Auto merge of #74195 - Manishearth:rollup-h3m0sl8, r=Manishearth
Rollup of 14 pull requests
Successful merges:
- #73292 (Fixing broken link for the Eq trait)
- #73791 (Allow for parentheses after macro intra-doc-links)
- #74070 ( Use for<'tcx> fn pointers in Providers, instead of having Providers<'tcx>.)
- #74077 (Use relative path for local links to primitives)
- #74079 (Eliminate confusing "globals" terminology.)
- #74107 (Hide `&mut self` methods from Deref in sidebar if there are no `DerefMut` impl for the type.)
- #74136 (Fix broken link in rustdocdoc)
- #74137 (Update cargo)
- #74142 (Liballoc use vec instead of vector)
- #74143 (Try remove unneeded ToString import in liballoc slice)
- #74146 (update miri)
- #74150 (Avoid "blacklist")
- #74184 (Add docs for intra-doc-links)
- #74188 (Tweak `::` -> `:` typo heuristic and reduce verbosity)
Rollup merge of #74137 - ehuss:update-cargo, r=ehuss
Update cargo
6 commits in fede83ccf973457de319ba6fa0e36ead454d2e20..4f74d9b2a771c58b7ef4906b2668afd075bc8081
2020-07-02 21:51:34 +0000 to 2020-07-08 17:13:00 +0000
- Disable long_file_names test if not supported on Windows. (rust-lang/cargo#8469)
- Add support for deserializing enums in config files (rust-lang/cargo#8454)
- Write GNU tar files, supporting long names. (rust-lang/cargo#8453)
- Don't overwrite existing `rustdoc` args with --document-private-items (rust-lang/cargo#8449)
- Add some help about rustup's +toolchain syntax. (rust-lang/cargo#8455)
- Update metadata man page. (rust-lang/cargo#8451)
Rollup merge of #74079 - nnethercote:session-globals, r=nikomatsakis
Eliminate confusing "globals" terminology.
There are some structures that are called "globals", but are they global
to a compilation session, and not truly global. I have always found this
highly confusing, so this commit renames them as "session globals" and
adds a comment explaining things.
Also, the commit fixes an unnecessary nesting of `set()` calls
`src/librustc_errors/json/tests.rs`
Rollup merge of #74070 - eddyb:forall-tcx-providers, r=nikomatsakis
Use for<'tcx> fn pointers in Providers, instead of having Providers<'tcx>.
In order to work around normalization-under-HRTB (for `provide!` in `rustc_metadata`), we ended up with this:
```rust
struct Providers<'tcx> {
type_of: fn(TyCtxt<'tcx>, DefId) -> Ty<'tcx>,
// ...
}
```
But what I initially wanted to do, IIRC, was this:
```rust
struct Providers {
type_of: for<'tcx> fn(TyCtxt<'tcx>, DefId) -> Ty<'tcx>,
// ...
}
```
This PR moves to the latter, for the simple reason that only the latter allows keeping a `Providers` value, or a subset of its `fn` pointer fields, around in a `static` or `thread_local!`, which can be really useful for custom drivers that override queries.
(@jyn514 and I came across a concrete usecase of that in `rustdoc`)
The `provide!` macro in `rustc_metadata` is fixed by making the query key/value types available as type aliases under `ty::query::query_{keys,values}`, not just associated types (this is the first commit).
There are some structures that are called "globals", but are they global
to a compilation session, and not truly global. I have always found this
highly confusing, so this commit renames them as "session globals" and
adds a comment explaining things.
Also, the commit fixes an unnecessary nesting of `set()` calls
`src/librustc_errors/json/tests.rs`
Auto merge of #74153 - ehuss:fix-bootstrap-test-librustc, r=Mark-Simulacrum
Fix x.py test for librustc crates.
#73352 introduced a bug where `x.py test src/librustc_ast` would fail to actually run the tests. The issue is that `krate` and `all_krates` were changed to return relative paths. This caused the code to do a test of "relative_path ends with absolute path" which is always false. The solution is to swap that around.
The change to `Crate` isn't necessary, it just simplifies the code and makes it uniform with `CrateLibrustc`.
There don't seem to be any other compiletests that are 1) building a standalone "no_core" create and then 2) trying to link against it. There seems to be a platform-specific limitation in doing so:
```
2020-07-08T16:07:42.9419409Z = note: Creating library D:\a\rust\rust\build\i686-pc-windows-msvc\test\rustdoc\intra-link-prim-methods-external-core\auxiliary\my_core.dll.lib and object D:\a\rust\rust\build\i686-pc-windows-msvc\test\rustdoc\intra-link-prim-methods-external-core\auxiliary\my_core.dll.exp
2020-07-08T16:07:42.9419810Z LINK : error LNK2001: unresolved external symbol __DllMainCRTStartup@12
2020-07-08T16:07:42.9420032Z D:\a\rust\rust\build\i686-pc-windows-msvc\test\rustdoc\intra-link-prim-methods-external-core\auxiliary\my_core.dll : fatal error LNK1120: 1 unresolved externals
```
Possibly this could be resolved by adding a `__DllMainCRTStartup` or `__DllMainCRTStartup@12` symbol in an architecture- and platform-specific way.
Auto merge of #74124 - ehuss:fix-doc-dry-run-up-to-date, r=Mark-Simulacrum
Fix occasional bootstrap panic in docs.
I am occasionally running into this panic when running `x.py`:
> thread 'main' panicked at 'source "/Users/eric/Proj/rust/rust/build/x86_64-apple-darwin/md-doc/unstable-book" failed to get metadata: No such file or directory (os error 2)', src/build_helper/lib.rs:173:19
I have not been able to figure out the exact sequence of commands that leads to this error (I tried for quite a while to reproduce it). I think it may involve updating my tree, but I am uncertain. An artificial way to trigger it is to build the documentation, and then delete the `md-doc` directory manually.
The cause is that bootstrap does a "dry run" before every command, and in this case `up_to_date` panics because the destination exists (`build/x86_64-apple-darwin/doc/unstable-book/index.html `) but the source does not (`build/x86_64-apple-darwin/md-doc/unstable-book`).
I am uncertain if it is important that the last line `builder.run(…)` needs to be called during the dry run. This patch seems to fix the issue, though.
Auto merge of #74117 - Manishearth:rollup-ds7z0kx, r=Manishearth
Rollup of 14 pull requests
Successful merges:
- #70563 ([rustdoc] Page hash handling)
- #73856 (Edit librustc_lexer top-level docs)
- #73870 (typeck: adding type information to projection)
- #73953 (Audit hidden/short code suggestions)
- #73962 (libstd/net/tcp.rs: #![deny(unsafe_op_in_unsafe_fn)])
- #73969 (mir: mark mir construction temporaries as internal)
- #73974 (Move A|Rc::as_ptr from feature(weak_into_raw) to feature(rc_as_ptr))
- #74067 (rustdoc: Restore underline text decoration on hover for FQN in header)
- #74074 (Fix the return type of Windows' `OpenOptionsExt::security_qos_flags`.)
- #74078 (Always resolve type@primitive as a primitive, not a module)
- #74089 (Add rust-analyzer to the build manifest)
- #74090 (Remove unused RUSTC_DEBUG_ASSERTIONS)
- #74102 (Fix const prop ICE)
- #74112 (Expand abbreviation in core::ffi description)
Rollup merge of #74089 - matklad:ship-rust-analyzer-some-more, r=pietroalbini
Add rust-analyzer to the build manifest
Does not die locally, produces `rust-analyzer-0.1.0-dev-x86_64-unknown-linux-gnu.tar.gz.sha256` and add something about rust-analyzer to some `.toml` file. Seems like a success?
Rollup merge of #74078 - jyn514:lut, r=Manishearth
Always resolve type@primitive as a primitive, not a module
Previously, if there were a module in scope with the same name as the
primitive, that would take precedence. Coupled with
https://github.com/rust-lang/rust/issues/58699, this made it impossible
to link to the primitive when that module was in scope.
This approach could be extended so that `struct@foo` would no longer resolve
to any type, etc. However, it could not be used for glob imports:
```rust
pub mod foo {
pub struct Bar;
}
pub enum Bar {}
use foo::*;
// This is expected to link to `inner::Bar`, but instead it will link to the enum.
/// Link to [struct@Bar]
pub struct MyDocs;
```
The reason for this is that this change does not affect the resolution
algorithm of rustc_resolve at all. The only reason we could special-case
primitives is because we have a list of all possible primitives ahead of time.
Rollup merge of #74067 - rye:rustdoc-fqn-hover-underline, r=GuillaumeGomez
rustdoc: Restore underline text decoration on hover for FQN in header
This causes the components of FQN's (e.g. `std`, `net`, and `Ipv4Addr` of the FQN `std::net::Ipv4Addr`) to behave similarly to other links in the contents of rustdoc-styled pages. When the user hovers over them, more clearly indicating that they can be used for navigation.
I (and I hope others at least in part) have found the prior design to be somewhat confusing, as it is not clear (upon hovering) that the various parts of the FQN are actually links that the user can navigate to.
<details><summary>📸 Before, mouse hovered over "net" in the FQN</summary>
<img alt="A rustdoc page with the mouse hovered over the fully-qualified name in the page header, producing no visual change" src="https://user-images.githubusercontent.com/1566689/86538363-4c827000-bebb-11ea-8291-5ea6b85d7e19.png" />
</details>
<details><summary>📸 After, mouse hovered over "net" in the FQN</summary>
<img alt="A rustdoc page with the mouse hovered over the fully-qualified name in the page header, now with an underline showing up under the word hovered over by the mouse" src="https://user-images.githubusercontent.com/1566689/86538471-d3374d00-bebb-11ea-9bb3-7aa2d7a4800b.png" />
</details>
Rollup merge of #73974 - CAD97:rc-no-weak, r=dtolnay
Move A|Rc::as_ptr from feature(weak_into_raw) to feature(rc_as_ptr)
These were stabilized alongside the Weak versions, but having `feature = "weak_.."` on a fn definition for the non-weak pointers is potentially very misleading, especially in a review context where the impl header may not be immediately visible.
Rollup merge of #73969 - davidtwco:issue-73914-checkedadd-temp-generator-interior, r=matthewjasper
mir: mark mir construction temporaries as internal
Fixes #73914.
This PR marks temporaries from MIR construction as internal such that they are skipped in `sanitize_witness` (where each MIR local is checked to have been contained within the generator interior computed during typeck). This resolves an ICE whereby the construction of checked addition introduced a `(u64, bool)` temporary which was not in the HIR and thus not in the generator interior.
Rollup merge of #73953 - JohnTitor:audit-hidden-sugg, r=estebank
Audit hidden/short code suggestions
Should fix #73641.
Audit uses of `span_suggestion_short` and `tool_only_span_suggestion` (`span_suggestion_hidden` is already tested with `run-rustfix`). Leave some FIXMEs for futher improvements/fixes.
r? @estebank
Rollup merge of #73870 - sexxi-goose:projection-ty, r=nikomatsakis
typeck: adding type information to projection
This commit modifies the Place as follow:
* remove 'ty' from ProjectionKind
* add type information into to Projection
* replace 'ty' in Place with 'base_ty'
* introduce 'ty()' in `Place` to return the final type of the `Place`
* introduce `ty_before_projection()` in `Place` to return the type of
a `Place` before i'th projection is applied
Joshua Nelson [Mon, 6 Jul 2020 01:20:31 +0000 (21:20 -0400)]
Always resolve type@primitive as a primitive, not a module
Previously, if there were a module in scope with the same name as the
primitive, that would take precedence. Coupled with
https://github.com/rust-lang/rust/issues/58699, this made it impossible
to link to the primitive when that module was in scope.
This approach could be extended so that `struct@foo` would no longer resolve
to any type, etc. However, it could not be used for glob imports:
```rust
pub mod foo {
pub struct Bar;
}
pub enum Bar {}
use foo::*;
// This is expected to link to `inner::Bar`, but instead it will link to the enum.
/// Link to [struct@Bar]
pub struct MyDocs;
```
The reason for this is that this change does not affect the resolution
algorithm of rustc_resolve at all. The only reason we could special-case
primitives is because we have a list of all possible primitives ahead of time.
Rollup merge of #74025 - tmiasko:try-unwrap, r=Amanieu
Remove unnecessary release from Arc::try_unwrap
The thread that recovers the unique access to Arc inner value (e.g., drop
when ref-count strong reaches zero, successful try_unwrap), ensures that
other operations on Arc inner value happened before by synchronizing
with release operations performed when decrementing the reference counter.
When try_unwrap succeeds, the current thread recovers the unique access
to Arc inner value, so release is unnecessary.
Rollup merge of #74000 - lcnr:lazy_normalisation_consts, r=varkor
add `lazy_normalization_consts` feature gate
In #71973 I underestimated the amount of code which is influenced by lazy normalization of consts
and decided against having a separate feature flag for this.
Looking a bit more into this, the following issues are already working with lazy norm in its current state #47814 #57739 #73980
I therefore think it is worth it to enable lazy norm separately. Note that `#![feature(const_generics)]` still automatically activates
this feature, so using `#![feature(const_generics, lazy_normalization_consts)]` is redundant.
Rollup merge of #73937 - JohnTitor:note-exhaustive-sized-int, r=varkor
Explain exhaustive matching on {usize,isize} maximum values
The wording is taken from https://github.com/rust-lang/rfcs/blob/master/text/2591-exhaustive-integer-pattern-matching.md#reference-level-explanation.
Fixes #73919
Rollup merge of #73871 - da-x:private-types-2018-no-extern, r=petrochenkov
Fix try_print_visible_def_path for Rust 2018
The recursive check of `try_print_visible_def_path` did not properly handle the Rust 2018 case of crate-paths without 'extern crate'. Instead, it returned a "not found" via (false, self).
Rollup merge of #73834 - oli-obk:safe_intrinsics, r=ecstatic-morse
Some refactoring around intrinsic type checking
So... This PR went a bit overboard. I wanted to make the `rustc_peek` intrinsic safe (cc @ecstatic-morse ), and remembered a long-standing itch of mine. So I made that huge `&str` match for the intrinsic name a match on `Symbol`s (so basically `u32`s). This is unlikely to have a positive perf effect, even if it likely has better codegen (intrinsics are used rarely, mostly once in their wrapper), so it's mostly a consistency thing since other places actually match on the symbol name of the intrinsics.
I have not created an rfc, because i felt like this is a very minor change.
i have just set a random feature name and rust version as stability attribute, i expect to have to change that, i just don't know what the policy on that is. all guides i could find focused on contributing to the compiler, not contributing to the standard library.
drawbacks: more code in the standard library, could be replaced with specialization: base-implementation for AsRef\<str> and specialization for String and Cow. i can write that code if ppl want it.
advantages: using "real strings" i.e. Box\<str> is as ergonomic as string slices (&str) and string buffers (String) with iterators.
Auto merge of #73879 - ecstatic-morse:discr-switch-uninit, r=oli-obk
Handle inactive enum variants in `MaybeUninitializedPlaces`
Resolves the first part of #69715.
This is the equivalent of #68528 but for `MaybeUninitializedPlaces`. Because we now notify drop elaboration that inactive enum variants might be uninitialized, some drops get marked as ["open" that were previously "static"](https://github.com/rust-lang/rust/blob/e0e5d82e1677c82d209b214bbfc2cc5705c2336a/src/librustc_mir/transform/elaborate_drops.rs#L191). Unlike in #69715, this isn't strictly better: An "open" drop expands to more MIR than a simple call to the drop shim. However, because drop elaboration considers each field of an "open" drop separately, it can sometimes eliminate unnecessary drops of moved-from or unit-like enum variants. This is the case for `Option::unwrap`, which is reflected in the `mir-opt` test.
rustdoc: Restore underline text decoration on hover for FQN in header
This causes the components of FQN's to behave similarly to other links
in the contents of rustdoc-styled pages.
I (and I hope others at least in part) have found the prior design to be
somewhat confusing, as it is not clear (upon hovering) that the various
parts of the FQN are actually links that the user can navigate to.
In short, this patch makes links in the FQN have an underline when the
user hovers over them, more clearly indicating that they can be used for
navigation.