opt | full | -0.47% | 2.37x
projection-caching | debug | full | -0.47% | 2.35x
projection-caching | opt | incr-full | -0.44% | 2.21x
projection-caching | debug | incr-full | -0.42% | 2.08x
deeply-nested-multi | check | incr-full | 0.37% | 1.87x
wf-projection-stress-65510 | opt | full | -0.37% | 1.84x
deep-vector | debug | incr-patched: add vec item | -0.32% | 1.61x
projection-caching | debug | incr-unchanged | -0.32% | 1.60x
wf-projection-stress-65510 | check | full | -0.31% | 1.55x
projection-caching | opt | incr-unchanged | -0.31% | 1.53x
wf-projection-stress-65510 | debug | incr-full | -0.30% | 1.51x
wf-projection-stress-65510 | opt | incr-full | -0.30% | 1.51x
r? `@nnethercote`
2022-04-19 10:53:53 +00:00
Nikita Popov
b914678a35
Ensure existance of dist directory when creating tarball
2022-04-19 10:49:37 +02:00
Michael Woerister
c0be619724
incr. comp.: Don't export impl_stable_hash_via_hash!() and warn about using it.
2022-04-19 10:43:20 +02:00
Mara Bos
06a8f05b0c
Use futex locks on emscripten.
2022-04-19 09:24:51 +02:00
Mara Bos
8f2913cc24
Use futex locks on wasm+atomics.
2022-04-19 09:21:54 +02:00
Mara Bos
65987ae8f5
Make std::sys::wasm::futex consistent with unix::futex.
2022-04-19 09:21:54 +02:00
Mara Bos
6abdd0b6d4
Make std::sys::unix::futex consistent on emscripten.
2022-04-19 09:19:29 +02:00
Nicholas Nethercote
d235ac7801
Handle Delimited opening immediately.
...
Instead of letting the next iteration of the loop handle it.
2022-04-19 17:02:49 +10:00
Nicholas Nethercote
29c78cc086
Add {open,close}_delim arguments to TokenCursorFrame::new().
...
This will facilitate the change in the next commit.
`boolean` arguments aren't great, but the function is only used in three
places within this one file.
2022-04-19 17:02:48 +10:00
Nicholas Nethercote
02317542eb
Rearrange TokenCursor::inlined_next().
...
In particular, avoid wrapping a token within `TokenTree::Token` and then
immediately matching it and returning the token within. Just return the
token immediately.
2022-04-19 17:02:48 +10:00
Nicholas Nethercote
b1e6dee596
Merge TokenCursor::{next,next_desugared}.
...
And likewise for the inlined variants.
I did this for simplicity, but interesting it was a performance win as
well.
2022-04-19 17:02:48 +10:00
Nicholas Nethercote
89ec75b0e9
Inline and remove Parser::next_tok().
...
It has a single call site.
2022-04-19 17:02:48 +10:00
Nicholas Nethercote
aefbbeec34
Inline and remove TokenTree::{open_tt,close_tt}.
...
They both have a single call site.
2022-04-19 17:02:48 +10:00
Nicholas Nethercote
ad566b78f2
Tweak Cursor::next_with_spacing.
...
This makes it more like `CursorRef::next_with_spacing`. There is no
performance effect, just a consistency improvement.
2022-04-19 17:02:48 +10:00
Chris Morgan
0255398ff7
Improve AddrParseError description
...
The existing description was incorrect for socket addresses, and
misleading: users would see “invalid IP address syntax” and suppose they
were supposed to provide an IP address rather than a socket address.
I contemplated making it two variants (IP, socket), but realised we can
do still better for the IPv4 and IPv6 types, so here it is as six.
I contemplated more precise error descriptions (e.g. “invalid IPv6
socket address syntax: expected a decimal scope ID after %”), but that’s
a more invasive change, and probably not worthwhile anyway.
2022-04-19 13:02:20 +10:00
bors
e2661bac6d
Auto merge of #95379 - icewind1991:suggest-associated-type-more, r=jackh726
...
show suggestion to replace generic bounds with associated types in more cases
Moves the hint to replace generic parameters with associated type bounds from the "not all associated type bounds are specified"(`E0191`) to "to many generic type parameters provided"(`E0107`).
Since `E0191` is only emitted in places where all associated types must be specified (when creating `dyn` types), the suggesting is currently not shown for other generic type uses (such as in generic type bounds). With this change the suggesting is always emitted when the number of excess generic parameters matches the number of unbound associated types.
Main motivation for the change was a lack of useful suggesting when doing
```rust
fn foo<I: Iterator<usize>>(i: I) {}
```
2022-04-19 01:59:35 +00:00
Erik Desjardins
1bbae507d4
mark payload fields of ScalarPair enums as Scalar::Union when they're not always initialized
2022-04-18 21:43:20 -04:00
Joshua Nelson
b4ca06558a
Remove assertion that all paths in ShouldRun exist
...
This breaks on submodules (see #96188 ). Disable the assertion for now until I can think of a proper
fix.
This doesn't revert any of the changes in `Step`s themselves, only what
`ShouldRun::paths` does.
2022-04-18 20:21:19 -05:00
Dan Gohman
19ef182655
Update the expected stderr for coerce-issue-49593-box-never.
2022-04-18 17:32:25 -07:00
Dan Gohman
b7ff10378c
Update the expected stderr for coerce-issue-49593-box-never.
...
This test's expected stderr now includes a count of the number of types
that implment `Error`. This PR introduces two new types, so increment
the number by two.
2022-04-18 16:53:50 -07:00
Dan Gohman
890125d73e
Add a comment explaining the (()) idiom for empty structs.
2022-04-18 16:53:50 -07:00
Dan Gohman
f934043c17
Split NotHandle into NullHandleError and InvalidHandleError.
...
Also, make the display messages more specific, and remove the `Copy`
implementation.
2022-04-18 16:53:50 -07:00
Dan Gohman
67994b77fd
Move the Error impl for NotHandle out of platform-independent code.
2022-04-18 16:53:49 -07:00
Dan Gohman
5b3023c564
Fix an incorrect word in a comment.
2022-04-18 16:53:38 -07:00
Dan Gohman
703a33673d
Define a dedicated error type for HandleOrNull and HandleOrInvalid.
...
Define a `NotHandle` type, that implements `std::error::Error`, and use
it as the error type in `HandleOrNull` and `HandleOrInvalid`.
2022-04-18 16:53:36 -07:00
bors
d5ae66c12c
Auto merge of #92287 - JulianKnodt:slice_remainder, r=yaahc
...
Add slice::remainder
This adds a remainder function to the Slice iterator, so that a caller can access unused
elements if iteration stops.
Addresses #91733
2022-04-18 23:34:24 +00:00
est31
3c1e1661e7
Remove unused macro rules
2022-04-18 23:28:06 +02:00
Guillaume Gomez
8d0e10cfbe
Correctly handle single :
2022-04-18 23:11:42 +02:00
bors
311e2683e1
Auto merge of #96171 - nikic:llvm-update-2, r=cuviper
...
Update LLVM submodule
Merge upstream release/14.x branch.
Fixes #94762 .
2022-04-18 20:53:42 +00:00
Gary Guo
8693227e5e
Skip broken symbol visibility test in MinGW
2022-04-18 20:50:56 +01:00
Gary Guo
e2fdb84df7
Conditionally export msan symbols only if they are defined
...
* `__msan_keep_going` is defined when `-Zsanitizer-recover=memory`.
* `__msan_track_origins` is defined when `-Zsanitizer-memory-track-origins`.
2022-04-18 20:50:56 +01:00
Gary Guo
c475117ffb
Refactor exported_symbols and linked_symbols for code reuse
2022-04-18 20:50:56 +01:00
Gary Guo
78f0e72431
Ignore paths in temporary dir in reproducible build test
2022-04-18 20:50:56 +01:00
Gary Guo
773f533eae
Synthesis object file for #[used] and exported symbols
2022-04-18 20:50:56 +01:00
Gary Guo
a1dae4bc9d
Make #[used] considered reachable
2022-04-18 20:50:56 +01:00
Gary Guo
49cc6d1f84
Add SymbolExportInfo
...
This is currently a wrapper to `SymbolExportLevel` but it allows
later addition of extra information.
2022-04-18 20:50:56 +01:00
Gary Guo
419e3ba97b
Add a test for issue 47384
2022-04-18 20:50:56 +01:00
Guillaume Gomez
5c6c1e110f
Add test to ensure that keywords in path are working
2022-04-18 21:43:19 +02:00
Guillaume Gomez
299e8ee25e
Add an extra check over filter type
2022-04-18 21:43:19 +02:00
ouz-a
5364c86e8e
replace iter with into_iter()
2022-04-18 22:05:16 +03:00
Guillaume Gomez
d7d538a8c4
Fix some corner cases
2022-04-18 20:59:09 +02:00
Guillaume Gomez
da363b2fc9
Add more tests for quote errors
2022-04-18 20:59:09 +02:00
Guillaume Gomez
6f35475ba5
Parse idents the same way in both quote string elements and "normal" elements
2022-04-18 20:59:09 +02:00
Guillaume Gomez
a6051c76db
Remove unnecessary elem.name.length === 0 since the rustdoc search eBNF does not allow elements without a name
2022-04-18 20:59:09 +02:00
Guillaume Gomez
ab9cf32131
Add tests for paths
2022-04-18 20:59:09 +02:00
Guillaume Gomez
51b4005003
Update the eBNF to allow generics bracket to not be closed if it's EOF
2022-04-18 20:59:09 +02:00
Guillaume Gomez
e03a950747
Handle separators in their own functions and fix missing handling of tabs
2022-04-18 20:59:09 +02:00
Guillaume Gomez
da829d8d9d
Improve the BNF description a bit and fix some issues
2022-04-18 20:59:09 +02:00
Guillaume Gomez
1f5c4c2b30
Forbid rustdoc search query to end with ->
2022-04-18 20:59:09 +02:00
Guillaume Gomez
8e29ed43d9
Add isIdentCharacter function to ensure that unexpected characters are handled correctly
2022-04-18 20:59:09 +02:00
