]> git.lizzy.rs Git - rust.git/log
rust.git
5 years agorustc(codegen): uncache `def_symbol_name` prefix from `symbol_name`.
Eduard-Mihai Burtescu [Thu, 21 Mar 2019 16:06:04 +0000 (18:06 +0200)]
rustc(codegen): uncache `def_symbol_name` prefix from `symbol_name`.

5 years agoAuto merge of #59487 - Centril:rollup, r=Centril
bors [Thu, 28 Mar 2019 21:46:30 +0000 (21:46 +0000)]
Auto merge of #59487 - Centril:rollup, r=Centril

Rollup of 10 pull requests

Successful merges:

 - #58717 (Add FromStr impl for NonZero types)
 - #59091 (Combine input and eval_always query types)
 - #59216 (Type dependent defs wrappers)
 - #59318 (rustc: Update linker flavor inference from filename)
 - #59320 (rustc: Allow using `clang` for wasm32 targets)
 - #59363 (#59361 Moved rustc edition opt to short list)
 - #59371 (ffi: rename VaList::copy to VaList::with_copy)
 - #59398 (Add a way to track Rustfix UI test coverage)
 - #59408 (compiletest: make path normalization smarter)
 - #59429 (When moving out of a for loop head, suggest borrowing it in nll mode)

Failed merges:

r? @ghost

5 years agoRollup merge of #59429 - estebank:for-loop-move-nll, r=petrochenkov
Mazdak Farrokhzad [Thu, 28 Mar 2019 12:35:44 +0000 (13:35 +0100)]
Rollup merge of #59429 - estebank:for-loop-move-nll, r=petrochenkov

When moving out of a for loop head, suggest borrowing it in nll mode

Follow up to #59195 for NLL.

5 years agoRollup merge of #59408 - euclio:compiletest-normalization, r=oli-obk
Mazdak Farrokhzad [Thu, 28 Mar 2019 12:35:43 +0000 (13:35 +0100)]
Rollup merge of #59408 - euclio:compiletest-normalization, r=oli-obk

compiletest: make path normalization smarter

Fixes #59109.

5 years agoRollup merge of #59398 - phansch:rustfix_coverage, r=oli-obk
Mazdak Farrokhzad [Thu, 28 Mar 2019 12:35:41 +0000 (13:35 +0100)]
Rollup merge of #59398 - phansch:rustfix_coverage, r=oli-obk

Add a way to track Rustfix UI test coverage

This came out of the first Rustfix WG meeting.

One of the goals is to enable Rustfix tests for all UI tests that
trigger lints with `MachineApplicable` suggestions. In order to do that
we first want to create a tracking issue that lists all files with
missing `// run-rustfix` headers.

This PR adds a `--rustfix-coverage` flag to `./x.py` and compiletest to
list the files with the missing headers in `/tmp/rustfix_missing_coverage.txt`.

From that file we can create the tracking issue and at some point also
enforce the `// run-rustfix` flag on UI tests with `MachineApplicable`
lints.

5 years agoRollup merge of #59371 - dlrobertson:rename_va_list_copy, r=joshtriplett
Mazdak Farrokhzad [Thu, 28 Mar 2019 12:35:40 +0000 (13:35 +0100)]
Rollup merge of #59371 - dlrobertson:rename_va_list_copy, r=joshtriplett

ffi: rename VaList::copy to VaList::with_copy

Rename `VaList::copy` to `VaList::with_copy`

r? @joshtriplett

5 years agoRollup merge of #59363 - peterjoel:issue_59361_edition_opt_shortlist, r=nagisa
Mazdak Farrokhzad [Thu, 28 Mar 2019 12:35:38 +0000 (13:35 +0100)]
Rollup merge of #59363 - peterjoel:issue_59361_edition_opt_shortlist, r=nagisa

#59361 Moved rustc edition opt to short list

Addresses #59361, in case that issue is accepted.

5 years agoRollup merge of #59320 - alexcrichton:wasm-clang, r=sanxiyn
Mazdak Farrokhzad [Thu, 28 Mar 2019 12:35:36 +0000 (13:35 +0100)]
Rollup merge of #59320 - alexcrichton:wasm-clang, r=sanxiyn

rustc: Allow using `clang` for wasm32 targets

This commit adds support code for using `clang` directly to link the
wasm32-unknown-unknown target. Currently the target is only really
configured to link with LLD directly, but this ensures that `clang` can
be configured as well.

While not immediately useful in the near term it's likely that more
wasm32 targets will pop up over time with Clang's new native support for
WebAssembly in the 8.0.0 release. Getting support into rustc early
should make it easier to experiment with these targets and try out
various changes here and there.

5 years agoRollup merge of #59318 - alexcrichton:check-for-clang, r=michaelwoerister
Mazdak Farrokhzad [Thu, 28 Mar 2019 12:35:35 +0000 (13:35 +0100)]
Rollup merge of #59318 - alexcrichton:check-for-clang, r=michaelwoerister

rustc: Update linker flavor inference from filename

This commit fixes what is believed to be a preexisting bug in the linker
flavor inference and additionally adds a new features. Previously if the
linker didn't end in `exe` the entire file name was compared to infer
the linker's flavor. This commit fixes the code to instead
unconditionally inspect `file_stem()` which is the relevant part we're
looking at to figure out what the linker flavor is.

Additionally this commit now also adds recognition of `clang` and clang
wrappers that end in `-clang` (which look like gcc wrappers). This
should allow clang-specific wrappers to get correctly inferred to the
`Gcc` linker flavor rather than the default linker flavor configured for
a target.

5 years agoRollup merge of #59216 - stepnivlk:type_dependent_defs-wrappers, r=oli-obk
Mazdak Farrokhzad [Thu, 28 Mar 2019 12:35:33 +0000 (13:35 +0100)]
Rollup merge of #59216 - stepnivlk:type_dependent_defs-wrappers, r=oli-obk

Type dependent defs wrappers

First of all, forgive me if something would seem lame to you or I offend some rule (although I tried to read through docs), this is my first PR.

Issue: https://github.com/rust-lang/rust/issues/59094

This PR adds 3 helper methods to `TypeckTables`:
* `opt_type_dependent_def`
* `opt_type_dependent_def_id`
* `type_dependent_def_id`

I didn't add `type_dependent_def` as was proposed in the issue simply because it wasn't used anywhere in the code. Only non-option wrapped`type_dependent_defs()[]` accesses were found in clippy which always called `def_id()` on result.
Speaking of clippy, should I open separate PR in its own repo, given it's used as submodule here?

Sry it took me so long, as I said I'm new here and I had tough week :).

5 years agoRollup merge of #59091 - Zoxc:eval_always, r=michaelwoerister
Mazdak Farrokhzad [Thu, 28 Mar 2019 12:35:31 +0000 (13:35 +0100)]
Rollup merge of #59091 - Zoxc:eval_always, r=michaelwoerister

Combine input and eval_always query types

Hardcoding `Krate` as a dependency of `eval_always` queries doesn't really make sense if we want to use the query system before HIR lowering / hashing. Without that hardcoding they become pretty much identical to `input` queries, so I combined them to a single type. This will regress the `clean` incremental scenario, but that isn't terribly common.

r? @michaelwoerister

5 years agoRollup merge of #58717 - hellow554:nonzero_parse, r=oli-obk
Mazdak Farrokhzad [Thu, 28 Mar 2019 12:35:29 +0000 (13:35 +0100)]
Rollup merge of #58717 - hellow554:nonzero_parse, r=oli-obk

Add FromStr impl for NonZero types

This is a WIP implementation because I do have some questions regarding the solution.

Somebody should ping the lang team on this I guess.
Please see the annotations on the code for more details.

Closes #58604

5 years agofixed shift overflow
Marcel Hellwig [Thu, 28 Mar 2019 08:31:18 +0000 (09:31 +0100)]
fixed shift overflow

Fix according to oli-obk
(https://github.com/rust-lang/rust/pull/58717#issuecomment-477494457)

5 years agoAuto merge of #59478 - Centril:rollup, r=Centril
bors [Thu, 28 Mar 2019 08:56:34 +0000 (08:56 +0000)]
Auto merge of #59478 - Centril:rollup, r=Centril

Rollup of 12 pull requests

Successful merges:

 - #57987 (Fix some AArch64 typos)
 - #58581 (Refactor generic parameter encoder functions)
 - #58803 (fs::copy() unix: set file mode early)
 - #58848 (Prevent cache issues on version updates)
 - #59198 (Do not complain about unmentioned fields in recovered patterns)
 - #59351 (Include llvm-ar with llvm-tools component)
 - #59413 (HirIdify hir::ItemId)
 - #59441 (Remove the block on natvis for lld-link.)
 - #59448 (Use consistent phrasing for all macro summaries)
 - #59456 (Add documentation about `for` used as higher ranked trait bounds)
 - #59472 (Document that `std::io::BufReader` discards contents on drop)
 - #59474 (Fix link capitalization in documentation of std::io::BufWriter.)

Failed merges:

r? @ghost

5 years agoRollup merge of #59474 - czipperz:bufwriter-fix-link-capitalization, r=Centril
Mazdak Farrokhzad [Thu, 28 Mar 2019 07:43:43 +0000 (08:43 +0100)]
Rollup merge of #59474 - czipperz:bufwriter-fix-link-capitalization, r=Centril

Fix link capitalization in documentation of std::io::BufWriter.

5 years agoRollup merge of #59472 - czipperz:bufreader-document-drop-discards, r=Centril
Mazdak Farrokhzad [Thu, 28 Mar 2019 07:43:42 +0000 (08:43 +0100)]
Rollup merge of #59472 - czipperz:bufreader-document-drop-discards, r=Centril

Document that `std::io::BufReader` discards contents on drop

Resolves #55546

5 years agoRollup merge of #59456 - czipperz:for-keyword-document-higher-ranked-trait-bounds...
Mazdak Farrokhzad [Thu, 28 Mar 2019 07:43:40 +0000 (08:43 +0100)]
Rollup merge of #59456 - czipperz:for-keyword-document-higher-ranked-trait-bounds, r=Centril

Add documentation about `for` used as higher ranked trait bounds

Resolves #55416

5 years agoRollup merge of #59448 - benesch:macro-doc, r=Centril
Mazdak Farrokhzad [Thu, 28 Mar 2019 07:43:39 +0000 (08:43 +0100)]
Rollup merge of #59448 - benesch:macro-doc, r=Centril

Use consistent phrasing for all macro summaries

None

5 years agoRollup merge of #59441 - TheGoddessInari:natvis, r=alexcrichton
Mazdak Farrokhzad [Thu, 28 Mar 2019 07:43:37 +0000 (08:43 +0100)]
Rollup merge of #59441 - TheGoddessInari:natvis, r=alexcrichton

Remove the block on natvis for lld-link.

Since #56642 bumped minimum LLVM version to 6.0.0, Rust should be able to honor or ignore Windows natvis support with lld-link.

It looks like  https://github.com/llvm-mirror/lld/commit/9133ca57b was in LLVM 7.0, while https://github.com/llvm-mirror/lld/commit/27b9c4285 made it into 6.0, at least if the release branches are anything to go by.

Fixes #59383.

5 years agoRollup merge of #59413 - Zoxc:hirid, r=oli-obk
Mazdak Farrokhzad [Thu, 28 Mar 2019 07:43:36 +0000 (08:43 +0100)]
Rollup merge of #59413 - Zoxc:hirid, r=oli-obk

HirIdify hir::ItemId

Version of https://github.com/rust-lang/rust/pull/59092.

r? @oli-obk

5 years agoRollup merge of #59351 - phil-opp:llvm-ar, r=alexcrichton
Mazdak Farrokhzad [Thu, 28 Mar 2019 07:43:34 +0000 (08:43 +0100)]
Rollup merge of #59351 - phil-opp:llvm-ar, r=alexcrichton

Include llvm-ar with llvm-tools component

Adds the `llvm-ar` tool to the `llvm-tools` component. This is useful for [building and linking native code](https://doc.rust-lang.org/cargo/reference/build-scripts.html#case-study-building-some-native-code) in cargo build scripts without needing to use the platform specific `ar`. According to #58663 it is also useful for WASM.

 `llvm-ar` is very small (~82KB), so it does not significantly increase the size of the `llvm-tools` component.

Fixes #58663

5 years agoRollup merge of #59198 - estebank:recovered-pattern, r=zackmdavis
Mazdak Farrokhzad [Thu, 28 Mar 2019 07:43:32 +0000 (08:43 +0100)]
Rollup merge of #59198 - estebank:recovered-pattern, r=zackmdavis

Do not complain about unmentioned fields in recovered patterns

When the parser has to recover from malformed code in a pattern, do not
complain about missing fields.

Fix #59145.

5 years agoRollup merge of #58848 - GuillaumeGomez:fix-cache-issues, r=Mark-Simulacrum,ollie27
Mazdak Farrokhzad [Thu, 28 Mar 2019 07:43:30 +0000 (08:43 +0100)]
Rollup merge of #58848 - GuillaumeGomez:fix-cache-issues, r=Mark-Simulacrum,ollie27

Prevent cache issues on version updates

Fixes #58827.

cc @rust-lang/infra

5 years agoRollup merge of #58803 - haraldh:fs_copy_fix, r=alexcrichton
Mazdak Farrokhzad [Thu, 28 Mar 2019 07:43:29 +0000 (08:43 +0100)]
Rollup merge of #58803 - haraldh:fs_copy_fix, r=alexcrichton

fs::copy() unix: set file mode early

A convenience method like fs::copy() should try to prevent pitfalls a
normal user doesn't think about.

In case of an empty umask, setting the file mode early prevents
temporarily world readable or even writeable files,
because the default mode is 0o666.

In case the target is a named pipe or special device node, setting the
file mode can lead to unwanted side effects, like setting permissons on
`/dev/stdout` or for root setting permissions on `/dev/null`.

copy_file_range() returns EINVAL, if the destination is a FIFO/pipe or
a device like "/dev/null", so fallback to io::copy, too.

Fixes: https://github.com/rust-lang/rust/issues/26933
Fixed: https://github.com/rust-lang/rust/issues/37885

5 years agoRollup merge of #58581 - varkor:const-generics-encoder-refactor, r=eddyb
Mazdak Farrokhzad [Thu, 28 Mar 2019 07:43:27 +0000 (08:43 +0100)]
Rollup merge of #58581 - varkor:const-generics-encoder-refactor, r=eddyb

Refactor generic parameter encoder functions

Addresses https://github.com/rust-lang/rust/pull/58503#discussion_r257488950.

r? @eddyb

5 years agoRollup merge of #57987 - parched:va-args, r=joshtriplett
Mazdak Farrokhzad [Thu, 28 Mar 2019 07:43:23 +0000 (08:43 +0100)]
Rollup merge of #57987 - parched:va-args, r=joshtriplett

Fix some AArch64 typos

cc @dlrobertson

5 years agoAuto merge of #59336 - gnzlbg:hint_black_box, r=alexcrichton
bors [Thu, 28 Mar 2019 05:44:01 +0000 (05:44 +0000)]
Auto merge of #59336 - gnzlbg:hint_black_box, r=alexcrichton

Moves test::black_box to core::hint and fix black_box on wasm32 and asm.js

This changes removes a cyclic dependency between the "test" and "libtest"
crates, where "libtest" depends on "test" for "black_box", but "test" depends on
"libtest" for everything else.

I've chosen the "hint" module because there seems to be enough consensus in the
discussion of RFC2360 that this module is where such an intrinsic would belong,
but this PR does not implement that RFC! If that RFC ever gets merged, the API, docs,
etc. of this API will need to change. This PR just move the implementation of the
already existing API.

For backwards compatibility reasons I've chosen to also keep the "test" feature
gate for these instead of adding a new feature gate. If we change the feature
gate, we'll potentially all benchmarks, and while that's something that we could
do, it seems unnecessary to do that now - if RFC2360 gets merged, we'll need to
do that anyways. Backwards compatibility is also why we continue to re-export
"black_box" from the "test" crate.

This PR also fixes black_box on the wasm32 target, which now supports inline assembly, and uses volatile loads on the asm.js target.

r? @Amanieu (cc @rust-lang/libs)

5 years agoFix link capitalization in documentation of std::io::BufWriter.
Chris Gregory [Thu, 28 Mar 2019 03:01:24 +0000 (23:01 -0400)]
Fix link capitalization in documentation of std::io::BufWriter.

5 years agoDocument that `std::io::BufReader` discards contents on drop
Chris Gregory [Thu, 28 Mar 2019 02:11:13 +0000 (22:11 -0400)]
Document that `std::io::BufReader` discards contents on drop

Resolves #55546

5 years agoAuto merge of #59471 - cuviper:rollup, r=cuviper
bors [Thu, 28 Mar 2019 02:00:57 +0000 (02:00 +0000)]
Auto merge of #59471 - cuviper:rollup, r=cuviper

Rollup of 18 pull requests

Successful merges:

 - #57293 (Make some lints incremental)
 - #57565 (syntax: Remove warning for unnecessary path disambiguators)
 - #58253 (librustc_driver => 2018)
 - #58837 (librustc_interface => 2018)
 - #59268 (Add suggestion to use `&*var` when `&str: From<String>` is expected)
 - #59283 (Make ASCII case conversions more than 4× faster)
 - #59284 (adjust MaybeUninit API to discussions)
 - #59372 (add rustfix-able suggestions to trim_{left,right} deprecations)
 - #59390 (Make `ptr::eq` documentation mention fat-pointer behavior)
 - #59393 (Refactor tuple comparison tests)
 - #59420 ([CI] record docker image info for reuse)
 - #59421 (Reject integer suffix when tuple indexing)
 - #59430 (Renames `EvalContext` to `InterpretCx`)
 - #59439 (Generalize diagnostic for `x = y` where `bool` is the expected type)
 - #59449 (fix: Make incremental artifact deletion more robust)
 - #59451 (Add `Default` to `std::alloc::System`)
 - #59459 (Add some tests)
 - #59460 (Include id in Thread's Debug implementation)

Failed merges:

r? @ghost

5 years agoMove link to rust book to next line to pass 100 column limit
Chris Gregory [Thu, 28 Mar 2019 01:46:25 +0000 (21:46 -0400)]
Move link to rust book to next line to pass 100 column limit

5 years agoRollup merge of #59460 - xfix:include-id-in-thread-debug, r=Amanieu
Josh Stone [Thu, 28 Mar 2019 01:15:44 +0000 (18:15 -0700)]
Rollup merge of #59460 - xfix:include-id-in-thread-debug, r=Amanieu

Include id in Thread's Debug implementation

Since Rust 1.19.0, `id` is a stable method, so there is no reason to not include it in Debug implementation.

5 years agoRollup merge of #59459 - JohnTitor:add-tests, r=Centril
Josh Stone [Thu, 28 Mar 2019 01:15:42 +0000 (18:15 -0700)]
Rollup merge of #59459 - JohnTitor:add-tests, r=Centril

Add some tests

close #52977

It seems that there are no tests for this issue, so I opened this PR.
off-topic: I noticed [this test](https://github.com/rust-lang/rust/blob/master/src/test/ui/existential_types/nested_existential_types.rs)'s indents are bad, could I include commit to fix this, or should I separate?

r? @oli-obk

5 years agoRollup merge of #59451 - TimDiekmann:patch-1, r=sfackler
Josh Stone [Thu, 28 Mar 2019 01:15:41 +0000 (18:15 -0700)]
Rollup merge of #59451 - TimDiekmann:patch-1, r=sfackler

Add `Default` to `std::alloc::System`

`System` is a unit struct, thus, it can be constructed without any additional information. Therefore `Default` is a noop. However, in generic code, a `T: Default` may happen as in

```rust
#[derive(Default)]
struct Foo<A> {
    allocator: A
}
```

Does this need a feature gate?
Should I also add `PartialEq/Eq/PartialOrd/Ord/Hash`?

5 years agoRollup merge of #59449 - Marwes:issue_57958, r=michaelwoerister
Josh Stone [Thu, 28 Mar 2019 01:15:39 +0000 (18:15 -0700)]
Rollup merge of #59449 - Marwes:issue_57958, r=michaelwoerister

fix: Make incremental artifact deletion more robust

Should fix the intermittent errors reported in #57958

cc #48614

5 years agoRollup merge of #59439 - Centril:generalize-assign-to-bool-diagnostic, r=oli-obk
Josh Stone [Thu, 28 Mar 2019 01:15:38 +0000 (18:15 -0700)]
Rollup merge of #59439 - Centril:generalize-assign-to-bool-diagnostic, r=oli-obk

Generalize diagnostic for `x = y` where `bool` is the expected type

Extracted out of https://github.com/rust-lang/rust/pull/59288.

Currently we special case a diagnostic for `if x = y { ...` since the expected type is `bool` in this case and we instead suggest `if x == y`. This PR generalizes this such that given an expression of form `x = y` (`ExprKind::Assign(..)`) where the expected type is `bool`, we emit a suggestion `x == y`.

r? @oli-obk

Let's do a perf run to make sure this was not the source of regressions in #59288.

5 years agoRollup merge of #59430 - kenta7777:rename-evalcontext-to-interpretcx, r=oli-obk
Josh Stone [Thu, 28 Mar 2019 01:15:36 +0000 (18:15 -0700)]
Rollup merge of #59430 - kenta7777:rename-evalcontext-to-interpretcx, r=oli-obk

Renames `EvalContext` to `InterpretCx`

This PR renames `EvalContext` to `InterpretCx` in `src/librustc_mir`.
This PR is related to #54395 .

5 years agoRollup merge of #59421 - estebank:tuple-index-suffix, r=petrochenkov
Josh Stone [Thu, 28 Mar 2019 01:15:35 +0000 (18:15 -0700)]
Rollup merge of #59421 - estebank:tuple-index-suffix, r=petrochenkov

Reject integer suffix when tuple indexing

Fix #59418.

r? @varkor

5 years agoRollup merge of #59420 - cuviper:ci-image-info, r=alexcrichton
Josh Stone [Thu, 28 Mar 2019 01:15:33 +0000 (18:15 -0700)]
Rollup merge of #59420 - cuviper:ci-image-info, r=alexcrichton

[CI] record docker image info for reuse

This writes an extra `dist/image-$image.txt` which contains the S3 URL
of the cached image and the `sha256` digest of the docker entry point.
This will be uploaded with the rest of the deployed artifacts in the
Travis `after_success` script.

cc rust-lang/rustup.rs#1724
r? @alexcrichton

5 years agoRollup merge of #59393 - czipperz:refactor_tuple_comparison_tests, r=shepmaster
Josh Stone [Thu, 28 Mar 2019 01:15:32 +0000 (18:15 -0700)]
Rollup merge of #59393 - czipperz:refactor_tuple_comparison_tests, r=shepmaster

Refactor tuple comparison tests

5 years agoRollup merge of #59390 - czipperz:ptr_eq_smart_pointer, r=Centril,steveklabnik
Josh Stone [Thu, 28 Mar 2019 01:15:30 +0000 (18:15 -0700)]
Rollup merge of #59390 - czipperz:ptr_eq_smart_pointer, r=Centril,steveklabnik

Make `ptr::eq` documentation mention fat-pointer behavior

Resolves #59214

5 years agoRollup merge of #59372 - euclio:rename-trim, r=rkruppe
Josh Stone [Thu, 28 Mar 2019 01:15:28 +0000 (18:15 -0700)]
Rollup merge of #59372 - euclio:rename-trim, r=rkruppe

add rustfix-able suggestions to trim_{left,right} deprecations

Fixes #53802 (technically already fixed by #58002, but that issue is about these methods).

5 years agoRollup merge of #59284 - RalfJung:maybe-uninit, r=sfackler
Josh Stone [Thu, 28 Mar 2019 01:15:27 +0000 (18:15 -0700)]
Rollup merge of #59284 - RalfJung:maybe-uninit, r=sfackler

adjust MaybeUninit API to discussions

uninitialized -> uninit
into_initialized -> assume_init
read_initialized -> read
set -> write

5 years agoRollup merge of #59283 - SimonSapin:branchless-ascii-case, r=joshtriplett
Josh Stone [Thu, 28 Mar 2019 01:15:25 +0000 (18:15 -0700)]
Rollup merge of #59283 - SimonSapin:branchless-ascii-case, r=joshtriplett

Make ASCII case conversions more than 4× faster

Reformatted output of `./x.py bench src/libcore --test-args ascii` below. The `libcore` benchmark calls `[u8]::make_ascii_lowercase`. `lookup` has code (effectively) identical to that before this PR, and ~~`branchless`~~ `mask_shifted_bool_match_range` after this PR.

~~See [code comments](https://github.com/rust-lang/rust/pull/59283/commits/ce933f77c865a15670855ac5941fe200752b739f#diff-01076f91a26400b2db49663d787c2576R3796) in `u8::to_ascii_uppercase` in `src/libcore/num/mod.rs` for an explanation of the branchless algorithm.~~

**Update:** the algorithm was simplified while keeping the performance. See `branchless` v.s. `mask_shifted_bool_match_range` benchmarks.

Credits to @raphlinus for the idea in https://twitter.com/raphlinus/status/1107654782544736261, which extends this algorithm to “fake SIMD” on `u32` to convert four bytes at a time. The `fake_simd_u32` benchmarks implements this with [`let (before, aligned, after) = bytes.align_to_mut::<u32>()`](https://doc.rust-lang.org/std/primitive.slice.html#method.align_to_mut). Note however that this is buggy when addition carries/overflows into the next byte (which does not happen if the input is known to be ASCII).

This could be fixed (to optimize `[u8]::make_ascii_lowercase` and `[u8]::make_ascii_uppercase` in `src/libcore/slice/mod.rs`) either with some more bitwise trickery that I didn’t quite figure out, or by using “real” SIMD intrinsics for byte-wise addition. I did not pursue this however because the current (incorrect) fake SIMD algorithm is only marginally faster than the one-byte-at-a-time branchless algorithm. This is because LLVM auto-vectorizes the latter, as can be seen on https://rust.godbolt.org/z/anKtbR.

Benchmark results on Linux x64 with Intel i7-7700K: (updated from https://github.com/rust-lang/rust/pull/59283#issuecomment-474146863)

```rust
6830 bytes string:

alloc_only                          ... bench:    112 ns/iter (+/- 0) = 62410 MB/s
black_box_read_each_byte            ... bench:  1,733 ns/iter (+/- 8) = 4033 MB/s
lookup_table                        ... bench:  1,766 ns/iter (+/- 11) = 3958 MB/s
branch_and_subtract                 ... bench:    417 ns/iter (+/- 1) = 16762 MB/s
branch_and_mask                     ... bench:    401 ns/iter (+/- 1) = 17431 MB/s
branchless                          ... bench:    365 ns/iter (+/- 0) = 19150 MB/s
libcore                             ... bench:    367 ns/iter (+/- 1) = 19046 MB/s
fake_simd_u32                       ... bench:    361 ns/iter (+/- 2) = 19362 MB/s
fake_simd_u64                       ... bench:    361 ns/iter (+/- 1) = 19362 MB/s
mask_mult_bool_branchy_lookup_table ... bench:  6,309 ns/iter (+/- 19) = 1107 MB/s
mask_mult_bool_lookup_table         ... bench:  4,183 ns/iter (+/- 29) = 1671 MB/s
mask_mult_bool_match_range          ... bench:    339 ns/iter (+/- 0) = 20619 MB/s
mask_shifted_bool_match_range       ... bench:    339 ns/iter (+/- 1) = 20619 MB/s

32 bytes string:

alloc_only                          ... bench:     15 ns/iter (+/- 0) = 2133 MB/s
black_box_read_each_byte            ... bench:     29 ns/iter (+/- 0) = 1103 MB/s
lookup_table                        ... bench:     24 ns/iter (+/- 4) = 1333 MB/s
branch_and_subtract                 ... bench:     16 ns/iter (+/- 0) = 2000 MB/s
branch_and_mask                     ... bench:     16 ns/iter (+/- 0) = 2000 MB/s
branchless                          ... bench:     16 ns/iter (+/- 0) = 2000 MB/s
libcore                             ... bench:     15 ns/iter (+/- 0) = 2133 MB/s
fake_simd_u32                       ... bench:     17 ns/iter (+/- 0) = 1882 MB/s
fake_simd_u64                       ... bench:     16 ns/iter (+/- 0) = 2000 MB/s
mask_mult_bool_branchy_lookup_table ... bench:     42 ns/iter (+/- 0) = 761 MB/s
mask_mult_bool_lookup_table         ... bench:     35 ns/iter (+/- 0) = 914 MB/s
mask_mult_bool_match_range          ... bench:     16 ns/iter (+/- 0) = 2000 MB/s
mask_shifted_bool_match_range       ... bench:     16 ns/iter (+/- 0) = 2000 MB/s

7 bytes string:

alloc_only                          ... bench:     14 ns/iter (+/- 0) = 500 MB/s
black_box_read_each_byte            ... bench:     22 ns/iter (+/- 0) = 318 MB/s
lookup_table                        ... bench:     16 ns/iter (+/- 0) = 437 MB/s
branch_and_subtract                 ... bench:     16 ns/iter (+/- 0) = 437 MB/s
branch_and_mask                     ... bench:     16 ns/iter (+/- 0) = 437 MB/s
branchless                          ... bench:     19 ns/iter (+/- 0) = 368 MB/s
libcore                             ... bench:     20 ns/iter (+/- 0) = 350 MB/s
fake_simd_u32                       ... bench:     18 ns/iter (+/- 0) = 388 MB/s
fake_simd_u64                       ... bench:     21 ns/iter (+/- 0) = 333 MB/s
mask_mult_bool_branchy_lookup_table ... bench:     20 ns/iter (+/- 0) = 350 MB/s
mask_mult_bool_lookup_table         ... bench:     19 ns/iter (+/- 0) = 368 MB/s
mask_mult_bool_match_range          ... bench:     19 ns/iter (+/- 0) = 368 MB/s
mask_shifted_bool_match_range       ... bench:     19 ns/iter (+/- 0) = 368 MB/s
```

5 years agoRollup merge of #59268 - estebank:from-string, r=QuietMisdreavus
Josh Stone [Thu, 28 Mar 2019 01:15:24 +0000 (18:15 -0700)]
Rollup merge of #59268 - estebank:from-string, r=QuietMisdreavus

Add suggestion to use `&*var` when `&str: From<String>` is expected

Fix #53879.

5 years agoRollup merge of #58837 - Centril:librustc_interface_2018, r=petrochenkov
Josh Stone [Thu, 28 Mar 2019 01:15:22 +0000 (18:15 -0700)]
Rollup merge of #58837 - Centril:librustc_interface_2018, r=petrochenkov

librustc_interface => 2018

r? @oli-obk

This will likely produce an ICE for some reason... so super-WIP.

5 years agoRollup merge of #58253 - taiki-e:librustc_driver-2018, r=petrochenkov
Josh Stone [Thu, 28 Mar 2019 01:15:21 +0000 (18:15 -0700)]
Rollup merge of #58253 - taiki-e:librustc_driver-2018, r=petrochenkov

librustc_driver => 2018

Transitions `librustc_driver` to Rust 2018; cc #58099

r? @Centril

5 years agoRollup merge of #57565 - petrochenkov:turbowarn, r=Centril
Josh Stone [Thu, 28 Mar 2019 01:15:19 +0000 (18:15 -0700)]
Rollup merge of #57565 - petrochenkov:turbowarn, r=Centril

syntax: Remove warning for unnecessary path disambiguators

`rustfmt` is now stable and it removes unnecessary turbofishes, so removing the warning as discussed in https://github.com/rust-lang/rust/pull/43540 (where it was introduced).
One hardcoded warning less.

Closes https://github.com/rust-lang/rust/issues/58055

r? @nikomatsakis

5 years agoRollup merge of #57293 - Zoxc:incr-passes3, r=michaelwoerister
Josh Stone [Thu, 28 Mar 2019 01:15:18 +0000 (18:15 -0700)]
Rollup merge of #57293 - Zoxc:incr-passes3, r=michaelwoerister

Make some lints incremental

Blocked on https://github.com/rust-lang/rust/pull/57253

r? @michaelwoerister

5 years agoAdd higher-ranked trait bounds link
Chris Gregory [Wed, 27 Mar 2019 16:03:14 +0000 (12:03 -0400)]
Add higher-ranked trait bounds link

5 years agoClarify behavior of dbg macro
Nikhil Benesch [Wed, 27 Mar 2019 15:22:55 +0000 (11:22 -0400)]
Clarify behavior of dbg macro

5 years agoAuto merge of #59415 - varkor:values_since_snapshot, r=eddyb
bors [Wed, 27 Mar 2019 13:20:16 +0000 (13:20 +0000)]
Auto merge of #59415 - varkor:values_since_snapshot, r=eddyb

Refactor InferenceFudger (née RegionFudger)

- Rename `RegionFudger` (and related methods) to `InferenceFudger`.
- Take integer and float inference variables into account.
- Refactor `types_created_since_snapshot` and `vars_created_since_snapshot` with the [new version of ena](https://github.com/rust-lang-nursery/ena/pull/21).
- Some other refactoring in the area.

r? @eddyb

5 years agoInclude id in Thread's Debug implementation
Konrad Borowski [Wed, 27 Mar 2019 11:28:17 +0000 (12:28 +0100)]
Include id in Thread's Debug implementation

Since Rust 1.19.0, id is a stable method, so there is no reason to
not include it in Debug implementation.

5 years agoAdd some tests
Yuki OKUSHI [Wed, 27 Mar 2019 10:30:33 +0000 (19:30 +0900)]
Add some tests

5 years agoUse Vec instead of FxHashMap
varkor [Mon, 25 Mar 2019 22:26:37 +0000 (22:26 +0000)]
Use Vec instead of FxHashMap

5 years agoLookup region variable origin instead of choosing one
varkor [Mon, 25 Mar 2019 21:28:39 +0000 (21:28 +0000)]
Lookup region variable origin instead of choosing one

5 years agoStore type variable origins in InferenceFudger
varkor [Mon, 25 Mar 2019 21:28:15 +0000 (21:28 +0000)]
Store type variable origins in InferenceFudger

5 years agoPropitiate tidy
varkor [Mon, 25 Mar 2019 16:48:57 +0000 (16:48 +0000)]
Propitiate tidy

5 years agoUpdate ena to version 0.13.0
varkor [Mon, 25 Mar 2019 15:45:44 +0000 (15:45 +0000)]
Update ena to version 0.13.0

5 years agoUse `eq_relations`
varkor [Thu, 21 Mar 2019 17:11:42 +0000 (17:11 +0000)]
Use `eq_relations`

5 years agoAdd `next_int_var` and `next_float_var`
varkor [Thu, 21 Mar 2019 13:47:57 +0000 (13:47 +0000)]
Add `next_int_var` and `next_float_var`

5 years agoSimplify `fudge_inference_if_ok`
varkor [Thu, 21 Mar 2019 13:43:37 +0000 (13:43 +0000)]
Simplify `fudge_inference_if_ok`

5 years agoAdd int variables and float variables to `InferenceFudger`
varkor [Thu, 21 Mar 2019 13:39:57 +0000 (13:39 +0000)]
Add int variables and float variables to `InferenceFudger`

5 years agoRename `RegionFudger` to `InferenceFudger`
varkor [Thu, 21 Mar 2019 12:39:05 +0000 (12:39 +0000)]
Rename `RegionFudger` to `InferenceFudger`

5 years agoRemove `TypeVariableMap`
varkor [Thu, 21 Mar 2019 12:38:05 +0000 (12:38 +0000)]
Remove `TypeVariableMap`

5 years agoUse Ranges for vars_since_snapshot
varkor [Thu, 21 Mar 2019 12:37:31 +0000 (12:37 +0000)]
Use Ranges for vars_since_snapshot

5 years agoSimplify `TypeVariableTable::vars_since_snapshot`
varkor [Wed, 20 Mar 2019 10:07:30 +0000 (10:07 +0000)]
Simplify `TypeVariableTable::vars_since_snapshot`

5 years agoSimplify `RegionConstraintCollector::vars_since_snapshot`
varkor [Wed, 20 Mar 2019 01:54:53 +0000 (01:54 +0000)]
Simplify `RegionConstraintCollector::vars_since_snapshot`

5 years agoMake `vars_since_snapshot` naming consistent
varkor [Wed, 20 Mar 2019 00:57:25 +0000 (00:57 +0000)]
Make `vars_since_snapshot` naming consistent

5 years agoUpdate ena
varkor [Tue, 19 Mar 2019 21:30:07 +0000 (21:30 +0000)]
Update ena

5 years agoadd negative test case in assignment-expected-bool
Mazdak Farrokhzad [Wed, 27 Mar 2019 09:19:35 +0000 (10:19 +0100)]
add negative test case in assignment-expected-bool

5 years agoadjust assignment-in-if test accordingly.
Mazdak Farrokhzad [Tue, 26 Mar 2019 13:57:17 +0000 (14:57 +0100)]
adjust assignment-in-if test accordingly.

5 years agoadd test for assignment x = y where type bool is expected.
Mazdak Farrokhzad [Tue, 26 Mar 2019 13:56:32 +0000 (14:56 +0100)]
add test for assignment x = y where type bool is expected.

5 years agogeneralize diagnostic for x = y where type bool is expected.
Mazdak Farrokhzad [Tue, 26 Mar 2019 14:02:02 +0000 (15:02 +0100)]
generalize diagnostic for x = y where type bool is expected.

5 years agoAuto merge of #55780 - ogoffart:span_source_text, r=petrochenkov
bors [Wed, 27 Mar 2019 08:58:40 +0000 (08:58 +0000)]
Auto merge of #55780 - ogoffart:span_source_text, r=petrochenkov

Introduce proc_macro::Span::source_text

A function to extract the actual source behind a Span.

Background: I would like to use `syn` in a `build.rs` script to parse the rust code, and extract part of the source code. However, `syn` only gives access to proc_macro2::Span, and i would like to get the source code behind that.
I opened an issue on proc_macro2 bug tracker for this feature https://github.com/alexcrichton/proc-macro2/issues/110  and @alexcrichton said the feature should first go upstream in proc_macro.  So there it is!

Since most of the Span API is unstable anyway, this is guarded by the same `proc_macro_span` feature as everything else.

5 years agolibrustc_interface => 2018; rename rustc-rayon to rayon in Cargo.toml
Mazdak Farrokhzad [Wed, 27 Mar 2019 08:48:50 +0000 (09:48 +0100)]
librustc_interface => 2018; rename rustc-rayon to rayon in Cargo.toml

5 years agolibrustc_interface => 2018
Mazdak Farrokhzad [Tue, 26 Mar 2019 18:07:13 +0000 (19:07 +0100)]
librustc_interface => 2018

5 years agoMake into itemized list and fix some wording
Chris Gregory [Wed, 27 Mar 2019 05:52:55 +0000 (01:52 -0400)]
Make into itemized list and fix some wording

5 years agoMinor rewordings and add `dyn` keyword
Chris Gregory [Wed, 27 Mar 2019 05:46:24 +0000 (01:46 -0400)]
Minor rewordings and add `dyn` keyword

5 years agoAuto merge of #59447 - GuillaumeGomez:rollup, r=GuillaumeGomez
bors [Wed, 27 Mar 2019 05:25:50 +0000 (05:25 +0000)]
Auto merge of #59447 - GuillaumeGomez:rollup, r=GuillaumeGomez

Rollup of 7 pull requests

Successful merges:

 - #59004 ([rustdoc] Improve "in parameters" search and search more generally)
 - #59026 (Fix moving text in search tabs headers)
 - #59197 (Exclude old book redirect stubs from search engines)
 - #59330 (Improve the documentation for std::convert (From, Into, AsRef and AsMut))
 - #59424 (Fix code block display in portability element in dark theme)
 - #59427 (Link to PhantomData in NonNull documentation)
 - #59432 (Improve some compiletest documentation)

Failed merges:

r? @ghost

5 years agoAdd documentation about `for` used as higher ranked trait bounds
Chris Gregory [Wed, 27 Mar 2019 05:22:23 +0000 (01:22 -0400)]
Add documentation about `for` used as higher ranked trait bounds

Resolves #55416

5 years agorenames EvalContext to InterpretCx in docs and comments.
kenta7777 [Wed, 27 Mar 2019 01:57:03 +0000 (10:57 +0900)]
renames EvalContext to InterpretCx in docs and comments.

5 years agoAuto merge of #59285 - cuviper:llvm-8.0.0, r=alexcrichton
bors [Wed, 27 Mar 2019 01:56:14 +0000 (01:56 +0000)]
Auto merge of #59285 - cuviper:llvm-8.0.0, r=alexcrichton

Rebase LLVM to 8.0.0 final

r? @alexcrichton

5 years agoAdd `Default` to `std::alloc::System`
Tim Diekmann [Tue, 26 Mar 2019 23:09:12 +0000 (00:09 +0100)]
Add `Default` to `std::alloc::System`

5 years agoFix error index CSS file name
Guillaume Gomez [Tue, 26 Mar 2019 22:57:16 +0000 (23:57 +0100)]
Fix error index CSS file name

5 years agofix: Make incremental artifact deletion more robust
Markus Westerlind [Tue, 26 Mar 2019 22:45:40 +0000 (23:45 +0100)]
fix: Make incremental artifact deletion more robust

Should fix the intermittent errors reported in #57958

cc #48614

5 years agoUse consistent phrasing for all macro summaries
Nikhil Benesch [Tue, 26 Mar 2019 22:22:25 +0000 (18:22 -0400)]
Use consistent phrasing for all macro summaries

5 years agoHandle RUSTDOC_RESOURCE_SUFFIX env variable for rustdoc build
Guillaume Gomez [Mon, 25 Mar 2019 21:01:19 +0000 (22:01 +0100)]
Handle RUSTDOC_RESOURCE_SUFFIX env variable for rustdoc build

5 years agoAdd resource suffix for libtest and proc_macro as well
Guillaume Gomez [Tue, 12 Mar 2019 23:32:13 +0000 (00:32 +0100)]
Add resource suffix for libtest and proc_macro as well

5 years agoPrevent cache issues on version updates
Guillaume Gomez [Fri, 1 Mar 2019 16:03:15 +0000 (17:03 +0100)]
Prevent cache issues on version updates

5 years agoRollup merge of #59432 - phansch:compiletest_docs, r=alexcrichton
Guillaume Gomez [Tue, 26 Mar 2019 21:26:46 +0000 (22:26 +0100)]
Rollup merge of #59432 - phansch:compiletest_docs, r=alexcrichton

Improve some compiletest documentation

This adds some missing documentation for rustfix related things and adds
a test for the `is_test` function.

5 years agoRollup merge of #59427 - czipperz:non_null_doc_links, r=Mark-Simulacrum
Guillaume Gomez [Tue, 26 Mar 2019 21:26:45 +0000 (22:26 +0100)]
Rollup merge of #59427 - czipperz:non_null_doc_links, r=Mark-Simulacrum

Link to PhantomData in NonNull documentation

5 years agoRollup merge of #59424 - GuillaumeGomez:fix-stability-css, r=QuietMisdreavus
Guillaume Gomez [Tue, 26 Mar 2019 21:26:43 +0000 (22:26 +0100)]
Rollup merge of #59424 - GuillaumeGomez:fix-stability-css, r=QuietMisdreavus

Fix code block display in portability element in dark theme

Fixes #59261.

r? @QuietMisdreavus

A little screenshot:

<img width="521" alt="Screenshot 2019-03-26 at 00 37 49" src="https://user-images.githubusercontent.com/3050060/54961082-9a41c600-4f5f-11e9-8040-ae6f26d368ff.png">

5 years agoRollup merge of #59330 - DevQps:improve-std-convert-documentation, r=steveklabnik
Guillaume Gomez [Tue, 26 Mar 2019 21:26:42 +0000 (22:26 +0100)]
Rollup merge of #59330 - DevQps:improve-std-convert-documentation, r=steveklabnik

Improve the documentation for std::convert (From, Into, AsRef and AsMut)

# Description
In this PR I updated the documentation of From, Into, AsRef and AsMut, as well as the general std::convert module documentation. The discussion in #59163 provided information that was not yet present in the docs, or was not expressed clearly enough. I tried to clarify the examples that were already present in the docs as well as add more information about considered best-practices that came out of the discussion in #59163

@steveklabnik I hope I didn't change too much. This is an initial version! I will scan through everything tomorrow as well again to see if I made any typo's or errors, and maybe make some small changes here and there.

All suggestions are welcome!

closes #59163

5 years agoRollup merge of #59197 - kornelski:redir, r=steveklabnik
Guillaume Gomez [Tue, 26 Mar 2019 21:26:40 +0000 (22:26 +0100)]
Rollup merge of #59197 - kornelski:redir, r=steveklabnik

Exclude old book redirect stubs from search engines

Adds `<meta name="robots" content="noindex,follow">` to the `<head>` of old stub pages pointing to the second edition of the book.

This is continuation of https://github.com/rust-lang/book/pull/1788

5 years agoRollup merge of #59026 - GuillaumeGomez:search-tabs-header, r=QuietMisdreavus
Guillaume Gomez [Tue, 26 Mar 2019 21:26:39 +0000 (22:26 +0100)]
Rollup merge of #59026 - GuillaumeGomez:search-tabs-header, r=QuietMisdreavus

Fix moving text in search tabs headers

Fixes #59005.

Now, the text in the search tabs headers isn't moving anymore.

r? @QuietMisdreavus

5 years agoRollup merge of #59004 - GuillaumeGomez:generics-handling, r=QuietMisdreavus
Guillaume Gomez [Tue, 26 Mar 2019 21:26:36 +0000 (22:26 +0100)]
Rollup merge of #59004 - GuillaumeGomez:generics-handling, r=QuietMisdreavus

[rustdoc] Improve "in parameters" search and search more generally

Fixes #58230.

r? @QuietMisdreavus

5 years agoMake crate_variances a regular query
John Kåre Alsaker [Thu, 14 Mar 2019 02:10:47 +0000 (03:10 +0100)]
Make crate_variances a regular query

5 years agoCombine input and eval_always query types
John Kåre Alsaker [Sun, 10 Mar 2019 09:11:15 +0000 (10:11 +0100)]
Combine input and eval_always query types

5 years agolibrustc_driver => 2018
Taiki Endo [Tue, 26 Mar 2019 20:35:18 +0000 (05:35 +0900)]
librustc_driver => 2018