bors [Sat, 14 Jan 2017 19:47:23 +0000 (19:47 +0000)]
Auto merge of #38992 - nagisa:i128-minvallit, r=eddyb
Fix two const-eval issues related to i128 negation
First issue here was the fact that we’d only allow negating integers in i64 range in case the
integer was not infered yes. While this is not the direct cause of the issue, its still good to fix
it.
The real issue here is the code handling specifically the `min_value` literals. While I128_OVERFLOW
has the expected value (0x8000_..._0000), match using this value as a pattern is handled
incorrectly by the stage1 compiler (it seems to be handled correctly, by the stage2 compiler). So
what we do here is extract this pattern into an explicit `==` until the next snapshot.
bors [Sat, 14 Jan 2017 17:33:46 +0000 (17:33 +0000)]
Auto merge of #38982 - clarcharr:expect_err, r=aturon
expect_err for Result.
This adds an `expect_err` method to `Result`. Considering how `unwrap_err` already exists, this seems to make sense. Inconsistency noted in Manishearth/rust-clippy#1435.
bors [Sat, 14 Jan 2017 15:28:11 +0000 (15:28 +0000)]
Auto merge of #38952 - nrc:save-impl-fix, r=eddyb
save-analysis: handle paths in type/trait context more correctly
TBH, this is still not perfect, witness the FIXME, but it is an improvement. In particular it means we get information about trait references in impls.
bors [Sat, 14 Jan 2017 13:17:16 +0000 (13:17 +0000)]
Auto merge of #38944 - michaelwoerister:incr-generics-partitioning, r=nikomatsakis
trans: Treat generics like regular functions, not like #[inline] function, during CGU partitioning
This PR makes generics be treated just like regular functions during CGU partitioning:
+ the function instantiation is placed in a codegen unit based on the function's DefPath,
+ unless it is marked with `#[inline]` -- which causes a private copy of the function to be placed in every referencing codegen unit.
This has the following effects:
+ Multi codegen unit builds will become faster because code for generic functions is duplicated less.
+ Multi codegen unit builds might have lower runtime performance, since generics are not available for inlining automatically any more.
+ Single codegen unit builds are not affected one way or the other.
This partitioning scheme is particularly good for incremental compilation as it drastically reduces the number of false positives during codegen unit invalidation.
I'd love to have a benchmark suite for estimating the effect on runtime performance for changes like this one.
bors [Sat, 14 Jan 2017 08:48:28 +0000 (08:48 +0000)]
Auto merge of #38927 - petrochenkov:leven, r=jseyfried
resolve: Levenshtein-based suggestions for non-import paths
This patch addresses both items from https://github.com/rust-lang/rust/issues/30197#issuecomment-264846000 and therefore implements the largest part of https://github.com/rust-lang/rust/issues/30197.
bors [Sat, 14 Jan 2017 06:43:03 +0000 (06:43 +0000)]
Auto merge of #38914 - est31:tidy-gate-tests, r=nikomatsakis
Make tidy check for lang gate tests
Add gate tests to the checks that tidy performs. Excerpt from the commit message of the main commit:
Require compile-fail tests for new lang features
Its non trivial to test lang feature gates, and people
forget to add such tests. So we extend the features lint
of the tidy tool to ensure that all new lang features
contain a new compile-fail test.
Of course, one could drop this requirement and just
grep all tests in run-pass for #![feature(abc)] and
then run this test again, removing the mention,
requiring that it fails.
But this only tests for the existence of a compilation
failure. Manual tests ensure that also the correct lines
spawn the error, and also test the actual error message.
For library features, it makes no sense to require such
a test, as here code is used that is generic for all
library features.
The tidy lint extension now checks the compile-fail test suite for occurences of "gate-test-X" where X is a feature. Alternatively, it also accepts file names with the form "feature-gate-X.rs". If a lang feature is found that has no such check, we emit a tidy error.
I've applied the markings to all tests I could find in the test suite. I left a small (20 elements) whitelist of features that right now have no gate test, or where I couldn't find one. Once this PR gets merged, I'd like to close issue #22820 and open a new one on suggestion of @nikomatsakis to track the removal of all elements from that whitelist (already have a draft). Writing such a small test can be a good opportunity for a first contribution, so I won't touch it (let others have the fun xD).
cc @brson , @pnkfelix (they both discussed about this in the issue linked above).
bors [Sat, 14 Jan 2017 04:29:44 +0000 (04:29 +0000)]
Auto merge of #39021 - alexcrichton:try-debug-travis, r=brson
travis: Attempt to debug OSX linker segfaults
This commit attempts to debug the segfaults that we've been seeing on OSX on
Travis. I have no idea what's going on here mostly, but let's try to look at
core dumps and get backtraces to see what's going on. This commit itself is
mostly a complete shot in the dark, I'm not sure if this even works...
LLVM usually prefers using memcpys over direct loads/store of first
class aggregates. The check in type_is_immediate to mark certain small
structs was originally part of the code to handle such immediates in
function arguments, and it had a counterpart in load_ty/store_ty to
actually convert small aggregates to integers.
But since then, the ABI handling has been refactored and takes care of
converting small aggregates to integers. During that refactoring, the
code to handle small aggregates in load_ty/store_ty has been removed,
and so we accidentally started using loads/stores on FCA values.
Since type_is_immediate() is no longer responsible for ABI-related
conversions, and using a memcpy even for small aggregates is usually
better than performing a FCA load/store, we can remove that code part
and only handle simple types as immediates there.
bors [Fri, 13 Jan 2017 19:26:49 +0000 (19:26 +0000)]
Auto merge of #39036 - aidanhs:aphs-robust-docker, r=alexcrichton
Remove strictly-unnecessary flags for docker
cc #39035
In addition to `--tty` I've removed `--interactive` as I don't think there's any reason for it to be there (it only hooks up stdin, which shouldn't be used anyway).
If this looks like it's working over a few days then I'll also alter the libc scripts.
bors [Fri, 13 Jan 2017 16:59:25 +0000 (16:59 +0000)]
Auto merge of #38890 - petrochenkov:noresolve, r=nrc
resolve: Do not use "resolve"/"resolution" in error messages
Use less jargon-y wording instead.
`cannot find <struct> <S> in <this scope>` and `cannot find <struct> <S> in <module a::b>` are used for base messages (this also harmonizes nicely with "you can import it into scope" suggestions) and `not found in <this scope>` and `not found in <a::b>` are used for short labels in fall-back case.
I tweaked some other diagnostics to avoid using "resolve" (see, e.g., `librustc_resolve/macros.rs`), but haven't touched messages for imports.
Guillaume Gomez [Fri, 13 Jan 2017 09:42:31 +0000 (10:42 +0100)]
Rollup merge of #38994 - BenWiederhake:master, r=sanxiyn
Fix some typos in Nomicon
I waited a bit before creating this PR in case I find more typos – I didn't.
I've read `CONTRIBUTING.md` but didn't `make check`, and `make doc` takes incredibly long. (Among other things, `make doc` builds llvm from scratch. Not sure if that's intentional.)
bors [Fri, 13 Jan 2017 02:56:20 +0000 (02:56 +0000)]
Auto merge of #38748 - alexcrichton:upload, r=brson
travis: Start uploading artifacts on commits
This commit starts adding the infrastructure for uploading release artifacts
from AppVeyor/Travis on each commit. The idea is that eventually we'll upload a
full release to AppVeyor/Travis in accordance with plans [outlined earlier].
Right now this configures Travis/Appveyor to upload all tarballs in the `dist`
directory, and various images are updated to actually produce tarballs in these
directories. These are nowhere near ready to be actual release artifacts, but
this should allow us to play around with it and test it out. Once this commit
lands we should start seeing artifacts uploaded on each commit.
Alex Crichton [Sun, 1 Jan 2017 01:42:40 +0000 (17:42 -0800)]
travis: Start uploading artifacts on commits
This commit starts adding the infrastructure for uploading release artifacts
from AppVeyor/Travis on each commit. The idea is that eventually we'll upload a
full release to AppVeyor/Travis in accordance with plans [outlined earlier].
Right now this configures Travis/Appveyor to upload all tarballs in the `dist`
directory, and various images are updated to actually produce tarballs in these
directories. These are nowhere near ready to be actual release artifacts, but
this should allow us to play around with it and test it out. Once this commit
lands we should start seeing artifacts uploaded on each commit.
Alex Crichton [Thu, 12 Jan 2017 22:58:55 +0000 (14:58 -0800)]
travis: Attempt to debug OSX linker segfaults
This commit attempts to debug the segfaults that we've been seeing on OSX on
Travis. I have no idea what's going on here mostly, but let's try to look at
core dumps and get backtraces to see what's going on. This commit itself is
mostly a complete shot in the dark, I'm not sure if this even works...
est31 [Sun, 8 Jan 2017 02:24:40 +0000 (03:24 +0100)]
Require compile-fail tests for new lang features
Its non trivial to test lang feature gates, and people
forget to add such tests. So we extend the features lint
of the tidy tool to ensure that all new lang features
contain a new compile-fail test.
Of course, one could drop this requirement and just
grep all tests in run-pass for #![feature(abc)] and
then run this test again, removing the mention,
requiring that it fails.
But this only tests for the existence of a compilation
failure. Manual tests ensure that also the correct lines
spawn the error, and also test the actual error message.
For library features, it makes no sense to require such
a test, as here code is used that is generic for all
library features.
bors [Thu, 12 Jan 2017 20:44:02 +0000 (20:44 +0000)]
Auto merge of #38814 - Ralith:cfg-fields, r=jseyfried
syntax: enable attributes and cfg on struct fields
This enables conditional compilation of field initializers in a struct literal, simplifying construction of structs whose fields are themselves conditionally present. For example, the intializer for the constant in the following becomes legal, and has the intuitive effect:
It's not clear to me whether this calls for the full RFC process, but the implementation was simple enough that I figured I'd begin the conversation with code.
bors [Thu, 12 Jan 2017 18:32:28 +0000 (18:32 +0000)]
Auto merge of #38569 - chris-morgan:rustdoc-highlight-kw-2, r=steveklabnik
Fix rustdoc highlighting of `&` and `*`
Whitespace tokens were included, so the span check used with `&` was incorrect, and it was never highlighted as kw-2 (RefKeyword).
The `*` in `*foo` and `*const T` should also be highlighted kw-2, so I added them. Note that this *will* cause mishighlighting of code like `1*2`, but that should have been written `1 * 2`. Same deal with `1&2`.
bors [Thu, 12 Jan 2017 14:42:02 +0000 (14:42 +0000)]
Auto merge of #38779 - Craig-Macomber:bench, r=alexcrichton
Do not run outer setup part of benchmarks multiple times to fix issue 20142
Fix #20142
This is my first real rust code, so I expect the quality is quite bad. Please let me know in which ways it is horrible and I'll fix it.
Previously the whole benchmark function was rerun many times, but with this change, only the callback passed to iter is rerun. This improves performances by saving benchmark startup time. The setup used to be called a minimum of 101 times, and now only runs once.
I wasn't sure exactly what should be done for the case where iter is never called, so I left a FIXME for that: currently it does not error, and I added tests to cover that.
I have left the algorithm and statistics unchanged: I don't like how the minimum number of runs is 301 (that's bad for very slow benchmarks) but I consider such changes out of scope for this fix.
bors [Thu, 12 Jan 2017 08:31:50 +0000 (08:31 +0000)]
Auto merge of #38654 - alexcrichton:rustbuild-destdir, r=brson
rustbuild: Implement DESTDIR support
This commit primarily starts supporting the `DESTDIR` environment variable like
the old build system. Along the way this brings `config.toml` up to date with
support in `config.mk` with install options supported.
bors [Thu, 12 Jan 2017 03:13:48 +0000 (03:13 +0000)]
Auto merge of #38867 - alexcrichton:ignore-test-on-windows, r=brson
std: Ignore close_read_wakes_up on Windows
It looks like in practice at least this test will not pass on Windows.
Empirically it is prone to blocking forever, presumably because a call to
`shutdown` doesn't actually wake up other threads on Windows.
We don't document this as a guarantee for `shutdown`, nor do we internally rely
on it. This test originated in a time long since passed when it was leveraged
for canceling I/O, but nowadays there's nothing fancy happening in the standard
library so it's not really a productive test anyway, hence just ignoring it on
Windows.
bors [Wed, 11 Jan 2017 20:12:26 +0000 (20:12 +0000)]
Auto merge of #38313 - jseyfried:self_imports, r=nrc
resolve: clean up the semantics of `self` in an import list
Change `self` in an import list `use foo::bar::{self, ...};` to import `bar` only in the type namespace. Today, `bar` is imported in every namespace in which `foo::bar` is defined.
This is a [breaking-change], see https://github.com/rust-lang/rust/issues/38293#issue-194817974 for examples of code that would break.
Fix two const-eval issues related to i128 negation
First issue here was the fact that we’d only allow negating integers in i64 range in case the
integer was not infered yes. While this is not the direct cause of the issue, its still good to fix
it.
The real issue here is the code handling specifically the `min_value` literals. While I128_OVERFLOW
has the expected value (0x8000_..._0000), match using this value as a pattern is handled
incorrectly by the stage1 compiler (it seems to be handled correctly, by the stage2 compiler). So
what we do here is extract this pattern into an explicit `==` until the next snapshot.
bors [Wed, 11 Jan 2017 13:42:52 +0000 (13:42 +0000)]
Auto merge of #38984 - alexcrichton:less-debuginfo, r=brson
rustbuild: Don't enable debuginfo in rustc
In #37280 we enabled line number debugging information in release artifacts,
primarily to close out #36452 where debugging information was critical for MSVC
builds of Rust to be useful in production. This commit, however, apparently had
some unfortunate side effects.
Namely it was noticed in #37477 that if `RUST_BACKTRACE=1` was set then any
compiler error would take a very long time for the compiler to exit. The cause
of the problem here was somewhat deep:
* For all compiler errors, the compiler will `panic!` with a known value. This
tears down the main compiler thread and allows cleaning up all the various
resources. By default, however, this panic output is suppressed for "normal"
compiler errors.
* When `RUST_BACKTRACE=1` was set this caused every compiler error to generate a
backtrace.
* The libbacktrace library hits a pathological case where it spends a very long
time in its custom allocation function, `backtrace_alloc`, because the
compiler has so much debugging information. More information about this can be
found in #29293 with a summary at the end of #37477.
To solve this problem this commit simply removes debuginfo from the compiler but
not from the standard library. This should allow us to keep #36452 closed while
also closing #37477. I've measured the difference to be orders of magnitude
faster than it was before, so we should see a much quicker time-to-exit after a
compile error when `RUST_BACKTRACE=1` is set.
bors [Wed, 11 Jan 2017 05:27:11 +0000 (05:27 +0000)]
Auto merge of #38916 - estebank:pad-suggestion-list, r=nikomatsakis
Teach diagnostics to correct margin of multiline messages
Make the suggestion list have a correct padding:
```
error[E0308]: mismatched types
--> file.rs:3:20
|
3 | let x: usize = "";
| ^^ expected usize, found reference
|
= note: expected type `usize`
= note: found type `&'static str`
= help: here are some functions which might fulfill your needs:
- .len()
- .foo()
- .bar()
```
Alex Crichton [Wed, 11 Jan 2017 04:01:54 +0000 (20:01 -0800)]
rustbuild: Don't enable debuginfo in rustc
In #37280 we enabled line number debugging information in release artifacts,
primarily to close out #36452 where debugging information was critical for MSVC
builds of Rust to be useful in production. This commit, however, apparently had
some unfortunate side effects.
Namely it was noticed in #37477 that if `RUST_BACKTRACE=1` was set then any
compiler error would take a very long time for the compiler to exit. The cause
of the problem here was somewhat deep:
* For all compiler errors, the compiler will `panic!` with a known value. This
tears down the main compiler thread and allows cleaning up all the various
resources. By default, however, this panic output is suppressed for "normal"
compiler errors.
* When `RUST_BACKTRACE=1` was set this caused every compiler error to generate a
backtrace.
* The libbacktrace library hits a pathological case where it spends a very long
time in its custom allocation function, `backtrace_alloc`, because the
compiler has so much debugging information. More information about this can be
found in #29293 with a summary at the end of #37477.
To solve this problem this commit simply removes debuginfo from the compiler but
not from the standard library. This should allow us to keep #36452 closed while
also closing #37477. I've measured the difference to be orders of magnitude
faster than it was before, so we should see a much quicker time-to-exit after a
compile error when `RUST_BACKTRACE=1` is set.
bors [Wed, 11 Jan 2017 01:33:03 +0000 (01:33 +0000)]
Auto merge of #38960 - arielb1:blake-endian, r=michaelwoerister
Use little-endian encoding for Blake2 hashing on all architectures
Like many hash functions, the blake2 hash is mathematically defined on
a sequence of 64-bit words. As Rust's hash interface operates on
sequences of octets, some encoding must be used to bridge that
difference.
The Blake2 RFC (RFC 7693) specifies that:
```
Byte (octet) streams are interpreted as words in little-endian order,
with the least-significant byte first.
```
So use that encoding consistently.
Fixes #38891.
Beta-nominating since this is a regression since 1.15.
bors [Tue, 10 Jan 2017 21:16:06 +0000 (21:16 +0000)]
Auto merge of #38843 - Manishearth:proposed, r=GuillaumeGomez
Improved rustdoc rendering for unstable features
This replaces "unstable" with "this is an experimental API", and uses a `<details>` tag to expand to the reason.
The `<details>` tag renders as a regular div (with the details show) on browsers which don't support it, On browsers which do support it, it shows only the summary line with an expandy-arrow next to it, and on clicking it the details will turn up below it.
This is somewhat a strawman proposal. The main issue is that we need to improve our messaging around unstable APIs. Since they turn up in the docs, we should be clearer that they are experimental (and perhaps add something about nightly-only). I'm making this PR to kickstart discussion on this.
Example rendering: http://manishearth.github.io/rust-internals-docs/std/io/trait.Read.html#method.chars
Ariel Ben-Yehuda [Tue, 10 Jan 2017 14:24:06 +0000 (16:24 +0200)]
Use little-endian encoding for Blake2 hashing on all architectures
Like many hash functions, the blake2 hash is mathematically defined on
a sequence of 64-bit words. As Rust's hash interface operates on
sequences of octets, some encoding must be used to bridge that
difference.
The Blake2 RFC (RFC 7693) specifies that:
Byte (octet) streams are interpreted as words in little-endian order,
with the least-significant byte first.
Seo Sanghyeon [Tue, 10 Jan 2017 11:27:48 +0000 (20:27 +0900)]
Rollup merge of #38874 - derekdreery:patch-1, r=steveklabnik
Update vec.rs
Add a warning not to convert char* from c to Vec<u8> (I thought you could until I asked on irc).
Reasoning is that it will help people avoid an error that could cause crashes and undefined behaviour. Only drawback is that it could confuse someone not familiar with C, but beginners are unlikely to be using this function anyway.
Seo Sanghyeon [Tue, 10 Jan 2017 11:27:40 +0000 (20:27 +0900)]
Rollup merge of #38623 - alexcrichton:less-osx-warnings, r=aturon
std: Remove unused objects from compiler-builtins
We don't actually use trampoline_setup.c and all the `*tf3` business
seems related to f80/f128 business. Specifically this'll fix some
warnings showing up during builds on OSX.
bors [Tue, 10 Jan 2017 06:50:59 +0000 (06:50 +0000)]
Auto merge of #38138 - rkruppe:no_std-no_loop, r=steveklabnik
book: use abort() over loop {} for panic
Due to #28728 `loop {}` is very risky and can lead to fun debugging experiences such as #38136. Besides, aborting is probably better behavior than an infinite loop.
Nick Cameron [Tue, 10 Jan 2017 03:09:13 +0000 (16:09 +1300)]
save-analysis: handle paths in type/trait context more correctly
TBH, this is still not perfect, witness the FIXME, but it is an improvement. In particular it means we get information about trait references in impls.