Auto merge of #52564 - pnkfelix:issue-52126-lhs-of-assign-op-is-invariant, r=eddyb
LHS of assign op is invariant
This addresses a bug injected by #45435. That PR changed the way we type-check `LHS <op> RHS` to coerce the LHS to the expected supertype in much the same way that we coerce the RHS.
The problem is that when we have a case of `LHS <op>= RHS`, we do not want to coerce to a supertype; we need the type to remain invariant. Otherwise we risk leaking a value with short-lifetimes into a expression context that needs to satisfy a long lifetime.
Auto merge of #52069 - zackmdavis:elided_states_of_america—and_to_the_re-pub-lic, r=nikomatsakis
add structured suggestions and fix false-positive for elided-lifetimes-in-paths lint
This adds structured suggestions to the elided-lifetimes-in-paths lint (introduced in Nov. 2017's #46254), prevents it from emitting a false-positive on anonymous (underscore) lifetimes (!), and adds it to the idioms-2018 group (#52041).
~~As an aside, "elided-lifetimes-in-paths" seems like an unfortunate name, because it's not clear exactly what "elided" means. The motivation for this lint (see original issue #45992, and [RFC 2115](https://github.com/rust-lang/rfcs/blob/e978a8d3017a01d632f916140c98802505cd1324/text/2115-argument-lifetimes.md#motivation)) seems to be specifically about not supplying angle-bracketed lifetime arguments to non-`&` types, but (1) the phrase "lifetime elision" has historically also referred to the ability to not supply a lifetime name to `&` references, and (2) an `is_elided` method in the HIR returns true for anoymous/underscore lifetimes, which is _not_ what we're trying to lint here. (That naming confusion is almost certainly what led to the false positive addressed here.) Given that the lint is relatively new and is allow-by-default, is it too late to rename it ... um, _again_ (#50879)?~~
~~This does _not_ address a couple of other false positives discovered in https://github.com/rust-lang/rust/issues/52041#issuecomment-402547901.~~
- #51807 (Deprecation of str::slice_unchecked(_mut))
- #52051 (mem::swap the obvious way for types smaller than the SIMD optimization's block size)
- #52465 (Add CI test harness for `thumb*` targets. [IRR-2018-embedded])
- #52507 (Reword when `_` couldn't be inferred)
- #52508 (Document that Unique::empty() and NonNull::dangling() aren't sentinel values)
- #52521 (Fix links in rustdoc book.)
- #52581 (Avoid using `#[macro_export]` for documenting builtin macros)
- #52582 (Typo)
- #52587 (Add missing backtick in UniversalRegions doc comment)
- #52594 (Run the error index tool against the sysroot libdir)
- #52615 (Added new lines to .gitignore.)
Rollup merge of #52051 - scottmcm:swap-directly, r=alexcrichton
mem::swap the obvious way for types smaller than the SIMD optimization's block size
LLVM isn't able to remove the alloca for the unaligned block in the post-SIMD tail in some cases, so doing this helps SRoA work in cases where it currently doesn't. Found in the `replace_with` RFC discussion.
Examples of the improvements:
<details>
<summary>swapping `[u16; 3]` takes 1/3 fewer instructions and no stackalloc</summary>
This PR:
```asm
_ZN4blah7inc_opt17h1426dc215ecbdb19E:
xor eax, eax
cmp dword ptr [rcx], 0
setne al
mov dword ptr [rcx], eax
add dword ptr [rcx + 4], 1
ret
```
Where that add is beautiful -- using an addressing mode to not even need to explicitly go through a register -- and the remaining imperfection is well-known (https://github.com/rust-lang/rust/pull/49420#issuecomment-376805721).
</details>
Rollup merge of #52615 - davidtwco:add-to-gitignore, r=nikomatsakis,Mark-Simulacrum
Added new lines to .gitignore.
There are a handful of files that I often find in my local working directories that I never want to commit that aren't covered in the `.gitignore` file:
`/mir_dump`:
Default output location from `-Z mir-dump=all` for a specific test, I can't think of a reason why this should ever be commited.
`tags*`/`TAGS*`:
I use `vim-gutentags` which outputs `tags` and `tags.temp` which I don't want commited. I also collapsed the `TAGS`, `TAGS.vi`, `TAGS.emacs` into `TAGS*`.
`Session.vim`:
I use `vim-obsession` to save my current session in Vim, it outputs a `Session.vim` file, this also shouldn't be commited.
Rollup merge of #52594 - Mark-Simulacrum:error-index-stage0, r=alexcrichton
Run the error index tool against the sysroot libdir
Previously when building the error index tool in stage 0 we would
attempt to use stage 0 libraries, but because it depends on rustdoc,
those don't exist: rustdoc is built against stage 1 libraries.
This patch aligns those two and passes the stage 1 libdir to the error
index.
@GuillaumeGomez discovered that this hasn't worked (presumably for a long time now, but not sure).
Zack M. Davis [Tue, 17 Jul 2018 07:30:53 +0000 (00:30 -0700)]
in which the elided-lifetimes-in-paths lint undergoes a revolution
The existing elided-lifetimes-in-paths lint (introduced in Nov. 2017's accd997b5 / #46254) lacked stuctured suggestions and—much more
alarmingly—produced false positives on associated functions (like
`Ref::clone`) and on anonymous '_ lifetimes (!!—yes, the very
anonymous lifetimes that we meant to suggest "instead"). That this
went apparently unnoticed for so long maybe tells you something about
how many people actually bother to flip on allow-by-default lints.
After many hours of good old-fashioned American elbow grease—and a
little help from expert reviewers—it turns out that getting the right
answer is a lot easier if we fire the lint while lowering the Higher
Intermediate Representation.
The lint is promoted to the idioms-2018 group.
Also, in the matter of test filenames, "elided" only has one 'l' (see,
e.g., https://en.wiktionary.org/wiki/elide).
Rollup merge of #52508 - joshtriplett:dangling-not-sentinel, r=Mark-Simulacrum
Document that Unique::empty() and NonNull::dangling() aren't sentinel values
The documentation of Unique::empty() and NonNull::dangling() could
potentially suggest that they work as sentinel values indicating a
not-yet-initialized pointer. However, they both declare a non-null
pointer equal to the alignment of the type, which could potentially
reference a valid value of that type (specifically, the first such valid
value in memory). Explicitly document that the return value of these
functions does not work as a sentinel value.
Rollup merge of #52465 - sekineh:add-ci-thumb, r=alexcrichton
Add CI test harness for `thumb*` targets. [IRR-2018-embedded]
This pull request will do the following (rather trivial) changes:
- Fix #52163. In other words, we enabled `./x.py test src/test/run-make` for `no_std` targets.
- Modify `dist-various-1` Dockerfile.
- CI now performs `run-make` test run on the targets below:
- `thumbv6m-none-eabi`
- `thumbv7m-none-eabi`
- `thumbv7em-none-eabi`
- `thumbv7em-none-eabihf`.
- ~~Add `thumb-none` Dockerfile.~~
- ~~Initially, `thumbv7m-none-eabi`, `thumbv7em-none-eabi` and `thumbv7em-none-eabihf` are included as the tested target. `thumbv6m-none-eabi` is disabled for now because LLVM support is not certain.~~
- ~~Add `thumb-none` to .travis.yml~~
Note:
- `run-make` tests are not implemented yet. This PR is test harness only.
The amount of change is very small, but I'd like to open the pull request while the change is trivial.
Because I'm not very used to pull request process, I want to make a small progress first. This PR will be a foundation for later additions.
Rollup merge of #51807 - newpavlov:deprecate_str_slice, r=alexcrichton
Deprecation of str::slice_unchecked(_mut)
Closes #51715
I am not sure if 1.28.0 or 1.29.0 should be used for deprecation version, for now it's 1.28.0.
Additionally I've replaced `slice_unchecked` uses with `get_unchecked`. The only places where this method is still used are `src/liballoc/tests/str.rs` and `src/liballoc/tests/str.rs`.
Auto merge of #52359 - matthewjasper:combine-move-error-reporting, r=pnkfelix
[NLL] Small move error reporting improvements
* Use a MirBorrowckContext when reporting errors to be more uniform with other error reporting
* Add a special message for the case of trying to move from capture variables in `Fn` and `FnMut` closures.
Scott McMurray [Wed, 4 Jul 2018 09:48:30 +0000 (02:48 -0700)]
Don't use SIMD in mem::swap for types smaller than the block size
LLVM isn't able to remove the alloca for the unaligned block in the SIMD tail in some cases, so doing this helps SRoA work in cases where it currently doesn't. Found in the `replace_with` RFC discussion.
Auto merge of #52250 - nnethercote:no-SparseBitMatrix, r=nikomatsakis
Speed up `SparseBitMatrix` use in `RegionValues`.
In practice, these matrices range from 10% to 90%+ full once they are
filled in, so the dense representation is better.
This reduces the runtime of Check Nll builds of `inflate` by 32%, and
several other benchmarks by 1--5%.
It also increases max-rss of `clap-rs` by 30% and a couple of others by
up to 5%, while decreasing max-rss of `coercions` by 14%. I think the
speed-ups justify the max-rss increases.
Mark Rousskov [Sat, 21 Jul 2018 21:19:39 +0000 (15:19 -0600)]
Run the error index tool against the sysroot libdir
Previously when building the error index tool in stage 0 we would
attempt to use stage 0 libraries, but because it depends on rustdoc,
those don't exist: rustdoc is built against stage 1 libraries.
This patch aligns those two and passes the stage 1 libdir to the error
index.
Auto merge of #52405 - matthewjasper:mutability-errors, r=pnkfelix
[NLL] Mutability errors
cc #51028
cc #51170
cc #46559
Closes #46629
* Better explain why the place is immutable ("immutable item" is gone)
* Distinguish &T and *const T
* Use better spans when a mutable borrow is for a closure capture
Auto merge of #52555 - petrochenkov:mresfact, r=alexcrichton
resolve: Some renaming, refactoring and comments
Commits are self-descriptive.
The only functional change is https://github.com/rust-lang/rust/commit/34bf2f572e33d4df1459413b5014ca98fc9fa4e0 that tightens shadowing rules for macro paths (makes the second and third cases in `test/ui/imports/glob-shadowing.rs` an error).
Auto merge of #52552 - eddyb:proc-macro-prep, r=alexcrichton
Prepare proc_macro for decoupling it from the rest of the compiler.
This is #49219 up to the point where the bridge is introduced. Aside from moving some code around, the largest change is the rewrite of `proc_macro::quote` to be simpler and do less introspection.
I'd like to also extend `quote!` with `${stmt;...;expr}` instead of just `$variable` (and maybe even `$(... $iter ...)*`), which seems pretty straight-forward now, but I don't know if/when I should.
r? @alexcrichton or @dtolnay cc @jseyfried @petrochenkov
Auto merge of #52536 - alexcrichton:attr-spans, r=nikomatsakis
proc_macro: Preserve spans of attributes on functions
This commit updates the tokenization of items which are subsequently passed to
`proc_macro` to ensure that span information is preserved on attributes as much
as possible. Previously this area of the code suffered from #43081 where we
haven't actually implemented converting an attribute to to a token tree yet, but
a local fix was possible here.
Rollup merge of #52539 - alexcrichton:two-attrs, r=petrochenkov
rustc: Fix two custom attributes with custom derive
This commit fixes an issue where multiple custom attributes could not be fed
into a custom derive in some situations with the `use_extern_macros` feature
enabled. The problem was that the macro expander didn't consider that it was
making progress when we were deducing that attributes should be lumped in with
custom derive invocations.
The fix applied here was to track in the expander if our attribute is changing
(getting stashed away elsewhere and replaced with a new invocation). If it is
swapped then it's considered progress, otherwise behavior should remain the
same.
Rollup merge of #52505 - alexcrichton:remove-thinlto-hack, r=nikomatsakis
rustc: Remove a workaround in ThinLTO fixed upstream
This commit removes a hack in our ThinLTO passes which removes available
externally functions manually. The [upstream bug][1] has long since been fixed,
so we should be able to rely on LLVM natively for this now!
Rollup merge of #52502 - RalfJung:rotate, r=scottmcm
fix unsafety: don't call ptr_rotate for ZST
`rotate::ptr_rotate` has a comment saying
```
/// # Safety
///
/// The specified range must be valid for reading and writing.
/// The type `T` must have non-zero size.
```
So we better make sure we don't call it on ZST...
Cc @scottmcm (author of https://github.com/rust-lang/rust/pull/41670)
Auto merge of #52354 - QuietMisdreavus:rustdoc-lints, r=GuillaumeGomez
stabilize lint handling in rustdoc
When https://github.com/rust-lang/rust/pull/51732 added CLI flags to manipulate lints in rustdoc, they were added as unstable flags. This made sense as they were new additions, but since they mirrored the flags that rustc has, it's worth considering them to not need an unstable period.
Stabilizing them also provides the opportunity for a critical fix: allowing Cargo to pass `--cap-lints allow` when documenting dependencies, the same as when it compiles them.