Jonathan Turner [Wed, 31 Aug 2016 20:53:35 +0000 (13:53 -0700)]
Rollup merge of #36129 - eddyb:signal-exit-status, r=alexcrichton
Fix run-pass/signal-exit-status to not trigger UB by writing to NULL.
`run-pass/signal-exit-status` has had UB (NULL dereference) since it was introduced in #10109.
Fixes the test failure found by @camlorn while running under Windows Subsystem for Linux.
Jonathan Turner [Wed, 31 Aug 2016 20:53:34 +0000 (13:53 -0700)]
Rollup merge of #36114 - zjhmale:fix-E0393, r=jonathandturner
Update E0393 to new error format
Fixes #35632.
Part of #35233.
r? @jonathandturner
and a wired thing is that if i add another label
```rust
.span_label(span, &format!("missing reference to `{}`", def.name))
.span_label(span, &format!("because of the default `Self` reference, type parameters must be specified on object types"))
```
and add a new note in the test case like
```rust
trait A<T=Self> {}
fn together_we_will_rule_the_galaxy(son: &A) {}
//~^ ERROR E0393
//~| NOTE missing reference to `T`
//~| NOTE because of the default `Self` reference, type parameters must be specified on object types
```
it will complain that
```
running 1 test
test [compile-fail] compile-fail/E0393.rs ... FAILED
error: /Users/zjh/Documents/rustspace/rust/src/test/compile-fail/E0393.rs:13: unexpected "error": '13:43: 13:44: the type parameter `T` must be explicitly specified [E0393]'
unexpected errors (from JSON output): [
Error {
line_num: 13,
kind: Some(
Error
),
msg: "13:43: 13:44: the type parameter `T` must be explicitly specified [E0393]"
}
]
```
it is a little bit confusing and through the blog post we can use `//~^` and `//~|` to support multiple notes, @jonathandturner am i missing something here?
Jonathan Turner [Wed, 31 Aug 2016 13:29:11 +0000 (06:29 -0700)]
Rollup merge of #36148 - birryree:E0194_bonus_format, r=jonathandturner
Bonus format for E0194
Bonus fix for #35280. Part of #35233. Fixes #36057. Adding expanded notes/context for what trait a parameter shadows as part of E0194 error messages.
Errors for E0194 now look like this:
```
$> ./build/x86_64-apple-darwin/stage1/bin/rustc src/test/compile-fail/E0194.rs
error[E0194]: type parameter `T` shadows another type parameter of the same name
--> src/test/compile-fail/E0194.rs:13:26
|
11 | trait Foo<T> { //~ NOTE first `T` declared here
| - first `T` declared here
12 | fn do_something(&self) -> T;
13 | fn do_something_else<T: Clone>(&self, bar: T);
| ^ shadows another type parameter
Jonathan Turner [Wed, 31 Aug 2016 13:29:08 +0000 (06:29 -0700)]
Rollup merge of #35926 - matthew-piziak:bit-or-xor-examples, r=GuillaumeGomez
add evocative examples for `BitOr` and `BitXor`
These are exactly equivalent to PR #35809, with one caveat: I do not believe there is a non-bitwise binary XOR operator in Rust, so here it's expressed as `(a || b) && !(a && b)`.
Alternative decompositions are `(a && !b) || (!a && b)` and `(a || b) && (!a || !b)`. Let me know if you think one of those would be clearer.
Jonathan Turner [Wed, 31 Aug 2016 13:29:08 +0000 (06:29 -0700)]
Rollup merge of #35758 - matthew-piziak:vec-assert-over-println-remaining, r=GuillaumeGomez
accumulate vector and assert for RangeFrom and RangeInclusive examples
PR #35695 for `Range` was merged, so it seems that this side-effect-free style is preferred for Range* examples. This PR performs the same translation for `RangeFrom` and `RangeInclusive`. It also removes what looks to be an erroneously commented line for `#![feature(step_by)]`, and an unnecessary primitive-type annotation in `0u8..`.
bors [Tue, 30 Aug 2016 23:28:32 +0000 (16:28 -0700)]
Auto merge of #35048 - tmiasko:monotonic-wait-timeout, r=alexcrichton
Use monotonic time in condition variables.
Configure condition variables to use monotonic time using
pthread_condattr_setclock on systems where this is possible.
This fixes the issue when thread waiting on condition variable is
woken up too late when system time is moved backwards.
Matthew Piziak [Tue, 23 Aug 2016 15:23:42 +0000 (11:23 -0400)]
add evocative examples for `BitOr` and `BitXor`
These are exactly equivalent to PR #35809, with one caveat: I do not believe there is a non-bitwise binary "xor" operator in Rust, so here it's expressed as (a || b) && !(a && b).
Guillaume Gomez [Tue, 30 Aug 2016 08:39:05 +0000 (10:39 +0200)]
Rollup merge of #35810 - matthew-piziak:fn-trait-example, r=steveklabnik
improve documentation for `Fn*` traits
This PR is not yet a serious attempt at contribution. Rather, I'm opening this for discussion. I can think of a few things we may want to accomplish with the documentation of the `Fn`, `FnMut`, and `FnOnce` traits:
- the relationship between these traits and the closures that implement them
- examples of non-closure implementations
- the relationship between these traits and Rust's ownership semantics
bors [Tue, 30 Aug 2016 07:36:19 +0000 (00:36 -0700)]
Auto merge of #36066 - jseyfried:rollup, r=Manishearth
Batch up libsyntax breaking changes
Batch of the following syntax-[breaking-change] changes:
- #35591: Add a field `span: Span` to `ast::Generics`.
- #35618: Remove variant `Mod` of `ast::PathListItemKind` and refactor the remaining variant `ast::PathListKind::Ident` to a struct `ast::PathListKind_`.
- #35480: Change uses of `Constness` in the AST to `Spanned<Constness>`.
- c.f. `MethodSig`, `ItemKind`
- #35728: Refactor `cx.pat_enum()` into `cx.pat_tuple_struct()` and `cx.pat_path()`.
- #35850: Generalize the elements of lists in attributes from `MetaItem` to a new type `NestedMetaItem` that can represent a `MetaItem` or a literal.
- #35917: Remove traits `AttrMetaMethods`, `AttributeMethods`, and `AttrNestedMetaItemMethods`.
- Besides removing imports of these traits, this won't cause fallout.
- Add a variant `Union` to `ItemKind` to future proof for `union` (c.f. #36016).
- Remove inherent methods `attrs` and `fold_attrs` of `Annotatable`.
- Use methods `attrs` and `map_attrs` of `HasAttrs` instead.
Fixes #36023 by backporting @majnemer's LLVM patch fixing [the LLVM bug](https://llvm.org/bugs/show_bug.cgi?id=29163.) where SimplifyCFG hoisted instructions andkept their metadata (conditional `!nonnull` loads could kill a null check later if hoisted).
bors [Mon, 29 Aug 2016 04:57:51 +0000 (21:57 -0700)]
Auto merge of #36062 - japaric:smarter-submodules, r=alexcrichton
rustbuild: smarter `git submodule`-ing
With this commit, if one bootstraps rust against system llvm then the
src/llvm submodule is not updated/checked-out. This saves considerable
network bandwith when starting from a fresh clone of rust-lang/rust as
the llvm submodule is never cloned.
cc #30107
r? @alexcrichton
cc @petevine
~~We could also avoid updating the jemalloc submodule if --disable-jemalloc is used. It just hasn't been implemented.~~ Done
This probably doesn't handle "recursive" submodules correctly but I think we don't have any of those right now.
I'm still testing a bootstrap but already confirmed that the llvm submodule doesn't get updated when `--llvm-root` is passed to `configure`.
bors [Sun, 28 Aug 2016 17:19:52 +0000 (10:19 -0700)]
Auto merge of #35984 - jonas-schievink:reproducible-builds, r=eddyb
Steps towards reproducible builds
cc #34902
Running `make dist` twice will result in a rustc tarball where only `librustc_back.so`, `librustc_llvm.so` and `librustc_trans.so` differ. Building `libstd` and `libcore` twice with the same compiler and flags produces identical artifacts.