]> git.lizzy.rs Git - rust.git/log
rust.git
5 years agoexplain why pick-constraints can be an empty vector
Niko Matsakis [Wed, 12 Jun 2019 15:47:48 +0000 (11:47 -0400)]
explain why pick-constraints can be an empty vector

5 years agoadd a FIXME related to the non-free-region case
Niko Matsakis [Wed, 12 Jun 2019 15:16:59 +0000 (11:16 -0400)]
add a FIXME related to the non-free-region case

I don't think it would actually be harmful to just ignore such cases
but I'm inclined not to take chances.

5 years agoupdate the async-fn-multiple-lifetimes test
Niko Matsakis [Wed, 12 Jun 2019 14:47:04 +0000 (10:47 -0400)]
update the async-fn-multiple-lifetimes test

5 years agofix tests and remove outdated stderr files
Niko Matsakis [Wed, 12 Jun 2019 14:22:27 +0000 (10:22 -0400)]
fix tests and remove outdated stderr files

5 years agointegrate reverse graph and upper-bound computation
Niko Matsakis [Wed, 12 Jun 2019 14:22:07 +0000 (10:22 -0400)]
integrate reverse graph and upper-bound computation

5 years agoadd a `depth_first_search` helper function
Niko Matsakis [Tue, 11 Jun 2019 20:29:27 +0000 (16:29 -0400)]
add a `depth_first_search` helper function

5 years agointroduce a `VecGraph` abstraction that cheaply stores graphs
Niko Matsakis [Tue, 11 Jun 2019 17:40:24 +0000 (13:40 -0400)]
introduce a `VecGraph` abstraction that cheaply stores graphs

This is perhaps better than the linked list approach I was using
before. Lower memory overhead, Theta(N+E) storage. Does require a
sort. =)

5 years agoimplement the graph traits for SCC
Niko Matsakis [Tue, 11 Jun 2019 12:48:40 +0000 (08:48 -0400)]
implement the graph traits for SCC

5 years agoimprove tests with migration mode, mir mode
Niko Matsakis [Mon, 10 Jun 2019 21:22:02 +0000 (17:22 -0400)]
improve tests with migration mode, mir mode

5 years agoremove old error and add an explanation
Niko Matsakis [Mon, 10 Jun 2019 21:17:56 +0000 (17:17 -0400)]
remove old error and add an explanation

5 years agopreliminary integration of "pick constraints" into nll solver
Niko Matsakis [Fri, 7 Jun 2019 19:57:09 +0000 (15:57 -0400)]
preliminary integration of "pick constraints" into nll solver

5 years agoconstruct pick-constraints and give them to region inference
Niko Matsakis [Wed, 5 Jun 2019 23:33:29 +0000 (19:33 -0400)]
construct pick-constraints and give them to region inference

5 years agoimplement PickConstraintSet type
Niko Matsakis [Wed, 5 Jun 2019 23:07:15 +0000 (19:07 -0400)]
implement PickConstraintSet type

5 years agopass more than outlives constraints to constraint conversion
Niko Matsakis [Wed, 5 Jun 2019 14:21:31 +0000 (10:21 -0400)]
pass more than outlives constraints to constraint conversion

5 years ago[WIP] fix `Lift` impl for `Rc`
Niko Matsakis [Mon, 24 Jun 2019 20:06:17 +0000 (16:06 -0400)]
[WIP] fix `Lift` impl for `Rc`

5 years agorename `ConstraintSet` to `OutlivesConstraintSet`
Niko Matsakis [Wed, 5 Jun 2019 14:17:30 +0000 (10:17 -0400)]
rename `ConstraintSet` to `OutlivesConstraintSet`

5 years agopropagate the pick-constraints through queries
Niko Matsakis [Wed, 5 Jun 2019 09:58:08 +0000 (05:58 -0400)]
propagate the pick-constraints through queries

5 years agopass a `&mut QueryRegionConstraints` not just outlives constraints
Niko Matsakis [Wed, 5 Jun 2019 09:52:58 +0000 (05:52 -0400)]
pass a `&mut QueryRegionConstraints` not just outlives constraints

5 years agoremove deref impl and add an index impl
Niko Matsakis [Tue, 4 Jun 2019 22:05:23 +0000 (18:05 -0400)]
remove deref impl and add an index impl

The constraint set is going to be more than just a set of outlives
constraints.

5 years agorename `ConstraintIndex` to `OutlivesConstraintIndex`
Niko Matsakis [Tue, 4 Jun 2019 21:39:12 +0000 (17:39 -0400)]
rename `ConstraintIndex` to `OutlivesConstraintIndex`

5 years agointroduce `QueryRegionConstraints` struct
Niko Matsakis [Mon, 3 Jun 2019 22:25:21 +0000 (18:25 -0400)]
introduce `QueryRegionConstraints` struct

5 years agointroduce `QueryRegionConstraints` struct (no-op)
Niko Matsakis [Mon, 3 Jun 2019 21:39:52 +0000 (17:39 -0400)]
introduce `QueryRegionConstraints` struct (no-op)

5 years agorename `QueryRegionConstraint` to `QueryOutlivesConstraint`
Niko Matsakis [Mon, 3 Jun 2019 21:24:30 +0000 (17:24 -0400)]
rename `QueryRegionConstraint` to `QueryOutlivesConstraint`

5 years agoenforce and report pick-constraint errors
Niko Matsakis [Mon, 3 Jun 2019 20:31:44 +0000 (16:31 -0400)]
enforce and report pick-constraint errors

The error message here is not great.

5 years agointegrate pick constraints into lexical solver more completely
Niko Matsakis [Mon, 3 Jun 2019 19:54:21 +0000 (15:54 -0400)]
integrate pick constraints into lexical solver more completely

5 years agorename from "in constraint" to "pick constraint"
Niko Matsakis [Mon, 3 Jun 2019 15:58:57 +0000 (11:58 -0400)]
rename from "in constraint" to "pick constraint"

5 years agoadd some tests, currently ICE-ing
Niko Matsakis [Sat, 1 Jun 2019 11:01:49 +0000 (07:01 -0400)]
add some tests, currently ICE-ing

5 years agomake `dup_vec` optional
Niko Matsakis [Fri, 31 May 2019 18:30:06 +0000 (14:30 -0400)]
make `dup_vec` optional

5 years agolexical_region_resolve: rustfmt
Niko Matsakis [Fri, 31 May 2019 18:29:43 +0000 (14:29 -0400)]
lexical_region_resolve: rustfmt

5 years agointroduce an "in" constraint instead of error
Niko Matsakis [Thu, 30 May 2019 14:36:35 +0000 (10:36 -0400)]
introduce an "in" constraint instead of error

5 years agointroduce `constrain_regions` helper
Niko Matsakis [Tue, 28 May 2019 21:02:07 +0000 (17:02 -0400)]
introduce `constrain_regions` helper

5 years agoopaque_types/mod.rs: rustfmt
Niko Matsakis [Fri, 28 Jun 2019 20:47:53 +0000 (16:47 -0400)]
opaque_types/mod.rs: rustfmt

5 years agoUse a single CtxtInterners
John Kåre Alsaker [Fri, 31 May 2019 08:23:22 +0000 (10:23 +0200)]
Use a single CtxtInterners

5 years agoHashMap is UnwindSafe
Simon Sapin [Tue, 2 Jul 2019 13:19:19 +0000 (15:19 +0200)]
HashMap is UnwindSafe

Fixes https://github.com/rust-lang/rust/issues/62301, a regression in 1.36.0 which was caused by hashbrown using `NonZero<T>` where the older hashmap used `Unique<T>`.

5 years agoAuto merge of #61922 - tmandry:moar-generator-optimization, r=matthewjasper
bors [Tue, 2 Jul 2019 12:25:00 +0000 (12:25 +0000)]
Auto merge of #61922 - tmandry:moar-generator-optimization, r=matthewjasper

Don't store locals that have been moved from in generators

This avoids reserving storage in generators for locals that are moved
out of (and not re-initialized) prior to yield points. Fixes #59123.

This adds a new dataflow analysis, `RequiresStorage`, to determine whether the storage of a local can be destroyed without being observed by the program. The rules are:

1. StorageLive(x) => mark x live
2. StorageDead(x) => mark x dead
3. If a local is moved from, _and has never had its address taken_, mark it dead
4. If (any part of) a local is initialized, mark it live'

This is used to determine whether to save a local in the generator object at all, as well as which locals can be overlapped in the generator layout.

Here's the size in bytes of all testcases included in the change, before and after the change:

async fn test    |Size before |Size after
-----------------|------------|----------
single           | 1028       | 1028
single_with_noop | 2056       | 1032
joined           | 5132       | 3084
joined_with_noop | 8208       | 3084

generator test              |Size before |Size after
----------------------------|------------|----------
move_before_yield           | 1028       | 1028
move_before_yield_with_noop | 2056       | 1032
overlap_move_points         | 3080       | 2056

## Future work

Note that there is a possible extension to this optimization, which modifies rule 3 to read: "If a local is moved from, _**and either has never had its address taken, or is Freeze and has never been mutably borrowed**_, mark it dead." This was discussed at length in #59123 and then #61849. Because this would cause some behavior to be UB which was not UB before, it's a step that needs to be taken carefully.

A more immediate priority for me is inlining `std::mem::size_of_val(&x)` so it becomes apparent that the address of `x` is not taken. This way, using `size_of_val` to look at the size of your inner futures does not affect the size of your outer future.

cc @cramertj @eddyb @Matthias247 @nikomatsakis @RalfJung @Zoxc

5 years agorefactor check_for_substitution
Aleksey Kladov [Tue, 2 Jul 2019 08:06:01 +0000 (11:06 +0300)]
refactor check_for_substitution

No behavior change, just flatter and simpler code

5 years agoAuto merge of #61871 - Zoxc:no-lift-branch, r=eddyb
bors [Tue, 2 Jul 2019 08:09:15 +0000 (08:09 +0000)]
Auto merge of #61871 - Zoxc:no-lift-branch, r=eddyb

Don't use lift to detect local types

This overlaps with https://github.com/rust-lang/rust/pull/61392.

r? @eddyb

5 years agoFix import of take in collapse_docs.rs
Chris Gregory [Tue, 2 Jul 2019 03:19:13 +0000 (20:19 -0700)]
Fix import of take in collapse_docs.rs

5 years agoRevert change in compiletest
Chris Gregory [Sun, 30 Jun 2019 21:51:14 +0000 (14:51 -0700)]
Revert change in compiletest

5 years agoEnable mem_take feature in relevant crates
Chris Gregory [Sun, 30 Jun 2019 18:38:06 +0000 (11:38 -0700)]
Enable mem_take feature in relevant crates

5 years agoConvert more usages over
Chris Gregory [Sun, 30 Jun 2019 18:30:01 +0000 (11:30 -0700)]
Convert more usages over

5 years agoUse mem::take instead of mem::replace with default
Chris Gregory [Sun, 30 Jun 2019 18:13:41 +0000 (11:13 -0700)]
Use mem::take instead of mem::replace with default

5 years agoFix michaelwoerister's mailmap
Mark Rousskov [Tue, 2 Jul 2019 00:23:35 +0000 (20:23 -0400)]
Fix michaelwoerister's mailmap

5 years agoAuto merge of #62279 - pietroalbini:fix-azure-crlf, r=alexcrichton
bors [Mon, 1 Jul 2019 22:33:54 +0000 (22:33 +0000)]
Auto merge of #62279 - pietroalbini:fix-azure-crlf, r=alexcrichton

ci: explicitly disable CRLF conversion on Windows

The Azure image enables CRLF conversion on Windows builders, but that caused regressions both in our test suite (the miri test suite broke) and in the ecosystem, since we started shipping install scripts with CRLF endings instead of the old LF. The [Godbolt Compiler Explorer](https://godbolt.org/) is one such case of breakage.

This adds a step to the build explicitly disabling the conversion before the repository is checked out.

r? @alexcrichton
cc @gnzlbg

5 years agomiri realloc: do not require giving old size+align
Ralf Jung [Mon, 1 Jul 2019 21:48:58 +0000 (23:48 +0200)]
miri realloc: do not require giving old size+align

5 years agoci: explicitly disable CRLF conversion on Windows
Pietro Albini [Mon, 1 Jul 2019 19:57:08 +0000 (21:57 +0200)]
ci: explicitly disable CRLF conversion on Windows

The Azure image enables CRLF conversion on Windows builders, but that
caused regressions both in our test suite (the miri test suite broke)
and in the ecosystem, since we started shipping install scripts with
CRLF endings instead of the old LF. The Godbolt Compiler Explorer is one
such case of breakage.

This adds a step to the build explicitly disabling the conversion before
the repository is checked out.

5 years agoClean up extra lifetime, add assertions
Tyler Mandry [Mon, 1 Jul 2019 19:19:21 +0000 (12:19 -0700)]
Clean up extra lifetime, add assertions

5 years agoAuto merge of #62267 - GuillaumeGomez:revert-implicit-option-return, r=Centril
bors [Mon, 1 Jul 2019 18:49:30 +0000 (18:49 +0000)]
Auto merge of #62267 - GuillaumeGomez:revert-implicit-option-return, r=Centril

Revert "implicit `Option`-returning doctests"

Reverts #61279 as discussed in #61360.

cc @Centril

5 years agoAuto merge of #62247 - pietroalbini:finish-azure-migration, r=Mark-Simulacrum
bors [Mon, 1 Jul 2019 15:14:16 +0000 (15:14 +0000)]
Auto merge of #62247 - pietroalbini:finish-azure-migration, r=Mark-Simulacrum

ci: finish the migration to azure

This moves to azure the last builders as we discussed on Discord last night.

r? @alexcrichton

5 years agobump miri to fix line endings on azure
Pietro Albini [Mon, 1 Jul 2019 13:30:27 +0000 (15:30 +0200)]
bump miri to fix line endings on azure

5 years agoUpdate RLS to disable spurious client_find_definitions test
Igor Matuszewski [Sun, 30 Jun 2019 23:36:43 +0000 (01:36 +0200)]
Update RLS to disable spurious client_find_definitions test

Since switching CI to Azure Pipelines it seems that this test seems
to fail more consistently, so let's disable that for now. It helps
that we have less than a week before release - we disallow PRs that
break the tools to land in this period, so this makes landing critical
PRs smoother  now.

r? @alexcrichton

5 years agoci: finish the migration to azure
Pietro Albini [Sun, 30 Jun 2019 17:12:56 +0000 (19:12 +0200)]
ci: finish the migration to azure

5 years agoRevert "implicit `Option`-returning doctests"
Guillaume Gomez [Mon, 1 Jul 2019 14:41:37 +0000 (16:41 +0200)]
Revert "implicit `Option`-returning doctests"

This reverts commit 6bb6c001be34d0932a014df981ee18f165c43374.

5 years agoClean up inherent_impls
John Kåre Alsaker [Tue, 16 Apr 2019 08:45:01 +0000 (10:45 +0200)]
Clean up inherent_impls

5 years agorustc: use a separate copy of P for HIR than for AST.
Eduard-Mihai Burtescu [Wed, 12 Jun 2019 06:41:00 +0000 (09:41 +0300)]
rustc: use a separate copy of P for HIR than for AST.

5 years agosyntax: use `box` instead of `Box::new` in `ptr::P`.
Eduard-Mihai Burtescu [Wed, 12 Jun 2019 06:36:12 +0000 (09:36 +0300)]
syntax: use `box` instead of `Box::new` in `ptr::P`.

5 years agorustc: always keep `hir::Path` behind a `P<...>`.
Eduard-Mihai Burtescu [Wed, 19 Jun 2019 17:10:37 +0000 (20:10 +0300)]
rustc: always keep `hir::Path` behind a `P<...>`.

5 years agoAddress review comments
Vadim Petrochenkov [Mon, 1 Jul 2019 09:20:44 +0000 (12:20 +0300)]
Address review comments

5 years agosyntax: Unsupport `foo! bar { ... }` macros in the parser
Vadim Petrochenkov [Sun, 30 Jun 2019 22:08:49 +0000 (01:08 +0300)]
syntax: Unsupport `foo! bar { ... }` macros in the parser

Unreserve `macro_rules` as a macro name

5 years agoAuto merge of #62253 - Centril:rollup-115uuuq, r=Centril
bors [Mon, 1 Jul 2019 06:41:48 +0000 (06:41 +0000)]
Auto merge of #62253 - Centril:rollup-115uuuq, r=Centril

Rollup of 8 pull requests

Successful merges:

 - #62062 (Use a more efficient iteration order for forward dataflow)
 - #62063 (Use a more efficient iteration order for backward dataflow)
 - #62224 (rustdoc: remove unused derives and variants)
 - #62228 (Extend the #[must_use] lint to boxed types)
 - #62235 (Extend the `#[must_use]` lint to arrays)
 - #62239 (Fix a typo)
 - #62241 (Always parse 'async unsafe fn' + properly ban in 2015)
 - #62248 (before_exec actually will only get deprecated with 1.37)

Failed merges:

r? @ghost

5 years agoAuto merge of #61682 - Centril:stabilize-type_alias_enum_variants, r=petrochenkov
bors [Mon, 1 Jul 2019 03:11:19 +0000 (03:11 +0000)]
Auto merge of #61682 - Centril:stabilize-type_alias_enum_variants, r=petrochenkov

Stabilize `type_alias_enum_variants` in Rust 1.37.0

Stabilize `#![feature(type_alias_enum_variants)]` which allows type-relative resolution with highest priority to `enum` variants in both expression and pattern contexts. For example, you may now write:

```rust
enum Option<T> {
    None,
    Some(T),
}

type OptAlias<T> = Option<T>;

fn work_on_alias(x: Option<u8>) -> u8 {
    match x {
        OptAlias::Some(y) => y + 1,
        OptAlias::None => 0,
    }
}
```

Closes https://github.com/rust-lang/rfcs/issues/2218
Closes https://github.com/rust-lang/rust/issues/52118

r? @petrochenkov

5 years agoSwitch tracking issue for 'slice_patterns'.
Mazdak Farrokhzad [Sun, 30 Jun 2019 21:23:11 +0000 (23:23 +0200)]
Switch tracking issue for 'slice_patterns'.

5 years agoRollup merge of #62248 - RalfJung:release-notes, r=Mark-Simulacrum
Mazdak Farrokhzad [Sun, 30 Jun 2019 19:57:32 +0000 (21:57 +0200)]
Rollup merge of #62248 - RalfJung:release-notes, r=Mark-Simulacrum

before_exec actually will only get deprecated with 1.37

Not sure if we usually fix old release notes, but I just found this when scrolling over them.

5 years agoRollup merge of #62241 - Centril:fix-async-unsafe-order, r=petrochenkov
Mazdak Farrokhzad [Sun, 30 Jun 2019 19:57:30 +0000 (21:57 +0200)]
Rollup merge of #62241 - Centril:fix-async-unsafe-order, r=petrochenkov

Always parse 'async unsafe fn' + properly ban in 2015

Parse `async unsafe fn` not `unsafe async fn` in implementations. We also take the opportunity to properly ban `async fn` in Rust 2015 when they are inside implementations.

Closes https://github.com/rust-lang/rust/issues/62232.

cc https://github.com/rust-lang/rust/pull/61319, https://github.com/rust-lang/rust/issues/62121, and https://github.com/rust-lang/rust/issues/62149.

r? @petrochenkov

5 years agoRollup merge of #62239 - lcolaholicl:lcolaholicl-patch-1, r=kennytm
Mazdak Farrokhzad [Sun, 30 Jun 2019 19:57:29 +0000 (21:57 +0200)]
Rollup merge of #62239 - lcolaholicl:lcolaholicl-patch-1, r=kennytm

Fix a typo

The definition of 京 seems to be capital, but not capitol.
[reference](https://en.wiktionary.org/wiki/%E4%BA%AC#Etymology_1)
[another reference](https://jisho.org/word/%E4%BA%AC)

5 years agoRollup merge of #62235 - varkor:must_use-adt-components, r=Centril
Mazdak Farrokhzad [Sun, 30 Jun 2019 19:57:28 +0000 (21:57 +0200)]
Rollup merge of #62235 - varkor:must_use-adt-components, r=Centril

Extend the `#[must_use]` lint to arrays

Based on top of https://github.com/rust-lang/rust/pull/62228.

r? @Centril

5 years agoRollup merge of #62228 - varkor:must_use-trait-in-box, r=Centril
Mazdak Farrokhzad [Sun, 30 Jun 2019 19:57:27 +0000 (21:57 +0200)]
Rollup merge of #62228 - varkor:must_use-trait-in-box, r=Centril

Extend the #[must_use] lint to boxed types

Fixes https://github.com/rust-lang/rust/issues/55506#issuecomment-506940181 (cc @Nemo157).

This should have been included as part of https://github.com/rust-lang/rust/pull/55663, but was overlooked.

5 years agoRollup merge of #62224 - euclio:remove-derives, r=GuillaumeGomez
Mazdak Farrokhzad [Sun, 30 Jun 2019 19:57:25 +0000 (21:57 +0200)]
Rollup merge of #62224 - euclio:remove-derives, r=GuillaumeGomez

rustdoc: remove unused derives and variants

Though many structs in rustdoc derive `RustcEncodable` and `RustcDecodable`, the impls do not appear to be used by the crate or its dependents. Removing them revealed some enum variants that are never constructed, too.

r? @GuillaumeGomez

5 years agoRollup merge of #62063 - ecstatic-morse:dataflow-backward-order, r=nagisa
Mazdak Farrokhzad [Sun, 30 Jun 2019 19:57:24 +0000 (21:57 +0200)]
Rollup merge of #62063 - ecstatic-morse:dataflow-backward-order, r=nagisa

Use a more efficient iteration order for backward dataflow

This applies the same basic principle as #62062 to the reverse dataflow analysis used to compute liveness information. It is functionally equivalent, except that post-order is used instead of reverse post-order.

In the long-term, `BitDenotation` should probably be extended to support both forward and backward dataflow, but there's some more work needed to get to that point.

5 years agoRollup merge of #62062 - ecstatic-morse:dataflow-order, r=nagisa
Mazdak Farrokhzad [Sun, 30 Jun 2019 19:57:22 +0000 (21:57 +0200)]
Rollup merge of #62062 - ecstatic-morse:dataflow-order, r=nagisa

Use a more efficient iteration order for forward dataflow

Currently, dataflow begins by visiting each block in order of ID (`BasicBlock(0)`, `BasicBlock(1)`, etc.). This PR changes that initial iteration to reverse post-order (see [this blog post](https://eli.thegreenplace.net/2015/directed-graph-traversal-orderings-and-applications-to-data-flow-analysis/#data-flow-analysis) for more info). This ensures that the effects of all predecessors will be applied before a basic block is visited if the CFG has no back-edges, and should result in less total iterations even when back-edges exist. This should not change the results of dataflow analysis.

The current ordering for basic blocks may be pretty close to RPO already--`BasicBlock(0)` is already the start block--in which case the cost of doing the traversal up front will outweigh the efficiency gains.
A perf run is needed to check this.

r? @pnkfelix (I think).

5 years agoUpdate mem::replace example to not be identical to mem::take
Chris Gregory [Sun, 30 Jun 2019 19:27:32 +0000 (12:27 -0700)]
Update mem::replace example to not be identical to mem::take

This also adds assertions that the operations work as expected.

5 years agoImprove box clone doctests to ensure the documentation is valid
Chris Gregory [Sun, 30 Jun 2019 18:56:21 +0000 (11:56 -0700)]
Improve box clone doctests to ensure the documentation is valid

5 years agoExtend #[must_use] lint to arrays
varkor [Sat, 29 Jun 2019 15:23:15 +0000 (16:23 +0100)]
Extend #[must_use] lint to arrays

5 years agoUpdate miri
varkor [Sun, 30 Jun 2019 17:18:44 +0000 (18:18 +0100)]
Update miri

5 years agoFix run-pass tests
varkor [Sat, 29 Jun 2019 15:51:44 +0000 (16:51 +0100)]
Fix run-pass tests

5 years agoImprove error messages for boxed trait objects in tuples
varkor [Sat, 29 Jun 2019 13:57:12 +0000 (14:57 +0100)]
Improve error messages for boxed trait objects in tuples

5 years agoExtend the #[must_use] lint to boxed types
varkor [Sat, 29 Jun 2019 12:53:46 +0000 (13:53 +0100)]
Extend the #[must_use] lint to boxed types

5 years agobefore_exec actually will only get deprecated with 1.37
Ralf Jung [Sun, 30 Jun 2019 16:29:39 +0000 (18:29 +0200)]
before_exec actually will only get deprecated with 1.37

5 years agoAuto merge of #59722 - Zoxc:query-cache, r=eddyb
bors [Sun, 30 Jun 2019 12:15:14 +0000 (12:15 +0000)]
Auto merge of #59722 - Zoxc:query-cache, r=eddyb

Clean up query cache code

There are a couple of queries for which we do not promote saved results, which have been removed from caching here. This was likely the cause of the regression in https://github.com/rust-lang/rust/pull/57293#issuecomment-476421298.

r? @michaelwoerister

5 years agoKeep caching for non-promoted queries
John Kåre Alsaker [Wed, 26 Jun 2019 08:16:56 +0000 (10:16 +0200)]
Keep caching for non-promoted queries

5 years agoClean up query cache code
John Kåre Alsaker [Fri, 5 Apr 2019 11:11:44 +0000 (13:11 +0200)]
Clean up query cache code

5 years agoAuto merge of #61252 - indygreg:upgrade-musl, r=alexcrichton
bors [Sun, 30 Jun 2019 08:42:58 +0000 (08:42 +0000)]
Auto merge of #61252 - indygreg:upgrade-musl, r=alexcrichton

Update musl to 1.1.22

This is the latest available version. I noticed Rust wasn't using the
latest version when I attempted to have Cargo link object files produced
outside of Cargo / Rust's toolchain and was getting missing symbol
errors. Those missing symbols were added in 1.1.21 and 1.1.22.

I'm not fully sure of the implications of this change or how
comprehensive the test coverage is. Upstream changes in 1.1.21 and
1.1.22 can be found at
https://git.musl-libc.org/cgit/musl/tree/WHATSNEW?h=v1.1.22&id=e97681d6f2c44bf5fa9ecdd30607cb63c780062e#n1989.

5 years agoAuto merge of #61767 - JohnTitor:update-smallvec, r=nikomatsakis
bors [Sun, 30 Jun 2019 05:21:51 +0000 (05:21 +0000)]
Auto merge of #61767 - JohnTitor:update-smallvec, r=nikomatsakis

Update new_debug_unreachable

In latest master, smallvec has been updated. So we only update new_debug_unreachable to remove unmaintained dependencies.

5 years agoReplace error by bug macro
Christian Poveda [Sun, 30 Jun 2019 03:50:37 +0000 (22:50 -0500)]
Replace error by bug macro

5 years agoAuto merge of #61459 - GuillaumeGomez:fix-rustdoc-sysroot-panic, r=ollie27,bjorn3...
bors [Sun, 30 Jun 2019 01:44:44 +0000 (01:44 +0000)]
Auto merge of #61459 - GuillaumeGomez:fix-rustdoc-sysroot-panic, r=ollie27,bjorn3,QuietMisdreavus

Prevent panic when sysroot cannot be computed

Fixes #61377.

cc @rotty @rust-lang/rustdoc

r? @Manishearth

5 years agoAdd missing type urls in Into trait
Guillaume Gomez [Thu, 27 Jun 2019 16:01:31 +0000 (18:01 +0200)]
Add missing type urls in Into trait

5 years agoAdd missing links for TryFrom docs
Guillaume Gomez [Sat, 29 Jun 2019 20:24:32 +0000 (22:24 +0200)]
Add missing links for TryFrom docs

5 years agowfcheck: resolve the type-vars in `AdtField` types
Ariel Ben-Yehuda [Sat, 29 Jun 2019 20:05:58 +0000 (23:05 +0300)]
wfcheck: resolve the type-vars in `AdtField` types

Normalization can leave some type-vars unresolved in its return type.
Make sure to resolve them so we have an infcx-independent type that can
be used with `needs_drop`.

Fixes #61402.

5 years agoAlways parse 'async unsafe fn' + properly ban in 2015.
Mazdak Farrokhzad [Sat, 29 Jun 2019 19:38:26 +0000 (21:38 +0200)]
Always parse 'async unsafe fn' + properly ban in 2015.

5 years agoSimplify control flow
Christian Poveda [Sat, 29 Jun 2019 18:19:37 +0000 (13:19 -0500)]
Simplify control flow

5 years agofix the same typo in doctest
lcolaholicl [Sat, 29 Jun 2019 17:46:45 +0000 (02:46 +0900)]
fix the same typo in doctest

5 years agoFix a typo
lcolaholicl [Sat, 29 Jun 2019 17:02:03 +0000 (02:02 +0900)]
Fix a typo
Fix a typo in `libcore/char/methods.rs`

5 years agoInline rust_printer
Mark Rousskov [Mon, 24 Jun 2019 18:54:48 +0000 (14:54 -0400)]
Inline rust_printer

5 years agoRemove unused import
Mark Rousskov [Mon, 24 Jun 2019 20:05:38 +0000 (16:05 -0400)]
Remove unused import

5 years agoRemove io::Result from syntax::print
Mark Rousskov [Mon, 24 Jun 2019 18:15:11 +0000 (14:15 -0400)]
Remove io::Result from syntax::print

Since we're now writing directly to the vector, there's no need to
thread results through the whole printing infrastructure

5 years agoReplace pretty-printer Box<dyn Write> with &mut String
Mark Rousskov [Mon, 24 Jun 2019 16:42:21 +0000 (12:42 -0400)]
Replace pretty-printer Box<dyn Write> with &mut String

5 years agoDelete now-unused methods
Mark Rousskov [Mon, 24 Jun 2019 16:13:22 +0000 (12:13 -0400)]
Delete now-unused methods

5 years agoPrivatize a bunch of methods
Mark Rousskov [Mon, 24 Jun 2019 16:12:56 +0000 (12:12 -0400)]
Privatize a bunch of methods

5 years agoUse pointer size as the source size
Christian Poveda [Sat, 29 Jun 2019 12:59:45 +0000 (07:59 -0500)]
Use pointer size as the source size

5 years agoAuto merge of #62226 - Centril:rollup-rcy1alx, r=Centril
bors [Sat, 29 Jun 2019 11:45:11 +0000 (11:45 +0000)]
Auto merge of #62226 - Centril:rollup-rcy1alx, r=Centril

Rollup of 7 pull requests

Successful merges:

 - #61199 (Revert "Set test flag when rustdoc is running with --test option" )
 - #61755 (Add `--pass $mode` to compiletest through `./x.py`)
 - #61818 (Issue #60709 test)
 - #62023 (publish_toolstate: don't use 'new' from inside the loop)
 - #62104 (Inform the query system about properties of queries at compile time)
 - #62163 (Avoid mem::uninitialized() in std::sys::unix)
 - #62204 (doc(libcore) Fix CS)

Failed merges:

r? @ghost