Corey Farwell [Thu, 23 Mar 2017 03:37:59 +0000 (23:37 -0400)]
Rollup merge of #40518 - michaelwoerister:hir-id, r=eddyb
Introduce HirId, a replacement for ast::NodeId after lowering to HIR
This is the first step towards implementing #40303. This PR introduces the `HirId` type and generates a `HirId` for everything that would be assigned one (i.e. stuff in the HIR), but the HIR data types still use `NodeId` for now. Changing that is a big refactoring that I want to do in a separate PR.
A `HirId` uniquely identifies a node in the HIR of the current crate. It is composed of the `owner`, which is the `DefIndex` of the directly enclosing `hir::Item`, `hir::TraitItem`, or `hir::ImplItem` (i.e. the closest "item-like"), and the `local_id` which is unique within the given owner.
This PR is also running a number of consistency checks for the generated `HirId`s:
- Does `NodeId` in the HIR have a corresponding `HirId`?
- Is the `owner` part of each `HirId` consistent with its position in the HIR?
- Do the numerical values of the `local_id` part all lie within a dense range of integers?
Corey Farwell [Thu, 23 Mar 2017 03:37:58 +0000 (23:37 -0400)]
Rollup merge of #39891 - shepmaster:emit-mir, r=nikomatsakis
Teach rustc --emit=mir
I'm opening this PR to discuss:
1. Is this a good idea?
1. Is this a good implementation?
I'm sure people will have opinions on both points!
This spawned from https://github.com/rust-lang/rust/issues/31847#issuecomment-279179057, so I figured a prototype implementation could help provide a seed to talk about.
Corey Farwell [Wed, 22 Mar 2017 23:30:35 +0000 (19:30 -0400)]
Rollup merge of #40732 - petrochenkov:booktidy, r=steveklabnik
Update the book submodule and fix tidy
When the book was included into https://github.com/rust-lang/rust as a submodule, tidy started failing on Windows.
https://github.com/rust-lang/book/pull/549 fixed the problem, now the submodule needs to be updated.
Corey Farwell [Wed, 22 Mar 2017 23:30:31 +0000 (19:30 -0400)]
Rollup merge of #40704 - omtcyfz:clang_version_bump, r=alexcrichton
Nit: LLVM & Clang latest version is 4.0
Small nit: since latest Clang version is 4.0 it's nice to reflect this in the documentation.
Also, I couldn't find anything, but there might be any hard-coded check that Clang version matches "3.X" anywhere in the build system; if there is one, it'd be great to bump that one too.
Corey Farwell [Wed, 22 Mar 2017 23:30:30 +0000 (19:30 -0400)]
Rollup merge of #40692 - SamWhited:consistent_str_docs_punctuation, r=bstrie
str: Make docs consistently punctuated
Every so slightly pointless one character PR, but this was driving me nuts while reading the docs a moment ago (all the [other public structs](https://doc.rust-lang.org/std/str/index.html#structs) have descriptions that end in a full-stop).
Corey Farwell [Wed, 22 Mar 2017 23:30:26 +0000 (19:30 -0400)]
Rollup merge of #40578 - michaelwoerister:shorter-compiletest-stamps, r=alexcrichton
Make the filenames of .stamp files generated by compiletest shorter
Otherwise we run into filename length limitations on some file systems. See https://bugs.launchpad.net/ecryptfs/+bug/344878 for an example where we only can have ~145 characters for filenames.
Corey Farwell [Wed, 22 Mar 2017 23:30:25 +0000 (19:30 -0400)]
Rollup merge of #40548 - alexcrichton:appveyor-ninja, r=brson
appveyor: Use Ninja to build LLVM on MinGW
I have a suspicion that MinGW's make is the cause of #40546 rather than anything
else, but that's purely a suspicion without any facts to back it up. In any case
we'll eventually be moving the MSVC build over to Ninja in order to leverage
sccache regardless, so this commit simply jumpstarts that process by downloading
Ninja for use by MinGW anyway.
I'm not sure if this closes #40546 for real, but this is my current best shot at
closing it out, so...
Corey Farwell [Wed, 22 Mar 2017 23:30:23 +0000 (19:30 -0400)]
Rollup merge of #40509 - jseyfried:duplicate_check_macro_exports, r=nrc
Forbid conflicts between macros 1.0 exports and macros 2.0 exports
This PR forbids for conflicts between `#[macro_export]`/`#[macro_reexport]` macro exports and `pub use` macro exports. For example,
```rust
// crate A:
pub use macros::foo;
//^ This is allowed today, will be forbidden by this PR.
// crate B:
extern crate A; // This triggers a confusing error today.
use A::foo; // This could refer to refer to either macro export in crate A.
```
Introduce HirId, a replacement for NodeId after lowering to HIR.
HirId has a more stable representation than NodeId, meaning that
modifications to one item don't influence (part of) the IDs within
other items. The other part is a DefIndex for which there already
is a way of stable hashing and persistence.
This commit introduces the HirId type and generates a HirId for
every NodeId during HIR lowering, but the resulting values are
not yet used anywhere, except in consistency checks.
Alex Crichton [Wed, 15 Mar 2017 14:35:35 +0000 (07:35 -0700)]
appveyor: Use Ninja to build LLVM on MinGW
I have a suspicion that MinGW's make is the cause of #40546 rather than anything
else, but that's purely a suspicion without any facts to back it up. In any case
we'll eventually be moving the MSVC build over to Ninja in order to leverage
sccache regardless, so this commit simply jumpstarts that process by downloading
Ninja for use by MinGW anyway.
I'm not sure if this closes #40546 for real, but this is my current best shot at
closing it out, so...
bors [Wed, 22 Mar 2017 02:00:16 +0000 (02:00 +0000)]
Auto merge of #40043 - petrochenkov:objpars, r=nikomatsakis
Refactor parsing of trait object types
Bugs are fixed and code is cleaned up.
User visible changes:
- `ty` matcher in macros accepts trait object types like `Write + Send` (https://github.com/rust-lang/rust/issues/39080)
- Buggy priority of `+` in trait object types starting with `for` is fixed (https://github.com/rust-lang/rust/issues/39317). `&for<'a> Trait<'a> + Send` is now parsed as `(&for<'a> Trait<'a>) + Send` and requires parens `&(for<'a> Trait<'a> + Send)`. For comparison, `&Send + for<'a> Trait<'a>` was parsed like this since [Nov 27, 2014](https://github.com/rust-lang/rust/pull/19298).
- Trailing `+`s are supported in trait objects, like in other bounds.
- Better error reporting for trait objects starting with `?Sized`.
Fixes https://github.com/rust-lang/rust/issues/39080
Fixes https://github.com/rust-lang/rust/issues/39317 [breaking-change]
Closes https://github.com/rust-lang/rust/issues/39298
cc https://github.com/rust-lang/rust/issues/39085 (fixed, then reverted https://github.com/rust-lang/rust/pull/40043#issuecomment-286570653)
cc https://github.com/rust-lang/rust/issues/39318 (fixed, then reverted https://github.com/rust-lang/rust/pull/40043#issuecomment-284493890)
Jeffrey Seyfried [Tue, 14 Mar 2017 05:16:54 +0000 (05:16 +0000)]
Check for conflicts between macros 1.0 exports (`#[macro_export]`, `#[macro_reexport]`)
and macros 2.0 exports (`pub use` macro re-exports and `pub macro` (once implemented)
at the crate root.
bors [Tue, 21 Mar 2017 19:50:17 +0000 (19:50 +0000)]
Auto merge of #40601 - stjepang:sort-unstable, r=alexcrichton
Implement feature sort_unstable
Tracking issue for the feature: #40585
This is essentially integration of [pdqsort](https://github.com/stjepang/pdqsort) into libcore.
There's plenty of unsafe blocks to review. The heart of pdqsort is `fn partition_in_blocks` and is probably the most challenging function to understand. It requires some patience, but let me know if you find it too difficult - comments could always be improved.
#### Changes
* Added `sort_unstable` feature.
* Tweaked insertion sort constants for stable sort. Sorting integers is now up to 5% slower, but sorting big elements is much faster (in particular, `sort_large_big_random` is 35% faster). The old constants were highly optimized for sorting integers, so overall the configuration is more balanced now. A minor regression in case of integers is forgivable as we recently had performance improvements (#39538) that completely make up for it.
* Removed some uninteresting sort benchmarks.
* Added a new sort benchmark for string sorting.
Interesting cases:
* Expensive comparison function and string sorting - it's a really close race, but timsort performs a slightly smaller number of comparisons. This is a natural difference of bottom-up merging versus top-down partitioning.
* `large_descending` - unstable sort is faster, but both sorts should have equivalent performance. Both just check whether the slice is descending and if so, they reverse it. I blame LLVM for the discrepancy.
bors [Tue, 21 Mar 2017 14:35:18 +0000 (14:35 +0000)]
Auto merge of #40686 - alexcrichton:less-logging, r=brson
travis: Don't set `RUST_LOG` globally
I have a suspicion that this caused a large regression in cycle times by forcing
the compiler to perform more checks on every `debug!` statement, so let's test
this out by removing the `RUST_LOG` env var globally.
This regression in cycle time was witnessed between [two] [builds] where the
[PR] in question didn't do much suspicious. Judging by how the stage0 times
*also* regressed though then this is my best guess.
Corey Farwell [Tue, 21 Mar 2017 03:45:00 +0000 (23:45 -0400)]
Rollup merge of #40576 - dwrensha:mir-terminator-kind-doc-typo, r=nikomatsakis
fix innacuracy in mir TerminatorKind::SwitchInt docs
Each index of `values` corresponds to an index of `targets`, and `targets` additionally has a "default case" element at its end, so `targets.len() == values.len() + 1`, not the other way around. For example, [here](https://github.com/rust-lang/rust/blob/0aeb9c12979e6da753701a798d04105b6b1a8c28/src/librustc/mir/mod.rs#L549-L550) is a concrete instance of `SwitchInt` being constructed with `targets.len() == 2` and `values.len() == 1`.
Corey Farwell [Tue, 21 Mar 2017 03:44:57 +0000 (23:44 -0400)]
Rollup merge of #40332 - steveklabnik:extract-book, r=alexcrichton
Extract book into a submodule
Part of https://github.com/rust-lang/rust/issues/39588
We probably don't want to land this till after the beta branches on friday, but would still ❤️ a review from @alexcrichton , since I am a rustbuild noob.
This pr:
1. removes the book
2. adds it back in as a submodule
3. the submodule includes both the old book and the new book
4. it also includes an index page explaining the difference in editions
5. it also includes redirect pages for the old book URLs.
6. so we build all that stuff too.
Alex Crichton [Mon, 20 Mar 2017 21:29:14 +0000 (14:29 -0700)]
travis: Don't set `RUST_LOG` globally
I have a suspicion that this caused a large regression in cycle times by forcing
the compiler to perform more checks on every `debug!` statement, so let's test
this out by removing the `RUST_LOG` env var globally.
This regression in cycle time was witnessed between [two] [builds] where the
[PR] in question didn't do much suspicious. Judging by how the stage0 times
*also* regressed though then this is my best guess.
bors [Mon, 20 Mar 2017 05:36:36 +0000 (05:36 +0000)]
Auto merge of #40281 - jimmycuadra:try-from-from-str, r=aturon
Rename TryFrom's associated type and implement str::parse using TryFrom.
Per discussion on the tracking issue, naming `TryFrom`'s associated type `Error` is generally more consistent with similar traits in the Rust ecosystem, and what people seem to assume it should be called. It also helps disambiguate from `Result::Err`, the most common "Err".
See https://github.com/rust-lang/rust/issues/33417#issuecomment-269108968.
`TryFrom<&str>` and `FromStr` are equivalent, so have the latter provide the former to ensure that. Using `TryFrom` in the implementation of `str::parse` means types that implement either trait can use it. When we're ready to stabilize `TryFrom`, we should update `FromStr` to
suggest implementing `TryFrom<&str>` instead for new code.
See https://github.com/rust-lang/rust/issues/33417#issuecomment-277175994
and https://github.com/rust-lang/rust/issues/33417#issuecomment-277253827.
Corey Farwell [Mon, 20 Mar 2017 00:51:10 +0000 (20:51 -0400)]
Rollup merge of #40554 - nrc:rls-data, r=alexcrichton
Use rls-data crate
This basically pulls out a bunch of data structures used by save-analysis for serialization into an external crate, and pulls that crate in using Rustbuild. The RLS can then share these data structures with the compiler which in some cases will allow more efficient communication between the compiler and the RLS (i.e., without serialisation).
Along the way, I have to pull in rls-span, which is the RLS's way of defining spans (more type-safe than the compiler's built-in way). This is basically just to convert from compiler spans to RLS spans.
I also pull in the crates.io version of rustc-serialize, which is a bit annoying, but seems to be the only way to have serialisable data in an external crate. To make this work, all of the save-analysis crate has to use this version too (cc #40527).
Finally I pull in a line from #40347 to make the unstable crate checking stuff working.
There are a lot of changes to save-analysis but they are all mechanical and trivial - changing from using `From` to `Into` (because of orphan rules) being the main thing.
Corey Farwell [Mon, 20 Mar 2017 00:51:09 +0000 (20:51 -0400)]
Rollup merge of #40532 - jseyfried:improve_tokenstream_quoter, r=nrc
macros: improve the `TokenStream` quoter
This PR
- renames the `TokenStream` quoter from `qquote!` to `quote!`,
- uses `$` instead of `unquote` (e.g. `let toks: TokenStream = ...; quote!([$toks])`),
- allows unquoting `Token`s as well as `TokenTree`s and `TokenStream`s (fixes #39746), and
- to preserve syntactic space, requires that `$` be followed by
- a single identifier to unquote, or
- another `$` to produce a literal `$`.
Corey Farwell [Mon, 20 Mar 2017 00:51:08 +0000 (20:51 -0400)]
Rollup merge of #40521 - TimNN:panic-free-shift, r=alexcrichton
Implemente overflowing_sh* with new unchecked_sh* intrinsics
Also update some 128 bit builtins to not rely on the constant evaluator to avoid checked operations.
Fixes #40508.
cc @nagisa, @alexcrichton
Note: I still have a build running to see if the 128 bit changes worked (unoptimized builds take *forever* to compile), however at least the overflowing builtins no longer reference `core::panicking::panic`.
Corey Farwell [Mon, 20 Mar 2017 00:51:07 +0000 (20:51 -0400)]
Rollup merge of #40398 - eddyb:struct-hint, r=nikomatsakis
Propagate expected type hints through struct literals.
Partial fix for #31260 to maximize backwards-compatibility, i.e. the hint is provided but not coerced to.
The added test works because `{...; x}` with a hint of `T` coerces `x` to `T`, and the reasoning why that is slightly different has to do with DSTs: `&Struct { tail: [x] }: &Struct<[T]>` has a hint of `[T]` for `[x]`, but the inferred type should be `[T; 1]` to succeed later, so `[x]` shouldn't be *forced* to be `[T]`.
*However*, implementing that complete behavior in a backwards-compatible way may be non-trivial, and has not yet been fully investigated, while this PR fixes #40355 and can be backported.
Corey Farwell [Mon, 20 Mar 2017 00:51:06 +0000 (20:51 -0400)]
Rollup merge of #40281 - jimmycuadra:try-from-from-str, r=aturon
Rename TryFrom's associated type and implement str::parse using TryFrom.
Per discussion on the tracking issue, naming `TryFrom`'s associated type `Error` is generally more consistent with similar traits in the Rust ecosystem, and what people seem to assume it should be called. It also helps disambiguate from `Result::Err`, the most common "Err".
See https://github.com/rust-lang/rust/issues/33417#issuecomment-269108968.
`TryFrom<&str>` and `FromStr` are equivalent, so have the latter provide the former to ensure that. Using `TryFrom` in the implementation of `str::parse` means types that implement either trait can use it. When we're ready to stabilize `TryFrom`, we should update `FromStr` to
suggest implementing `TryFrom<&str>` instead for new code.
See https://github.com/rust-lang/rust/issues/33417#issuecomment-277175994
and https://github.com/rust-lang/rust/issues/33417#issuecomment-277253827.
Corey Farwell [Mon, 20 Mar 2017 00:51:06 +0000 (20:51 -0400)]
Rollup merge of #40241 - Sawyer47:fix-39997, r=alexcrichton
Change how the `0` flag works in format!
Now it always implies right-alignment, so that padding zeroes are placed after the sign (if any) and before the digits. In other words, it always takes precedence over explicitly specified `[[fill]align]`. This also affects the '#' flag: zeroes are placed after the prefix (0b, 0o, 0x) and before the digits.
Here's a short summary of how similar format strings work in Python and Rust:
bors [Sun, 19 Mar 2017 21:02:53 +0000 (21:02 +0000)]
Auto merge of #39799 - dpc:create_dir_all, r=alexcrichton
Fix race condition in fs::create_dir_all
The code would crash if the directory was created after create_dir_all
checked whether the directory already existed. This was contrary to
the documentation which claimed to create the directory if it doesn't
exist, implying (but not stating) that there would not be a failure
due to the directory existing.
Corey Farwell [Sun, 19 Mar 2017 14:18:21 +0000 (10:18 -0400)]
Rollup merge of #40621 - jswalden:dependant-spelling-fix, r=sfackler
Fix a spelling error in HashMap documentation, and slightly reword surrounding text for precision
Noticed while reading docs just now.
It's possible that the prior wording *meant* to state that the seed's randomness depends on the exact instant that the system RNG was created, I guess. But unless there's an API guarantee that this is the case, the wording seems over-precise. Is there a formal API guarantee that would forbid, say, the system RNG from generating all output using the Intel RDRAND instruction? I don't think the quality of output in that case would depend on when the RNG was created. Yet it seems to me like it could well be a valid source of randomness when computing the initial seed.
For that reason, tying the randomness of the seed, to the quality of the RNG's output *at the precise instant the seed is computed*, seems less confining. That instantaneous quality level could be determined by the quality at the instant the RNG was created -- but instantaneous quality need not be low for that precise reason.
Corey Farwell [Sun, 19 Mar 2017 14:18:19 +0000 (10:18 -0400)]
Rollup merge of #40603 - QuietMisdreavus:slice-ptr-docs, r=GuillaumeGomez
minor wording tweak to slice::{as_ptr, as_mut_ptr}
Per #37334, the slice-as-pointer methods mentioned that "modifying the slice may cause its buffer to be reallocated", when in fact modifying the *slice* itself would cause no such change. (It is a borrow, after all!) This is a tweak to the wording of that line to stress it's the *collection* that could cause the buffer to be reallocated.