]> git.lizzy.rs Git - rust.git/log
rust.git
7 years agoRollup merge of #41442 - tbu-:pr_writeall_interrupted, r=aturon
Corey Farwell [Wed, 26 Apr 2017 03:05:55 +0000 (23:05 -0400)]
Rollup merge of #41442 - tbu-:pr_writeall_interrupted, r=aturon

Specify behavior of `write_all` for `ErrorKind::Interrupted` errors

Also spell out that read and write operations should be retried on
`ErrorKind::Interrupted` errors.

Fixes #38494.

7 years agoRollup merge of #39983 - AndrewGaspar:rust-windbg, r=brson
Corey Farwell [Wed, 26 Apr 2017 03:05:54 +0000 (23:05 -0400)]
Rollup merge of #39983 - AndrewGaspar:rust-windbg, r=brson

Adds rust-windbg.cmd script

Adds rust-gdb/rust-lldb equivalent for windbg that loads the Rust .natvis files on start.

This change modifies the bootstrap code to add rust-windbg to bin and the .natvis files to lib/rustlib/etc.

Example usage from cmd or PowerShell:
```
rust-windbg -c "bu rs_f442289d74765418!rs::main;g" target\debug\rs.exe
```

7 years agoAuto merge of #41332 - petrochenkov:privti, r=eddyb
bors [Tue, 25 Apr 2017 23:46:37 +0000 (23:46 +0000)]
Auto merge of #41332 - petrochenkov:privti, r=eddyb

Check privacy of trait items in all contexts

Fixes https://github.com/rust-lang/rust/issues/28514

This is a sufficiently rare scenario and it's currently guarded by `private_in_public` lint, so it shouldn't be a [breaking-change] in practice.

7 years agoFix object safety violations in the test
Vadim Petrochenkov [Tue, 25 Apr 2017 19:38:21 +0000 (22:38 +0300)]
Fix object safety violations in the test

7 years agoprivacy: Rename and cleanup PrivacyVisitor
Vadim Petrochenkov [Sun, 16 Apr 2017 18:19:24 +0000 (21:19 +0300)]
privacy: Rename and cleanup PrivacyVisitor

7 years agoCheck privacy of trait items in all contexts
Vadim Petrochenkov [Sun, 16 Apr 2017 16:06:26 +0000 (19:06 +0300)]
Check privacy of trait items in all contexts

7 years agoAuto merge of #41362 - alexcrichton:run-cargot-ests, r=aturon
bors [Tue, 25 Apr 2017 15:51:40 +0000 (15:51 +0000)]
Auto merge of #41362 - alexcrichton:run-cargot-ests, r=aturon

Run tests for the cargo submodule in tree

Previously the `cargotest` suite would run some arbitrary revision of Cargo's
test suite, but now that we're bundling it in tree we should be running the
Cargo submodule's test suite instead.

7 years agoAuto merge of #40434 - mattico:splice-update, r=alexcrichton
bors [Tue, 25 Apr 2017 10:34:07 +0000 (10:34 +0000)]
Auto merge of #40434 - mattico:splice-update, r=alexcrichton

Implement Vec::splice and String::splice (RFC 1432)

RFC: rust-lang/rfcs#1432, tracking issue: #32310
A rebase of https://github.com/rust-lang/rust/pull/32355 with a few more tests.

Let me know if you have any ideas for more tests.

cc @SimonSapin

7 years agoAdds rust-windbg.cmd script
Andrew Gaspar [Mon, 20 Feb 2017 19:10:32 +0000 (11:10 -0800)]
Adds rust-windbg.cmd script

7 years agoAuto merge of #40123 - TimNN:llvm40, r=alexcrichton
bors [Mon, 24 Apr 2017 22:18:16 +0000 (22:18 +0000)]
Auto merge of #40123 - TimNN:llvm40, r=alexcrichton

LLVM 4.0 Upgrade

Since nobody has done this yet, I decided to get things started:

**Todo:**

* [x] push the relevant commits to `rust-lang/llvm` and `rust-lang/compiler-rt`
* [x] cleanup `.gitmodules`
* [x] Verify if there are any other commits from `rust-lang/llvm` which need backporting
* [x] Investigate / fix debuginfo ("`<optimized out>`") failures
* [x] Use correct emscripten version in docker image

---

Closes #37609.

---

**Test results:**

Everything is green ðŸŽ‰

7 years agoRun tests for the cargo submodule in tree
Alex Crichton [Tue, 18 Apr 2017 00:24:05 +0000 (17:24 -0700)]
Run tests for the cargo submodule in tree

Previously the `cargotest` suite would run some arbitrary revision of Cargo's
test suite, but now that we're bundling it in tree we should be running the
Cargo submodule's test suite instead.

7 years agoAdd Splice forget test
Matt Ickstadt [Mon, 24 Apr 2017 14:49:29 +0000 (09:49 -0500)]
Add Splice forget test

7 years agoAuto merge of #40634 - GuillaumeGomez:refcell-docs, r=steveklabnik
bors [Mon, 24 Apr 2017 13:09:35 +0000 (13:09 +0000)]
Auto merge of #40634 - GuillaumeGomez:refcell-docs, r=steveklabnik

Add more explanation on RefCell::get_mut

Fixes #40203.

r? @rust-lang/docs

7 years agoAdd more explanation on RefCell::get_mut
Guillaume Gomez [Sat, 18 Mar 2017 14:33:56 +0000 (15:33 +0100)]
Add more explanation on RefCell::get_mut

7 years agoAuto merge of #41494 - withoutboats:associated-consts-are-not-object-safe, r=eddyb
bors [Mon, 24 Apr 2017 10:06:41 +0000 (10:06 +0000)]
Auto merge of #41494 - withoutboats:associated-consts-are-not-object-safe, r=eddyb

Associated consts are not object safe.

fixes #26847

r? @eddyb

7 years agoFeature gate in test.
Without Boats [Mon, 24 Apr 2017 09:42:36 +0000 (02:42 -0700)]
Feature gate in test.

7 years agoStyle.
Without Boats [Mon, 24 Apr 2017 08:40:17 +0000 (01:40 -0700)]
Style.

7 years agoFix type error.
Without Boats [Mon, 24 Apr 2017 08:20:36 +0000 (01:20 -0700)]
Fix type error.

7 years agoAdd compile-fail test.
Without Boats [Mon, 24 Apr 2017 08:19:12 +0000 (01:19 -0700)]
Add compile-fail test.

7 years agoAssociated consts are not object safe.
Without Boats [Mon, 24 Apr 2017 05:00:09 +0000 (22:00 -0700)]
Associated consts are not object safe.

7 years agoAdd splice to the unstable book.
Matt Ickstadt [Sat, 8 Apr 2017 21:58:47 +0000 (16:58 -0500)]
Add splice to the unstable book.

7 years agoUse Vec::splice impl in string::Splice::drop()
Matt Ickstadt [Sat, 8 Apr 2017 21:43:30 +0000 (16:43 -0500)]
Use Vec::splice impl in string::Splice::drop()

7 years agoImprove splice docs and tests
Matt Ickstadt [Sat, 8 Apr 2017 21:04:30 +0000 (16:04 -0500)]
Improve splice docs and tests

7 years agoUpdate splice impl
Matt Ickstadt [Sat, 8 Apr 2017 21:12:58 +0000 (16:12 -0500)]
Update splice impl

7 years agoAdd Vec::splice and String::splice
Simon Sapin [Sat, 8 Apr 2017 20:55:53 +0000 (15:55 -0500)]
Add Vec::splice and String::splice

7 years agoAuto merge of #41486 - arielb1:select-where, r=eddyb
bors [Sun, 23 Apr 2017 23:06:50 +0000 (23:06 +0000)]
Auto merge of #41486 - arielb1:select-where, r=eddyb

traits::select: quickly filter out predicates from other traits

this improves most pre-trans passes's performance by ~1%.

That missed the spring cleaning PR because I wanted to ship it.

r? @eddyb

7 years agoAuto merge of #41485 - arielb1:dtorck-constraint, r=eddyb
bors [Sun, 23 Apr 2017 20:11:50 +0000 (20:11 +0000)]
Auto merge of #41485 - arielb1:dtorck-constraint, r=eddyb

cache dtorck constraints on ADTs

This avoids visiting the fields of all structs multiple times, improving item-bodies checking time by 10% (!).

Not sure whether we want this in 1.18 or 1.19. It's a big-ish patch, but the 10% win is very tempting.

r? @eddyb

7 years agocache ADT dtorck results
Ariel Ben-Yehuda [Sun, 23 Apr 2017 14:43:23 +0000 (17:43 +0300)]
cache ADT dtorck results

This avoids visiting the fields of all structs multiple times, improving
item-bodies checking time by 10% (!).

7 years agotraits::select: quickly filter out predicates from other traits
Ariel Ben-Yehuda [Sun, 23 Apr 2017 20:05:25 +0000 (23:05 +0300)]
traits::select: quickly filter out predicates from other traits

this improves most pre-trans passes's performance by ~1%.

7 years agoFIN: windows-gnu: statically link gcc_s, pthread with llvm
Tim Neumann [Thu, 13 Apr 2017 20:27:53 +0000 (22:27 +0200)]
FIN: windows-gnu: statically link gcc_s, pthread with llvm

7 years agoFIN: Compile LLVM with -fno-omit-frame-pointer on 32bit MinGW builds
Tim Neumann [Mon, 17 Apr 2017 08:24:33 +0000 (10:24 +0200)]
FIN: Compile LLVM with -fno-omit-frame-pointer on 32bit MinGW builds

to work around an apparently bad optimization.

7 years agoFIN: switch appveryor to mingw with posix threads
Tim Neumann [Fri, 14 Apr 2017 09:27:56 +0000 (11:27 +0200)]
FIN: switch appveryor to mingw with posix threads

7 years agoFIN: ignore failing test on emscripten, see #41299
Tim Neumann [Fri, 14 Apr 2017 09:29:05 +0000 (11:29 +0200)]
FIN: ignore failing test on emscripten, see #41299

7 years agoFIN: update appveyor mingw description
Tim Neumann [Fri, 14 Apr 2017 09:27:35 +0000 (11:27 +0200)]
FIN: update appveyor mingw description

7 years agoFIN: build comiler-rt wihout Thumb on arm
Tim Neumann [Sun, 2 Apr 2017 10:36:20 +0000 (12:36 +0200)]
FIN: build comiler-rt wihout Thumb on arm

7 years agoFIN: disable backtrace printing for panic-runtime/abort* on ARM
Tim Neumann [Sun, 2 Apr 2017 09:24:22 +0000 (11:24 +0200)]
FIN: disable backtrace printing for panic-runtime/abort* on ARM

7 years agoFIN: update emscripten builder
Tim Neumann [Mon, 20 Mar 2017 16:42:25 +0000 (17:42 +0100)]
FIN: update emscripten builder

7 years agoFIN/CRT: adopt all applicable rust compiler-rt patches
Tim Neumann [Sun, 23 Apr 2017 19:58:44 +0000 (21:58 +0200)]
FIN/CRT: adopt all applicable rust compiler-rt patches

7 years agoFIN/LLVM: new rust patches required for 4.0
Tim Neumann [Sun, 23 Apr 2017 18:50:28 +0000 (20:50 +0200)]
FIN/LLVM: new rust patches required for 4.0

7 years agoFIN/LLVM: adopt all applicable rust llvm patches
Tim Neumann [Sun, 23 Apr 2017 18:21:02 +0000 (20:21 +0200)]
FIN/LLVM: adopt all applicable rust llvm patches

7 years agoFIN: update to upstream (llvm 4.0.1 + fastcomp 1.37.10) & compiler-rt 4.0
Tim Neumann [Mon, 20 Mar 2017 16:35:57 +0000 (17:35 +0100)]
FIN: update to upstream (llvm 4.0.1 + fastcomp 1.37.10) & compiler-rt 4.0

7 years agoclean-up adt_sized_constraint now that it uses on-demand
Ariel Ben-Yehuda [Fri, 21 Apr 2017 17:24:32 +0000 (20:24 +0300)]
clean-up adt_sized_constraint now that it uses on-demand

7 years agoAuto merge of #41408 - eddyb:poly-const-eval, r=arielb1
bors [Sun, 23 Apr 2017 08:35:00 +0000 (08:35 +0000)]
Auto merge of #41408 - eddyb:poly-const-eval, r=arielb1

rustc: generalize monomorphic_const_eval to polymorphic constants.

With the addition of `Substs` to the query key, we can now evaluate *and cache* polymorphic constants.

Fixes #23898 by replacing the crippled explicit-discriminant-only local-crate-only `lookup_variant_by_id` with `ConstVal::Variant` which can describe variants irrespective of their discriminant.

Fixes #41394 by fixing #23898 (for the original testcase) and by not looping past the first discriminant.

7 years agorustc_const_eval: support all unit enum variants.
Eduard-Mihai Burtescu [Wed, 19 Apr 2017 12:16:19 +0000 (15:16 +0300)]
rustc_const_eval: support all unit enum variants.

7 years agorustc_const_eval: CallOn isn't needed, typeck/const-qualif handle those cases.
Eduard-Mihai Burtescu [Wed, 19 Apr 2017 12:15:12 +0000 (15:15 +0300)]
rustc_const_eval: CallOn isn't needed, typeck/const-qualif handle those cases.

7 years agorustc: make the const-eval cache polymorphic.
Eduard-Mihai Burtescu [Tue, 18 Apr 2017 20:38:15 +0000 (23:38 +0300)]
rustc: make the const-eval cache polymorphic.

7 years agoAuto merge of #41437 - cuviper:remove-unstable-deprecated, r=alexcrichton
bors [Sun, 23 Apr 2017 02:13:55 +0000 (02:13 +0000)]
Auto merge of #41437 - cuviper:remove-unstable-deprecated, r=alexcrichton

Remove items that are unstable and deprecated

This removes unstable items that have been deprecated for more than one cycle.

- Since 1.16.0, `#![feature(enumset)]`
    - All of `mod collections::enum_set`
- Since 1.15.0, `#![feature(borrow_state)]`
    - `cell::BorrowState`
    - `RefCell::borrow_state()`
- Since 1.15.0, `#![feature(is_unique)]`
    - `Rc::is_unique()` (made private like `Arc::is_unique()`)
- Since 1.15.0, `#![feature(rc_would_unwrap)]`
    - `Rc::would_wrap()`
- Since 1.13.0, `#![feature(binary_heap_extras)]`
    - `BinaryHeap::push_pop()`
    - `BinaryHeap::replace()`
- Since 1.12.0, `#![feature(as_unsafe_cell)]`
    - `Cell::as_unsafe_cell()`
    - `RefCell::as_unsafe_cell()`
- Since 1.12.0, `#![feature(map_entry_recover_keys)]`
    - `btree_map::OccupiedEntry::remove_pair()`
    - `hash_map::OccupiedEntry::remove_pair()`
- Since 1.11.0, `#![feature(float_extras)]`
    - `Float::nan()`
    - `Float::infinity()`
    - `Float::neg_infinity()`
    - `Float::neg_zero()`
    - `Float::zero()`
    - `Float::one()`
    - `Float::integer_decode()`
    - `f32::integer_decode()`
    - `f32::ldexp()`
    - `f32::frexp()`
    - `f32::next_after()`
    - `f64::integer_decode()`
    - `f64::ldexp()`
    - `f64::frexp()`
    - `f64::next_after()`
- Since 1.11.0, `#![feature(zero_one)]`
    - `num::Zero`
    - `num::One`

7 years agoRemove unused import.
Corey Farwell [Sun, 23 Apr 2017 02:13:22 +0000 (22:13 -0400)]
Remove unused import.

7 years agoAuto merge of #41469 - arielb1:rustc-spring-cleaning, r=eddyb
bors [Sat, 22 Apr 2017 20:07:01 +0000 (20:07 +0000)]
Auto merge of #41469 - arielb1:rustc-spring-cleaning, r=eddyb

Performance audit, Spring 2017

Fix up some quite important performance "surprises" I've found running callgrind on rustc.

7 years agobail out of selection when there are multiple surviving candidates
Ariel Ben-Yehuda [Sat, 22 Apr 2017 12:05:18 +0000 (15:05 +0300)]
bail out of selection when there are multiple surviving candidates

In some cases (e.g. <[int-var] as Add<[int-var]>>), selection can turn up
a large number of candidates. Bailing out early avoids O(n^2) performance.

This improves item-type checking time by quite a bit, resulting in ~2% of total
time-to-typeck.

7 years agoadd a cache to impl_polarity
Ariel Ben-Yehuda [Thu, 20 Apr 2017 17:37:02 +0000 (20:37 +0300)]
add a cache to impl_polarity

this is another one of these things that looks *much* worse on valgrind.

7 years agoallocate less strings in `symbol_names`
Ariel Ben-Yehuda [Thu, 20 Apr 2017 15:38:35 +0000 (18:38 +0300)]
allocate less strings in `symbol_names`

this improves trans performance by *another* 10%.

7 years agoweak_lang_items: check for `lang` attribute before calling `value_str`
Ariel Ben-Yehuda [Thu, 20 Apr 2017 14:17:13 +0000 (17:17 +0300)]
weak_lang_items: check for `lang` attribute before calling `value_str`

improves trans performance by *another* 10%.

7 years agocache attributes of items from foreign crates
Ariel Ben-Yehuda [Thu, 20 Apr 2017 12:08:41 +0000 (15:08 +0300)]
cache attributes of items from foreign crates

this avoids parsing item attributes on each call to `item_attrs`, which takes
off 33% (!) of translation time and 50% (!) of trans-item collection time.

7 years agoshort-cut SharedCrateContext::layout_of
Ariel Ben-Yehuda [Thu, 20 Apr 2017 00:14:39 +0000 (03:14 +0300)]
short-cut SharedCrateContext::layout_of

That method is *incredibly* hot, so this ends up saving 10% of trans
time.

BTW, we really should be doing dependency tracking there - and possibly be
taking the respective perf hit (got to find a way to make DTMs fast), but
`layout_cache` is a non-dep-tracking map.

7 years agoavoid calling `mk_region` unnecessarily
Ariel Ben-Yehuda [Wed, 19 Apr 2017 22:58:12 +0000 (01:58 +0300)]
avoid calling `mk_region` unnecessarily

this improves typeck & trans performance by 1%. This looked hotter on
callgrind than it is on a CPU.

7 years agoremove cleanup branches to the resume block
Ariel Ben-Yehuda [Wed, 19 Apr 2017 22:20:50 +0000 (01:20 +0300)]
remove cleanup branches to the resume block

This improves LLVM performance by 10% lost during the shimmir transition.

7 years agoAuto merge of #41431 - GuillaumeGomez:hoedown-default, r=aturon
bors [Sat, 22 Apr 2017 17:37:04 +0000 (17:37 +0000)]
Auto merge of #41431 - GuillaumeGomez:hoedown-default, r=aturon

Re-enable hoedown by default

r? @rust-lang/docs

cc @aturon

7 years agoAuto merge of #41464 - frewsxcv:rollup, r=frewsxcv
bors [Sat, 22 Apr 2017 14:17:36 +0000 (14:17 +0000)]
Auto merge of #41464 - frewsxcv:rollup, r=frewsxcv

Rollup of 3 pull requests

- Successful merges: #41077, #41355, #41450
- Failed merges:

7 years agoRollup merge of #41450 - rillian:rustc-wrapper, r=alexcrichton
Corey Farwell [Sat, 22 Apr 2017 13:02:58 +0000 (09:02 -0400)]
Rollup merge of #41450 - rillian:rustc-wrapper, r=alexcrichton

Update cargo for 1.18.0.

Bump our associated cargo to pick up the RUSTC_WRAPPER feature
for use with build caches.

7 years agoRollup merge of #41355 - nikomatsakis:incr-comp-refactor-trans-2, r=eddyb
Corey Farwell [Sat, 22 Apr 2017 13:02:57 +0000 (09:02 -0400)]
Rollup merge of #41355 - nikomatsakis:incr-comp-refactor-trans-2, r=eddyb

Refactor trans some more to pave way for incremental compilation

Various refactorings paving the way for the newer approach to incremental compilation (And, in particular, to "query-ifying" trans). My partial goal is to remove `SharedCrateContext`; this PR gets about as far as I can easily get before starting to really want the red/green algorithm.

r? @eddyb
cc @michaelwoerister

7 years agoRollup merge of #41077 - petrochenkov:boundparen, r=nikomatsakis
Corey Farwell [Sat, 22 Apr 2017 13:02:56 +0000 (09:02 -0400)]
Rollup merge of #41077 - petrochenkov:boundparen, r=nikomatsakis

syntax: Support parentheses around trait bounds

An implementation for https://github.com/rust-lang/rust/issues/39318#issuecomment-290956826

r? @nikomatsakis

7 years agoFix line display
Guillaume Gomez [Sat, 22 Apr 2017 12:56:36 +0000 (14:56 +0200)]
Fix line display

7 years agoAuto merge of #39999 - bitshifter:struct_align, r=eddyb
bors [Sat, 22 Apr 2017 11:50:40 +0000 (11:50 +0000)]
Auto merge of #39999 - bitshifter:struct_align, r=eddyb

Implementation of repr struct alignment RFC 1358.

The main changes around rustc::ty::Layout::struct:
* Added abi_align field which stores abi alignment before repr align is applied
* align field contains transitive repr alignment
* Added padding vec which stores padding required after fields

The main user of this information is rustc_trans::adt::struct_llfields
which determines the LLVM fields to be used by LLVM, including padding
fields.

A possible future optimisation would be to put the padding Vec in an Option, since it will be unused unless you are using repr align.

7 years agoFix tests
Guillaume Gomez [Fri, 21 Apr 2017 16:41:47 +0000 (18:41 +0200)]
Fix tests

7 years agoFix invalid linkage
Guillaume Gomez [Fri, 21 Apr 2017 14:09:28 +0000 (16:09 +0200)]
Fix invalid linkage

7 years agoSet hoedown to generate error index
Guillaume Gomez [Fri, 21 Apr 2017 10:27:48 +0000 (12:27 +0200)]
Set hoedown to generate error index

7 years agoRe-enable hoedown by default
Guillaume Gomez [Thu, 20 Apr 2017 22:32:23 +0000 (00:32 +0200)]
Re-enable hoedown by default

7 years agoAuto merge of #41459 - frewsxcv:rollup, r=frewsxcv
bors [Sat, 22 Apr 2017 03:31:53 +0000 (03:31 +0000)]
Auto merge of #41459 - frewsxcv:rollup, r=frewsxcv

Rollup of 4 pull requests

- Successful merges: #37658, #41405, #41432, #41435
- Failed merges:

7 years agoRollup merge of #41435 - estebank:issue-33884, r=nikomatsakis
Corey Farwell [Sat, 22 Apr 2017 03:29:15 +0000 (23:29 -0400)]
Rollup merge of #41435 - estebank:issue-33884, r=nikomatsakis

Add test for issue 33884

Fix #33884.

r=nikomatsakis

7 years agoRollup merge of #41432 - abonander:issue_41211, r=jseyfried
Corey Farwell [Sat, 22 Apr 2017 03:29:15 +0000 (23:29 -0400)]
Rollup merge of #41432 - abonander:issue_41211, r=jseyfried

Don't panic if an attribute macro fails to resolve at crate root

Adds temporary regression test; this ideally should work as-is (#41430)

Closes #41211

r? @jseyfried

7 years agoRollup merge of #41405 - GuillaumeGomez:rustdoc-warning-improvement, r=steveklabnik
Corey Farwell [Sat, 22 Apr 2017 03:29:14 +0000 (23:29 -0400)]
Rollup merge of #41405 - GuillaumeGomez:rustdoc-warning-improvement, r=steveklabnik

Fix line display for hoedown

Fixes #41401.

r? @rust-lang/docs

7 years agoRollup merge of #37658 - GuillaumeGomez:ref_suggestion, r=nikomatsakis,eddyb
Corey Farwell [Sat, 22 Apr 2017 03:29:13 +0000 (23:29 -0400)]
Rollup merge of #37658 - GuillaumeGomez:ref_suggestion, r=nikomatsakis,eddyb

Ref suggestion

7 years agomove the uses of the trans caches into rustc::traits
Niko Matsakis [Sat, 22 Apr 2017 01:02:14 +0000 (21:02 -0400)]
move the uses of the trans caches into rustc::traits

This makes these routines more readily available for other bits of
code. It also will help when refactoring.

7 years agoUse primitive align for tagged enum fill.
Cameron Hart [Fri, 21 Apr 2017 23:38:23 +0000 (09:38 +1000)]
Use primitive align for tagged enum fill.

Hopefully will fix assert on ARM where vector types are being used as
the fill type for enums containing repr aligned types greater than the
largest possible native type, thus don't match the Layout's alignment
and triggers an assert.

7 years agouse Symbol/InternedString in the cache
Niko Matsakis [Fri, 21 Apr 2017 21:26:25 +0000 (17:26 -0400)]
use Symbol/InternedString in the cache

7 years agomove the trans trait caches into tcx
Niko Matsakis [Mon, 17 Apr 2017 16:35:53 +0000 (12:35 -0400)]
move the trans trait caches into tcx

Arguably these could become custom queries, but I chose not to do that
because the relationship of queries and trait system is not yet fleshed
out enough. For now it seems fine to have them be `DepTrackingMap` using
the memoize pattern.

7 years agosort `provide`
Niko Matsakis [Mon, 17 Apr 2017 13:08:34 +0000 (09:08 -0400)]
sort `provide`

7 years agomake `reachable_set` ref-counted
Niko Matsakis [Fri, 14 Apr 2017 19:33:52 +0000 (15:33 -0400)]
make `reachable_set` ref-counted

Once it is computed, no need to deep clone the set.

7 years agojust take `tcx` where we can
Niko Matsakis [Fri, 14 Apr 2017 19:30:06 +0000 (15:30 -0400)]
just take `tcx` where we can

The more we can things dependent on just tcx, the easier it will
be to make queries etc later on.

7 years agointroduce the rather simpler symbol-cache in place of symbol-map
Niko Matsakis [Fri, 14 Apr 2017 19:10:53 +0000 (15:10 -0400)]
introduce the rather simpler symbol-cache in place of symbol-map

The symbol map is not good for incremental: it has inputs from every fn
in existence, and it will change if anything changes. One could imagine
cheating with the symbol-map and exempting it from the usual dependency
tracking, since the results are fully deterministic. Instead, I opted to
just add a per-CGU cache, on the premise that recomputing some symbol
names is not going to be so very expensive.

7 years agoremove `translation_items` from `SharedCrateContext`
Niko Matsakis [Fri, 14 Apr 2017 17:27:39 +0000 (13:27 -0400)]
remove `translation_items` from `SharedCrateContext`

If we are going to hash `SharedCrateContext`, we don't want a list of
things that pertain to **every CGU** in there.

7 years agointroduce `is_foreign_item` query
Niko Matsakis [Fri, 14 Apr 2017 14:51:22 +0000 (10:51 -0400)]
introduce `is_foreign_item` query

This may seem like overkill, but it's exactly what we want/need for
incremental compilation I think. In particular, while generating code
for some codegen unit X, we can wind up querying about any number of
external items, and we only want to be forced to rebuild X is some of
those changed from a foreign item to otherwise. Factoring this into a
query means we would re-run only if some `false` became `true` (or vice
versa).

7 years agoUpdate ui test
Guillaume Gomez [Fri, 21 Apr 2017 11:28:31 +0000 (13:28 +0200)]
Update ui test

7 years agoUpdate cargo.
Ralph Giles [Fri, 21 Apr 2017 18:40:16 +0000 (11:40 -0700)]
Update cargo.

Bump our associated cargo to pick up the RUSTC_WRAPPER feature
for use with build caches.

7 years agoMove parse_remaining_bounds into a separate function
Vadim Petrochenkov [Fri, 21 Apr 2017 18:32:11 +0000 (21:32 +0300)]
Move parse_remaining_bounds into a separate function

7 years agoFix issue with single question mark or paren
Vadim Petrochenkov [Thu, 20 Apr 2017 21:19:06 +0000 (00:19 +0300)]
Fix issue with single question mark or paren

7 years agosyntax: Support parentheses around trait bounds
Vadim Petrochenkov [Tue, 4 Apr 2017 22:12:53 +0000 (01:12 +0300)]
syntax: Support parentheses around trait bounds

7 years agoAuto merge of #41445 - frewsxcv:rollup, r=frewsxcv
bors [Fri, 21 Apr 2017 16:36:41 +0000 (16:36 +0000)]
Auto merge of #41445 - frewsxcv:rollup, r=frewsxcv

Rollup of 4 pull requests

- Successful merges: #41372, #41376, #41426, #41429
- Failed merges:

7 years agoCreate a new method to run coercion inside probe
Guillaume Gomez [Wed, 9 Nov 2016 23:25:45 +0000 (00:25 +0100)]
Create a new method to run coercion inside probe

7 years agoAuto merge of #41349 - eddyb:ty-contents, r=nikomatsakis
bors [Fri, 21 Apr 2017 14:02:37 +0000 (14:02 +0000)]
Auto merge of #41349 - eddyb:ty-contents, r=nikomatsakis

rustc: replace TypeContents with two independent properties (is_freeze / needs_drop).

`InteriorUnsafe` / `interior_unsafe` was replaced with a private lang-item `Freeze` auto trait in libcore.

`OwnsDtor` / `needs_drop` was replaced with a specialized traversal that *doesn't* avoid caching results in case of a cycle, as the only cycles left can only occur in erroneous "types with infinite sizes", references and raw pointers not having destructors. Also, `Copy` is now checked at every step of the recursion.

r? @nikomatsakis

7 years agoUpdate #[no_core] users with the "freeze" lang item.
Eduard-Mihai Burtescu [Tue, 18 Apr 2017 08:02:21 +0000 (11:02 +0300)]
Update #[no_core] users with the "freeze" lang item.

7 years agoRollup merge of #41429 - jonathandturner:update_rls_submod, r=alexcrichton
Corey Farwell [Fri, 21 Apr 2017 12:40:36 +0000 (08:40 -0400)]
Rollup merge of #41429 - jonathandturner:update_rls_submod, r=alexcrichton

Bump the rls submodule revision

Bump the RLS version to include a few recent fixes.

7 years agoRollup merge of #41426 - malbarbo:android-x86_64, r=alexcrichton
Corey Farwell [Fri, 21 Apr 2017 12:40:35 +0000 (08:40 -0400)]
Rollup merge of #41426 - malbarbo:android-x86_64, r=alexcrichton

Add x86_64-linux-android target

7 years agoRollup merge of #41376 - mbrubeck:docs, r=frewsxcv
Corey Farwell [Fri, 21 Apr 2017 12:40:34 +0000 (08:40 -0400)]
Rollup merge of #41376 - mbrubeck:docs, r=frewsxcv

Expanded docs and examples for PathBuf::file_name and friends

This addresses some common surprises when `PathBuf::set_file_name` is called on the path of a directory rather than a file.

r? @steveklabnik

7 years agoRollup merge of #41372 - nbigaouette:master, r=alexcrichton
Corey Farwell [Fri, 21 Apr 2017 12:40:33 +0000 (08:40 -0400)]
Rollup merge of #41372 - nbigaouette:master, r=alexcrichton

Use an (over-writable) environment variable for the `gdb` command

Instead of hard-coding the command to run, using the environment
variable `GDB_CMD` (that defaults to `gdb`) allows using a different
debugger than the default `gdb` executable.

This gives the possibility to use `cgdb` as the debugger, which provides
a nicer user interface. Note that one has to use `GDB_CMD="cgdb --"` to
use cgdb (note the trailing `--`) to let cgdb pass the proper arguments
to `gdb`.

7 years agoUpdate tests
Guillaume Gomez [Tue, 8 Nov 2016 21:55:58 +0000 (22:55 +0100)]
Update tests

7 years agoAdd suggestion for & coercions
Guillaume Gomez [Tue, 8 Nov 2016 21:54:31 +0000 (22:54 +0100)]
Add suggestion for & coercions

7 years agoSpecify behavior of `write_all` for `ErrorKind::Interrupted` errors
Tobias Bucher [Fri, 21 Apr 2017 08:32:13 +0000 (10:32 +0200)]
Specify behavior of `write_all` for `ErrorKind::Interrupted` errors

Also spell out that read and write operations should be retried on
`ErrorKind::Interrupted` errors.

Fixes #38494.

7 years agoAuto merge of #41245 - estebank:multiline-trim, r=nikomatsakis
bors [Fri, 21 Apr 2017 06:59:25 +0000 (06:59 +0000)]
Auto merge of #41245 - estebank:multiline-trim, r=nikomatsakis

Reduce visual clutter of multiline start when possible

When a span starts on a line with nothing but whitespace to the left,
and there are no other annotations in that line, simplify the visual
representation of the span.

Go from:

```rust
error[E0072]: recursive type `A` has infinite size
 --> file2.rs:1:1
  |
1 |   struct A {
  |  _^ starting here...
2 | |     a: A,
3 | | }
  | |_^ ...ending here: recursive type has infinite size
  |
```

To:

```rust
error[E0072]: recursive type `A` has infinite size
 --> file2.rs:1:1
  |
1 | / struct A {
2 | |     a: A,
3 | | }
  | |_^ recursive type has infinite size
```

Re: #38246.

r? @nikomatsakis CC @jonathandturner