]> git.lizzy.rs Git - rust.git/log
rust.git
17 months agoRollup merge of #106466 - clubby789:relative-module-fix, r=notriddle
Matthias Krüger [Fri, 6 Jan 2023 20:26:11 +0000 (21:26 +0100)]
Rollup merge of #106466 - clubby789:relative-module-fix, r=notriddle

Fix rustdoc source code rendering for `#[path = "../path/to/mod.rs"]` links

Fixes #103517

While generating the location for modules source HTML to be saved at, a `..` path component appeared to be translated to `/up/`.
Additionally, while generating the navigation sidebar, `..` path components were ignored. This means that (as in the issue above), a *real* directory structure of:
```
sys/
  unix/
    mod.rs  <-- contains #![path = "../unix/mod.rs]
    cmath.rs
```
was rendered as:
```
sys/
  unix/
    mod.rs
    unix/
      cmath.rs  <-- links to sys/unix/unix/cmath.rs.html, 404
```
While the *files* were stored as
```
sys/
  unix/
    mod.rs.html
    up/
      unix/
        cmath.rs.html
```

17 months agoRollup merge of #106453 - coastalwhite:master, r=GuillaumeGomez
Matthias Krüger [Fri, 6 Jan 2023 20:26:10 +0000 (21:26 +0100)]
Rollup merge of #106453 - coastalwhite:master, r=GuillaumeGomez

Improve include macro documentation

As outlined in #106118, the `include!` macro is a SEO problem when it comes to the Rust documentation. Beginners may see it as a replacement to `include` syntax in other languages. I feel like this documentation should quite explicitly link to the modules' documentation.

The primary goal of this PR is to address that issue by adding a warning to the documentation. While I was here, I also added some other parts. This included a `Uses` section and some (intra doc) links to other relevant topics.

I hope this can help beginners to Rust more quickly understand some multi-file project intricacies.

# References
- Syntax for the warning: https://github.com/tokio-rs/tracing/blob/58accc6da3f04af3f6144fbe6d68af7225c70c02/tracing/src/lib.rs#L55

17 months agoRollup merge of #106341 - Ezrashaw:refactor-error-code-tidy-check, r=mejrs,klensy...
Matthias Krüger [Fri, 6 Jan 2023 20:26:10 +0000 (21:26 +0100)]
Rollup merge of #106341 - Ezrashaw:refactor-error-code-tidy-check, r=mejrs,klensy,GuillaumeGomez

refactor: clean up `errors.rs` and `error_codes_check.rs`

`errors.rs` is basically unused now, `error_codes_check.rs` is useful but not well commented, etc. It also doesn't check certain things which are certainly not correct. For example, `E0505` has a UI test in `src/test/ui/error-codes/` but that test actually outputs `E0504`?! Other issues like these exist. I've implemented these with "warnings" which are a bit rough around the edges but should be removed eventually.

r? `@GuillaumeGomez` (again not sure if you want to review but its relevant to you)

17 months agoRollup merge of #106287 - Nilstrieb:its-bugging-me-how-we-dont-have-docs, r=jyn514
Matthias Krüger [Fri, 6 Jan 2023 20:26:09 +0000 (21:26 +0100)]
Rollup merge of #106287 - Nilstrieb:its-bugging-me-how-we-dont-have-docs, r=jyn514

Add some docs to `bug`, `span_bug` and `delay_span_bug`

cc `@mejrs` as you wanted me to do this, does this look good and understandable?

17 months agoUpdate tests
clubby789 [Thu, 5 Jan 2023 04:17:57 +0000 (04:17 +0000)]
Update tests

17 months agoCorrectly render sidebar for relative module paths
clubby789 [Wed, 4 Jan 2023 20:56:46 +0000 (20:56 +0000)]
Correctly render sidebar for relative module paths

17 months agoAuto merge of #106474 - erikdesjardins:noalias, r=bjorn3
bors [Fri, 6 Jan 2023 15:20:58 +0000 (15:20 +0000)]
Auto merge of #106474 - erikdesjardins:noalias, r=bjorn3

cleanup: handle -Zmutable-noalias like -Zbox-noalias

r? `@bjorn3`

cc `@RalfJung` this will conflict with #106180

17 months agoRemove HTML tags around warning
Gijs Burghoorn [Fri, 6 Jan 2023 12:20:58 +0000 (13:20 +0100)]
Remove HTML tags around warning

17 months agoAuto merge of #105805 - yanchen4791:issue-105227-fix, r=estebank
bors [Fri, 6 Jan 2023 12:08:49 +0000 (12:08 +0000)]
Auto merge of #105805 - yanchen4791:issue-105227-fix, r=estebank

Suggest adding named lifetime when the return contains value borrowed from more than one lifetimes of function inputs

fix for #105227.

The problem: The suggestion of adding an explicit `'_` lifetime bound is **incorrect** when the function's return type contains a value which could be borrowed from more than one lifetimes of the function's inputs. Instead, a named lifetime parameter can be introduced in such a case.

The solution: Checking the number of elided lifetimes in the function signature. If more than one lifetimes found in the function inputs when the suggestion of adding explicit `'_` lifetime, change it to using named lifetime parameter `'a` instead.

17 months agoAuto merge of #106529 - Dylan-DPC:rollup-mvncmrk, r=Dylan-DPC
bors [Fri, 6 Jan 2023 09:17:14 +0000 (09:17 +0000)]
Auto merge of #106529 - Dylan-DPC:rollup-mvncmrk, r=Dylan-DPC

Rollup of 5 pull requests

Successful merges:

 - #106400 (Point at expressions where inference refines an unexpected type)
 - #106491 (Fix error-index redirect to work with the back button.)
 - #106494 (Add regression test for #58355)
 - #106499 (fix [type error] for error E0029 and E0277)
 - #106502 (rustdoc: remove legacy user-select CSS)

Failed merges:

r? `@ghost`
`@rustbot` modify labels: rollup

17 months agoRollup merge of #106502 - notriddle:notriddle/user-select, r=GuillaumeGomez
Dylan DPC [Fri, 6 Jan 2023 06:08:58 +0000 (11:38 +0530)]
Rollup merge of #106502 - notriddle:notriddle/user-select, r=GuillaumeGomez

rustdoc: remove legacy user-select CSS

According to [caniuse], the only supported browser that requires the vendor prefix, as defined in [RFC 1985], is Safari.

* The last version of Chrome that required a vendor prefix was version 53. The current version is 108.
* Firefox 68 is the last version that required a vendor prefix. The [current Firefox ESR] is version 102.
* The current version of Safari for Mac and iOS still requires a prefix.
* The last version of Edge that required a vendor frefix was 18. The current version of Edge is 108.
* UCAndroid support is unknown, but if it still requires a vendor prefix, it's more likely to be `-webkit-` than `-moz-` or `-ms-`, since they would want to emulate iOS for compatibility.

[caniuse]: https://caniuse.com/?search=user-select
[RFC 1985]: https://rust-lang.github.io/rfcs/1985-tiered-browser-support.html
[current Firefox ESR]: https://wiki.mozilla.org/Releases

17 months agoRollup merge of #106499 - lyming2007:issue-105946-fix, r=estebank
Dylan DPC [Fri, 6 Jan 2023 06:08:57 +0000 (11:38 +0530)]
Rollup merge of #106499 - lyming2007:issue-105946-fix, r=estebank

fix [type error] for error E0029 and E0277

check explicitly for the type references error
if ty.references_error() is true change the error to be err.delay_as_bug() and prevent the error E0029 and E0277 from emitting out this fix #105946

17 months agoRollup merge of #106494 - JohnTitor:issue-58355, r=compiler-errors
Dylan DPC [Fri, 6 Jan 2023 06:08:57 +0000 (11:38 +0530)]
Rollup merge of #106494 - JohnTitor:issue-58355, r=compiler-errors

Add regression test for #58355

Closes #58355
r? `@compiler-errors`
Signed-off-by: Yuki Okushi <jtitor@2k36.org>
17 months agoRollup merge of #106491 - ehuss:error-index-redirect, r=GuillaumeGomez,notriddle
Dylan DPC [Fri, 6 Jan 2023 06:08:56 +0000 (11:38 +0530)]
Rollup merge of #106491 - ehuss:error-index-redirect, r=GuillaumeGomez,notriddle

Fix error-index redirect to work with the back button.

This fixes the redirect page at https://doc.rust-lang.org/error-index.html so that it works with the browser's back-button. The solution is to use `window.location.replace()`, which avoids adding the page to the browser history stack.

This also cleans up the code a little bit, since it looks like it was written with different assumptions (maybe before f5857d5c5e1d2fde302f330d11c5cdea8005eb2a).  I don't think there is a need to have a redirect at https://doc.rust-lang.org/error_codes/error-index.html since I don't think fragment-based links were ever used there.

I have tested with Firefox, Chrome, and Safari. Going to `/error-index.html` redirects without adding to the stack, and can use the back button. Additionally, `/error-index.html#E0005` redirects correctly to the error page.

Finally, there is an unrelated commit to remove the 404 hack. There is an official way of avoiding the generation of the 404 page with setting input-404 to an empty value.

Fixes #106485

17 months agoRollup merge of #106400 - estebank:type-errs, r=compiler-errors
Dylan DPC [Fri, 6 Jan 2023 06:08:56 +0000 (11:38 +0530)]
Rollup merge of #106400 - estebank:type-errs, r=compiler-errors

Point at expressions where inference refines an unexpected type

Fix #106355. Fix #14007. (!)

```
error[E0308]: mismatched types
  --> src/test/ui/type/type-check/point-at-inference.rs:12:9
   |
9  |         foo.push(i);
   |                  - this is of type `&{integer}`, which makes `foo` to be inferred as `Vec<&{integer}>`
...
12 |     bar(foo);
   |     --- ^^^ expected `i32`, found `&{integer}`
   |     |
   |     arguments to this function are incorrect
   |
   = note: expected struct `Vec<i32>`
              found struct `Vec<&{integer}>`
note: function defined here
  --> src/test/ui/type/type-check/point-at-inference.rs:2:4
   |
2  | fn bar(_: Vec<i32>) {}
   |    ^^^ -----------
help: consider dereferencing the borrow
   |
9  |         foo.push(*i);
   |                  +
```

17 months agoAuto merge of #106429 - djkoloski:add_vendor_to_fuchsia_target_triple, r=nagisa
bors [Fri, 6 Jan 2023 06:05:40 +0000 (06:05 +0000)]
Auto merge of #106429 - djkoloski:add_vendor_to_fuchsia_target_triple, r=nagisa

Add vendor to Fuchsia's target triple

Historically, Rust's Fuchsia targets have been labeled x86_64-fuchsia and aarch64-fuchsia. However, they should technically contain vendor information. This CL changes Fuchsia's target triples to include the "unknown" vendor since Clang now does normalization and handles all triple spellings.

This was previously attempted in #90510, which was closed due to inactivity.

17 months agofix [type error] for error E0029 and E0277
Yiming Lei [Thu, 5 Jan 2023 17:12:37 +0000 (09:12 -0800)]
fix [type error] for error E0029 and E0277
check explicitly for the type references error
if ty.references_error() is true change the error to be err.delay_as_bug()
and prevent the error E0029 and E0277 from emitting out
this fix #105946

17 months agoSuggests adding named lifetime when the return contains value borrowed from more...
yanchen4791 [Fri, 16 Dec 2022 22:47:24 +0000 (14:47 -0800)]
Suggests adding named lifetime when the return contains value borrowed from more than one lifetimes of the function's inputs

17 months agoAuto merge of #105890 - Nilstrieb:inline-fmt-part-1, r=jackh726
bors [Fri, 6 Jan 2023 02:58:40 +0000 (02:58 +0000)]
Auto merge of #105890 - Nilstrieb:inline-fmt-part-1, r=jackh726

Fix `uninlined_format_args` in compiler crates with the diagnostic migration completed

Convert all the crates that have had their diagnostic migration completed (except save_analysis because that will be deleted soon and apfloat because of the licensing problem).

Some of them have been reviewed by myself and they were all correct (though I still recommend going over all of them again for review).

17 months agoAuto merge of #106416 - Nilstrieb:better-failure, r=compiler-errors
bors [Thu, 5 Jan 2023 23:27:51 +0000 (23:27 +0000)]
Auto merge of #106416 - Nilstrieb:better-failure, r=compiler-errors

Shrink `ParseResult` in the hot path.

#105570 increased the size, which caused regressions. This uses the existing generic infrastructure to differentiate between the hot path and the diagnostics path.

17 months agoShrink `ParseResult` in the hot path.
Nilstrieb [Tue, 3 Jan 2023 18:28:28 +0000 (19:28 +0100)]
Shrink `ParseResult` in the hot path.

A recent PR increased the size, which caused regressions. This uses the
existing generic infrastructure to differentiate between the hot path
and the diagnostics path.

17 months agoerror-index: Don't generate 404 instead of removing it.
Eric Huss [Thu, 5 Jan 2023 14:44:11 +0000 (06:44 -0800)]
error-index: Don't generate 404 instead of removing it.

17 months agoFix error-index redirect to work with back button.
Eric Huss [Thu, 5 Jan 2023 19:07:11 +0000 (11:07 -0800)]
Fix error-index redirect to work with back button.

17 months agorustdoc: remove legacy user-select CSS
Michael Howell [Thu, 5 Jan 2023 17:59:21 +0000 (10:59 -0700)]
rustdoc: remove legacy user-select CSS

According to [caniuse], the only supported browser that requires the vendor prefix,
as defined in [RFC 1985], is Safari.

* The last version of Chrome that required a vendor prefix was version 53.
  The current version is 108.
* Firefox 68 is the last version that required a vendor prefix. The
  [current Firefox ESR] is version 102.
* The current version of Safari for Mac and iOS still requires a prefix.
* The last version of Edge that required a vendor frefix was 18. The current
  version of Edge is 108.
* UCAndroid support is unknown, but if it still requires a vendor prefix,
  it's more likely to be `-webkit-` than `-moz-` or `-ms-`, since they
  would want to emulate iOS for compatibility.

[caniuse]: https://caniuse.com/?search=user-select
[RFC 1985]: https://rust-lang.github.io/rfcs/1985-tiered-browser-support.html
[current Firefox ESR]: https://wiki.mozilla.org/Releases

17 months agoFix `uninlined_format_args` for some compiler crates
nils [Mon, 19 Dec 2022 09:31:55 +0000 (10:31 +0100)]
Fix `uninlined_format_args` for some compiler crates

Convert all the crates that have had their diagnostic migration
completed (except save_analysis because that will be deleted soon and
apfloat because of the licensing problem).

17 months agofix rebase
Esteban Küber [Thu, 5 Jan 2023 17:06:34 +0000 (17:06 +0000)]
fix rebase

17 months agoreview comments: reword
Esteban Küber [Thu, 5 Jan 2023 01:26:44 +0000 (01:26 +0000)]
review comments: reword

17 months agoreview comment: potentially produce more suggestions for arg type mismatch
Esteban Küber [Thu, 5 Jan 2023 01:00:21 +0000 (01:00 +0000)]
review comment: potentially produce more suggestions for arg type mismatch

17 months agoSuggest changing argument on type error
Esteban Küber [Thu, 5 Jan 2023 00:37:53 +0000 (00:37 +0000)]
Suggest changing argument on type error

17 months agoAccount for type error on method arg caused by earlier inference
Esteban Küber [Thu, 5 Jan 2023 00:18:57 +0000 (00:18 +0000)]
Account for type error on method arg caused by earlier inference

```rust
fn main() {
    let v = Vec::new();
    v.push(0);
    v.push(0);
    v.push("");
}
```

now produces

```
error[E0308]: mismatched types
  --> $DIR/point-at-inference-3.rs:6:12
   |
LL |     v.push(0);
   |            - this is of type `{integer}`, which makes `v` to be inferred as `Vec<{integer}>`
...
LL |     v.push("");
   |       ---- ^^ expected integer, found `&str`
   |       |
   |       arguments to this function are incorrect
   |
note: associated function defined here
  --> $SRC_DIR/alloc/src/vec/mod.rs:LL:COL
```

17 months agoUse `BottomUpFolder`
Esteban Küber [Wed, 4 Jan 2023 04:49:47 +0000 (20:49 -0800)]
Use `BottomUpFolder`

17 months agoFormatting
Esteban Küber [Wed, 4 Jan 2023 04:32:15 +0000 (20:32 -0800)]
Formatting

17 months agoTweak output
Esteban Küber [Wed, 4 Jan 2023 04:20:31 +0000 (20:20 -0800)]
Tweak output

- Only point at a the single expression where the found type was first
  inferred.
- Find method call argument that might have caused the found type to be
  inferred.
- Provide structured suggestion.
- Apply some review comments.
- Tweak wording.

17 months agoreview comments: do not always point at init expr
Esteban Küber [Tue, 3 Jan 2023 07:34:26 +0000 (23:34 -0800)]
review comments: do not always point at init expr

17 months agoMore eagerly resolve expr `ty`s before writing them
Esteban Küber [Tue, 3 Jan 2023 06:25:50 +0000 (22:25 -0800)]
More eagerly resolve expr `ty`s before writing them

This allows the expressions to have more accurate types
when showing inference steps.

17 months agoSkip macros to avoid talking about bindings the user can't see
Esteban Küber [Tue, 3 Jan 2023 05:56:19 +0000 (21:56 -0800)]
Skip macros to avoid talking about bindings the user can't see

17 months agoPoint at expressions where inference refines an unexpected type
Esteban Küber [Tue, 3 Jan 2023 00:23:08 +0000 (16:23 -0800)]
Point at expressions where inference refines an unexpected type

Address #106355.

17 months agoBetter phrasing for hygiene of include macro
Gijs Burghoorn [Thu, 5 Jan 2023 16:09:05 +0000 (17:09 +0100)]
Better phrasing for hygiene of include macro

17 months agoAuto merge of #106487 - GuillaumeGomez:fix-kbd-var, r=notriddle
bors [Thu, 5 Jan 2023 15:47:39 +0000 (15:47 +0000)]
Auto merge of #106487 - GuillaumeGomez:fix-kbd-var, r=notriddle

Fix --kbd-color variable name in rustdoc.css

Interestingly enough, it only impacted the dark theme.

Before:

![Screenshot from 2023-01-05 11-03-17](https://user-images.githubusercontent.com/3050060/210754145-c3bb0f50-d35f-4543-bf73-010a4524a803.png)

After:

![Screenshot from 2023-01-05 11-03-05](https://user-images.githubusercontent.com/3050060/210754190-85c2f146-a774-4463-9cd3-9495b7c91bd7.png)

r? `@notriddle`

17 months agoAdd regression test for #58355
Yuki Okushi [Thu, 5 Jan 2023 15:24:42 +0000 (00:24 +0900)]
Add regression test for #58355

Signed-off-by: Yuki Okushi <jtitor@2k36.org>
17 months agoAdd vendor to Fuchsia's target triple
David Koloski [Tue, 3 Jan 2023 20:59:21 +0000 (15:59 -0500)]
Add vendor to Fuchsia's target triple

Historically, Rust's Fuchsia targets have been labeled x86_64-fuchsia
and aarch64-fuchsia. However, they should technically contain vendor
information. This CL changes Fuchsia's target triples to include the
"unknown" vendor since Clang now does normalization and handles all
triple spellings.

This was previously attempted in #90510, which was closed due to
inactivity.

17 months agoAuto merge of #106406 - nikic:update-llvm-11, r=cuviper
bors [Thu, 5 Jan 2023 12:45:43 +0000 (12:45 +0000)]
Auto merge of #106406 - nikic:update-llvm-11, r=cuviper

Update LLVM submodule

Fixes #104099.

17 months agoUpdate kbd GUI test
Guillaume Gomez [Thu, 5 Jan 2023 10:03:45 +0000 (11:03 +0100)]
Update kbd GUI test

17 months agoFix --kbd-color variable name in rustdoc.css
Guillaume Gomez [Thu, 5 Jan 2023 10:03:35 +0000 (11:03 +0100)]
Fix --kbd-color variable name in rustdoc.css

17 months agoAuto merge of #106404 - tmiasko:dedup-box-derefs, r=wesleywiser
bors [Thu, 5 Jan 2023 09:50:05 +0000 (09:50 +0000)]
Auto merge of #106404 - tmiasko:dedup-box-derefs, r=wesleywiser

Remove duplicated elaborate box derefs pass

The pass runs earlier as a part of `run_runtime_lowering_passes`.

The duplicate was added in #99102.

17 months agoAuto merge of #106482 - compiler-errors:rollup-g7n1p39, r=compiler-errors
bors [Thu, 5 Jan 2023 06:47:02 +0000 (06:47 +0000)]
Auto merge of #106482 - compiler-errors:rollup-g7n1p39, r=compiler-errors

Rollup of 6 pull requests

Successful merges:

 - #105846 (Account for return-position `impl Trait` in trait in `opt_suggest_box_span`)
 - #106385 (Split `-Zchalk` flag into `-Ztrait-solver=(classic|chalk|next)` flag)
 - #106403 (Rename `hir::Map::{get_,find_}parent_node` to `hir::Map::{,opt_}parent_id`, and add `hir::Map::{get,find}_parent`)
 - #106462 (rustdoc: remove unnecessary wrapper around sidebar and mobile logos)
 - #106464 (Update Fuchsia walkthrough with new configs)
 - #106478 (Tweak wording of fn call with wrong number of args)

Failed merges:

r? `@ghost`
`@rustbot` modify labels: rollup

17 months agoRollup merge of #106478 - estebank:tweak-fn-mismatch, r=compiler-errors
Michael Goulet [Thu, 5 Jan 2023 04:36:29 +0000 (20:36 -0800)]
Rollup merge of #106478 - estebank:tweak-fn-mismatch, r=compiler-errors

Tweak wording of fn call with wrong number of args

17 months agoRollup merge of #106464 - djkoloski:update_fuchsia_platform_support_docs, r=tmandry
Michael Goulet [Thu, 5 Jan 2023 04:36:28 +0000 (20:36 -0800)]
Rollup merge of #106464 - djkoloski:update_fuchsia_platform_support_docs, r=tmandry

Update Fuchsia walkthrough with new configs

The new `download-ci-llvm` configuration option dosn't work with `lld = true` (see #100853). The Fuchsia walkthrough should recommend setting it to `false`.

r? `@tmandry`

17 months agoRollup merge of #106462 - notriddle:notriddle/logo-container-sidebar, r=GuillaumeGomez
Michael Goulet [Thu, 5 Jan 2023 04:36:28 +0000 (20:36 -0800)]
Rollup merge of #106462 - notriddle:notriddle/logo-container-sidebar, r=GuillaumeGomez

rustdoc: remove unnecessary wrapper around sidebar and mobile logos

This commit changes `.sidebar a:hover:not(.logo-container)` to add the `:not()` pseudo-class, retaining the old appearance of the logo when mousing over it.

This didn't used to be necessary because the `a.sidebar-logo` was `display:inline`, and was what got the `background` changed on hover, while the `div.logo-container` inside it was `display:block`. This resulted in the hover rule not having any effect, because the logo-container box was not actually nested inside the sidebar-logo box:

https://www.w3.org/TR/CSS2/visuren.html#anonymous-block-level

> When an inline box contains an in-flow block-level box, the inline box (and its inline ancestors within the same line box) are broken around the block-level box (and any block-level siblings that are consecutive or separated only by collapsible whitespace and/or out-of-flow elements), splitting the inline box into two boxes (even if either side is empty), one on each side of the block-level box(es). The line boxes before the break and after the break are enclosed in anonymous block boxes, and the block-level box becomes a sibling of those anonymous boxes. When such an inline box is affected by relative positioning, any resulting translation also affects the block-level box contained in the inline box.

17 months agoRollup merge of #106403 - compiler-errors:rename-hir-methods, r=cjgillot
Michael Goulet [Thu, 5 Jan 2023 04:36:28 +0000 (20:36 -0800)]
Rollup merge of #106403 - compiler-errors:rename-hir-methods, r=cjgillot

Rename `hir::Map::{get_,find_}parent_node` to `hir::Map::{,opt_}parent_id`, and add `hir::Map::{get,find}_parent`

The `hir::Map::get_parent_node` function doesn't return a `Node`, and I think that's quite confusing. Let's rename it to something that sounds more like something that gets the parent hir id => `hir::Map::parent_id`. Same with `find_parent_node` => `opt_parent_id`.

Also, combine `hir.get(hir.parent_id(hir_id))` and similar `hir.find(hir.parent_id(hir_id))` function into new functions that actually retrieve the parent node in one call. This last commit is the only one that might need to be looked at closely.

17 months agoRollup merge of #106385 - compiler-errors:new-solver-flag, r=jackh726
Michael Goulet [Thu, 5 Jan 2023 04:36:27 +0000 (20:36 -0800)]
Rollup merge of #106385 - compiler-errors:new-solver-flag, r=jackh726

Split `-Zchalk` flag into `-Ztrait-solver=(classic|chalk|next)` flag

We'll eventually need a way to select more than chalk + not-chalk.

Does this need an MCP since it's touching a `-Z` flag? Or perhaps I should preserve `-Zchalk` for the time being... maybe I could make it a warning to use that flag? cc ``@rust-lang/types``

r? types

17 months agoRollup merge of #105846 - compiler-errors:issue-105838, r=jackh726
Michael Goulet [Thu, 5 Jan 2023 04:36:27 +0000 (20:36 -0800)]
Rollup merge of #105846 - compiler-errors:issue-105838, r=jackh726

Account for return-position `impl Trait` in trait in `opt_suggest_box_span`

RPITITs are the only types where their opaque bounds might normalize to some other self type than the opaque type itself. To avoid needing to do normalization, let's just match on either alias kind.

Ideally, we'd just get rid of `opt_suggest_box_span`. It's kind of a wart on type-checking `if`/`match`. I've recently refactored this expression for being confusing/wrong, but moving it into the error path is pretty hard.

Fixes #105838

17 months agoAuto merge of #105409 - compiler-errors:closure-infer-cycle, r=jackh726
bors [Thu, 5 Jan 2023 03:59:31 +0000 (03:59 +0000)]
Auto merge of #105409 - compiler-errors:closure-infer-cycle, r=jackh726

Don't deduce a signature that makes a closure cyclic

Sometimes when elaborating supertrait bounds for closure signature inference, we end up deducing a closure signature that is cyclical because either a parameter or the return type references a projection mentioning `Self` that also has escaping bound vars, which means that it's not eagerly replaced with an inference variable.

Interestingly, this is not *just* related to my PR that elaborates supertrait bounds for closure signature deduction. The committed test `supertrait-hint-cycle-3.rs` shows **stable** code that is fixed by this PR:

```rust
trait Foo<'a> {
    type Input;
}

impl<F: Fn(u32)> Foo<'_> for F {
    type Input = u32;
}

fn needs_super<F: for<'a> Fn(<F as Foo<'a>>::Input) + for<'a> Foo<'a>>(_: F) {}

fn main() {
    needs_super(|_: u32| {});
}
```

Fixes #105401
Fixes #105396

r? types

17 months agoTweak wording of fn call with wrong number of args
Esteban Küber [Thu, 5 Jan 2023 03:02:10 +0000 (03:02 +0000)]
Tweak wording of fn call with wrong number of args

17 months agoAuto merge of #106472 - matthiaskrgr:rollup-4w4v50e, r=matthiaskrgr
bors [Thu, 5 Jan 2023 01:06:05 +0000 (01:06 +0000)]
Auto merge of #106472 - matthiaskrgr:rollup-4w4v50e, r=matthiaskrgr

Rollup of 7 pull requests

Successful merges:

 - #106391 (rustdoc: allow popover child links to work)
 - #106398 (Fix a few clippy lints in libtest)
 - #106412 (Fix link generation for local primitive types in rustdoc JSON output)
 - #106437 (rustdoc: fix buggy JS check for absolute URL)
 - #106451 (Merge borrowck permission checks)
 - #106460 (Move tests)
 - #106461 (docs: fix broken link "search bar")

Failed merges:

r? `@ghost`
`@rustbot` modify labels: rollup

17 months agocleanup: handle -Zmutable-noalias like -Zbox-noalias
Erik Desjardins [Thu, 5 Jan 2023 00:24:42 +0000 (19:24 -0500)]
cleanup: handle -Zmutable-noalias like -Zbox-noalias

17 months agoRollup merge of #106461 - notriddle:notriddle/window-searchstate-focus, r=GuillaumeGomez
Matthias Krüger [Wed, 4 Jan 2023 22:39:52 +0000 (23:39 +0100)]
Rollup merge of #106461 - notriddle:notriddle/window-searchstate-focus, r=GuillaumeGomez

docs: fix broken link "search bar"

Regression introduced by #84150

17 months agoRollup merge of #106460 - c410-f3r:moar-errors, r=compiler-errors
Matthias Krüger [Wed, 4 Jan 2023 22:39:51 +0000 (23:39 +0100)]
Rollup merge of #106460 - c410-f3r:moar-errors, r=compiler-errors

Move tests

r? `@petrochenkov`

17 months agoRollup merge of #106451 - Zeegomo:merge-borrowck-access, r=estebank
Matthias Krüger [Wed, 4 Jan 2023 22:39:51 +0000 (23:39 +0100)]
Rollup merge of #106451 - Zeegomo:merge-borrowck-access, r=estebank

Merge borrowck permission checks

Merge `check_access_permission` and `check_if_reassignment_to_immutable_state`.
The goal of this commit is twofold:
* simplify the codebase by removing duplicate logic.
* avoid duplicate reporting of illegal reassignment errors by reusing the exiting de-duplicating logic of access_place.

17 months agoRollup merge of #106437 - notriddle:notriddle/http-url, r=GuillaumeGomez
Matthias Krüger [Wed, 4 Jan 2023 22:39:50 +0000 (23:39 +0100)]
Rollup merge of #106437 - notriddle:notriddle/http-url, r=GuillaumeGomez

rustdoc: fix buggy JS check for absolute URL

The old code did the wrong thing when faced with a crate named "http".

17 months agoRollup merge of #106412 - GuillaumeGomez:fix-links-to-primitive-rustdoc-json, r=aDotI...
Matthias Krüger [Wed, 4 Jan 2023 22:39:50 +0000 (23:39 +0100)]
Rollup merge of #106412 - GuillaumeGomez:fix-links-to-primitive-rustdoc-json, r=aDotInTheVoid

Fix link generation for local primitive types in rustdoc JSON output

Fixes https://github.com/rust-lang/rust/issues/104064.

As mentioned in the issue, I'm not super happy about this fix which is more a hack rather than a sound-proof solution. However I couldn't find a better way to fix it.

r? `@aDotInTheVoid`

17 months agoRollup merge of #106398 - jyn514:fix-clippy, r=thomcc
Matthias Krüger [Wed, 4 Jan 2023 22:39:49 +0000 (23:39 +0100)]
Rollup merge of #106398 - jyn514:fix-clippy, r=thomcc

Fix a few clippy lints in libtest

- Remove unnecessary references and dereferences
- Use `.contains` instead of `a <= x && x <= b`
- Use `mem::take` instead of `mem::replace` where possible

cc https://github.com/rust-lang/rust/pull/106394 :)

17 months agoRollup merge of #106391 - ardislu:fix-popover-child-link, r=GuillaumeGomez
Matthias Krüger [Wed, 4 Jan 2023 22:39:49 +0000 (23:39 +0100)]
Rollup merge of #106391 - ardislu:fix-popover-child-link, r=GuillaumeGomez

rustdoc: allow popover child links to work

No need to prevent default click behavior on a `<div>`, it will also disable all child click behavior.

Closes #106390

17 months agoAuto merge of #106224 - bjorn3:staticlib_fixes, r=wesleywiser
bors [Wed, 4 Jan 2023 21:35:15 +0000 (21:35 +0000)]
Auto merge of #106224 - bjorn3:staticlib_fixes, r=wesleywiser

Small fixes for --crate-type staticlib

The first commit doesn't have an effect until we start translating error messages. The second commit fixes potential linker errors when combining `--crate-type staticlib` with another crate type and I think `-Cprefer-dynamic`.

17 months agoUpdate Fuchsia walkthrough with new configs
David Koloski [Wed, 4 Jan 2023 20:55:48 +0000 (15:55 -0500)]
Update Fuchsia walkthrough with new configs

The new `download-ci-llvm` configuration option dosn't work with `lld =
true` (see #100853). The Fuchsia walkthrough should recommend setting it
to `false`.

17 months agoRemove unused match pattern for primitive types
Guillaume Gomez [Wed, 4 Jan 2023 20:54:32 +0000 (21:54 +0100)]
Remove unused match pattern for primitive types

17 months agorustdoc: remove unnecessary wrapper around sidebar and mobile logos
Michael Howell [Wed, 4 Jan 2023 20:04:30 +0000 (13:04 -0700)]
rustdoc: remove unnecessary wrapper around sidebar and mobile logos

This commit changes `.sidebar a:hover:not(.logo-container)` to add the `:not()` pseudo-class, retaining the old appearance of the logo when mousing over it.

This didn't used to be necessary because the `a.sidebar-logo` was `display:inline`, and was what got the `background` changed on hover, while the `div.logo-container` inside it was `display:block`. This resulted in the hover rule not having any effect, because the logo-container box was not actually nested inside the sidebar-logo box:

https://www.w3.org/TR/CSS2/visuren.html#anonymous-block-level

> When an inline box contains an in-flow block-level box, the inline box (and its inline ancestors within the same line box) are broken around the block-level box (and any block-level siblings that are consecutive or separated only by collapsible whitespace and/or out-of-flow elements), splitting the inline box into two boxes (even if either side is empty), one on each side of the block-level box(es). The line boxes before the break and after the break are enclosed in anonymous block boxes, and the block-level box becomes a sibling of those anonymous boxes. When such an inline box is affected by relative positioning, any resulting translation also affects the block-level box contained in the inline box.

17 months agodocs: fix broken link "search bar"
Michael Howell [Wed, 4 Jan 2023 19:47:24 +0000 (12:47 -0700)]
docs: fix broken link "search bar"

17 months agoMove tests
Caio [Wed, 4 Jan 2023 19:44:30 +0000 (16:44 -0300)]
Move tests

17 months agoAuto merge of #106330 - weihanglo:update-cargo, r=weihanglo
bors [Wed, 4 Jan 2023 18:43:32 +0000 (18:43 +0000)]
Auto merge of #106330 - weihanglo:update-cargo, r=weihanglo

Update cargo

8 commits in 2381cbdb4e9b07090f552d34a44a529b6e620e44..8c460b2237a6359a7e3335890db8da049bdd62fc
2022-12-23 12:19:27 +0000 to 2023-01-04 14:30:01 +0000
- test: revive nightly plugin tests to work (rust-lang/cargo#11534)
- Add note to release notes about rejecting multiple registries. (rust-lang/cargo#11531)
- Fix a typo `fresheness` -&gt; `freshness` (rust-lang/cargo#11529)
- Reasons for rebuilding (rust-lang/cargo#11407)
- Asymmetric tokens (rust-lang/cargo#10771)
- Use proper git URL for GitHub repos (rust-lang/cargo#11517)
- Add `registry.default` example (rust-lang/cargo#11516)
- Support vendoring with different revs from same git repo (rust-lang/cargo#10690)

Also update license exceptions and permitted dependencies
for new cargo dependency "pasetors".

A new dependency `getrandom` is added into `rustc-workspace-hacks`,
since it requires feature `js`.

r? `@ghost`

17 months agoRename stock solver to classic
Michael Goulet [Wed, 4 Jan 2023 16:49:38 +0000 (16:49 +0000)]
Rename stock solver to classic

17 months agoUpdate tests, etc
Michael Goulet [Mon, 2 Jan 2023 23:18:00 +0000 (23:18 +0000)]
Update tests, etc

17 months agoSplit `-Zchalk` flag into `-Ztrait-solver=(stock|chalk|next)` flag
Michael Goulet [Mon, 2 Jan 2023 23:12:47 +0000 (23:12 +0000)]
Split `-Zchalk` flag into `-Ztrait-solver=(stock|chalk|next)` flag

17 months agoUpdate cargo
Weihang Lo [Wed, 4 Jan 2023 15:33:59 +0000 (15:33 +0000)]
Update cargo

8 commits in 2381cbdb4e9b07090f552d34a44a529b6e620e44..8c460b2237a6359a7e3335890db8da049bdd62fc
2022-12-23 12:19:27 +0000 to 2023-01-04 14:30:01 +0000
- test: revive nightly plugin tests to work (rust-lang/cargo#11534)
- Add note to release notes about rejecting multiple registries. (rust-lang/cargo#11531)
- Fix a typo `fresheness` -&gt; `freshness` (rust-lang/cargo#11529)
- Reasons for rebuilding (rust-lang/cargo#11407)
- Asymmetric tokens (rust-lang/cargo#10771)
- Use proper git URL for GitHub repos (rust-lang/cargo#11517)
- Add `registry.default` example (rust-lang/cargo#11516)
- Support vendoring with different revs from same git repo (rust-lang/cargo#10690)

Also update license exceptions and permitted dependencies
for new cargo dependency "pasetors".

A new dependency `getrandom` is added into `rustc-workspace-hacks`,
since it requires feature `js`.

17 months agoAuto merge of #106382 - tmiasko:sdniwnu-daed, r=wesleywiser
bors [Wed, 4 Jan 2023 15:48:47 +0000 (15:48 +0000)]
Auto merge of #106382 - tmiasko:sdniwnu-daed, r=wesleywiser

Fix handling of dead unwinds in backward analyses

Dead unwinds set contains a head of an unreachable unwind edge.

17 months agoTidy up whitespace
Gijs Burghoorn [Wed, 4 Jan 2023 15:18:34 +0000 (16:18 +0100)]
Tidy up whitespace

17 months agoImprove include macro documentation
Gijs Burghoorn [Wed, 4 Jan 2023 15:07:33 +0000 (16:07 +0100)]
Improve include macro documentation

17 months agoMerge borrowck permission checks
Giacomo Pasini [Sat, 31 Dec 2022 12:45:22 +0000 (13:45 +0100)]
Merge borrowck permission checks

Merge `check_access_permission` and
`check_if_reassignment_to_immutable_state`.
The goal of this commit is twofold:
First, we simplify the codebase by removing duplicate logic.
Second, we avoid duplicate reporting of illegal reassignment
errors by reusing the exiting de-duplicating logic of
access_place.

17 months agoFix each_linked_rlib when the current crate has 0 crate types
bjorn3 [Wed, 4 Jan 2023 13:37:38 +0000 (13:37 +0000)]
Fix each_linked_rlib when the current crate has 0 crate types

17 months agoAuto merge of #106239 - LegionMammal978:thin-box-drop-guard, r=Amanieu
bors [Wed, 4 Jan 2023 12:55:22 +0000 (12:55 +0000)]
Auto merge of #106239 - LegionMammal978:thin-box-drop-guard, r=Amanieu

Deallocate ThinBox even if the value unwinds on drop

This makes it match the behavior of an ordinary `Box`.

17 months agoAuto merge of #106442 - matthiaskrgr:rollup-wivf7gh, r=matthiaskrgr
bors [Wed, 4 Jan 2023 09:41:43 +0000 (09:41 +0000)]
Auto merge of #106442 - matthiaskrgr:rollup-wivf7gh, r=matthiaskrgr

Rollup of 7 pull requests

Successful merges:

 - #106200 (Suggest `impl Fn*` and `impl Future` in `-> _` return suggestions)
 - #106274 (Add JSON output to -Zdump-mono-stats)
 - #106292 (Add codegen test for `Box::new(uninit)` of big arrays)
 - #106327 (Add tidy check for dbg)
 - #106361 (Note maximum integer literal for `IntLiteralTooLarge`)
 - #106396 (Allow passing a specific date to `bump-stage0`)
 - #106436 (Enable doctests for rustc_query_impl)

Failed merges:

r? `@ghost`
`@rustbot` modify labels: rollup

17 months agoAdd hardcoded wait time for external navigation to load
Ardis Lu [Wed, 4 Jan 2023 07:44:35 +0000 (23:44 -0800)]
Add hardcoded wait time for external navigation to load

17 months agoRollup merge of #106436 - CarsonV:doctest_query_impl, r=jyn514
Matthias Krüger [Wed, 4 Jan 2023 06:28:57 +0000 (07:28 +0100)]
Rollup merge of #106436 - CarsonV:doctest_query_impl, r=jyn514

Enable doctests for rustc_query_impl

doctests worked out of the box for query_impl on my local machine. Working towards #99144

17 months agoRollup merge of #106396 - jyn514:bump-stage-date, r=pietroalbini
Matthias Krüger [Wed, 4 Jan 2023 06:28:56 +0000 (07:28 +0100)]
Rollup merge of #106396 - jyn514:bump-stage-date, r=pietroalbini

Allow passing a specific date to `bump-stage0`

This allows regenerating `src/stage0.json` on changes to the tool, without needing to hard-code the date in the source. It was useful for https://github.com/rust-lang/rust/pull/106394, which added clippy to the list of required components.

r? `@pietroalbini`

17 months agoRollup merge of #106361 - clubby789:int-literal-too-large, r=estebank
Matthias Krüger [Wed, 4 Jan 2023 06:28:56 +0000 (07:28 +0100)]
Rollup merge of #106361 - clubby789:int-literal-too-large, r=estebank

Note maximum integer literal for `IntLiteralTooLarge`

Closes #105908

`@rustbot` label +A-diagnostics

17 months agoRollup merge of #106327 - gimbles:dbg, r=jyn514
Matthias Krüger [Wed, 4 Jan 2023 06:28:55 +0000 (07:28 +0100)]
Rollup merge of #106327 - gimbles:dbg, r=jyn514

Add tidy check for dbg

Fixes #106169

17 months agoRollup merge of #106292 - Nilstrieb:box-uninit-test, r=RalfJung
Matthias Krüger [Wed, 4 Jan 2023 06:28:55 +0000 (07:28 +0100)]
Rollup merge of #106292 - Nilstrieb:box-uninit-test, r=RalfJung

Add codegen test for `Box::new(uninit)` of big arrays

Closes #58201

r? `@RalfJung`

17 months agoRollup merge of #106274 - jyn514:dump-mono-stats, r=lqd
Matthias Krüger [Wed, 4 Jan 2023 06:28:54 +0000 (07:28 +0100)]
Rollup merge of #106274 - jyn514:dump-mono-stats, r=lqd

Add JSON output to -Zdump-mono-stats

Follow-up to https://github.com/rust-lang/rust/pull/105481

r? `@lqd` cc `@wesleywiser`

17 months agoRollup merge of #106200 - compiler-errors:suggest-impl-trait, r=estebank
Matthias Krüger [Wed, 4 Jan 2023 06:28:54 +0000 (07:28 +0100)]
Rollup merge of #106200 - compiler-errors:suggest-impl-trait, r=estebank

Suggest `impl Fn*` and `impl Future` in `-> _` return suggestions

Follow-up to #106172, only the last commit is relevant. Can rebase once that PR is landed for easier review.

Suggests `impl Future` and `impl Fn{,Mut,Once}` in `-> _` return suggestions.

r? `@estebank`

17 months agoAuto merge of #104376 - compiler-errors:thin-metadata-implies-thin-ptr, r=wesleywiser
bors [Wed, 4 Jan 2023 06:28:05 +0000 (06:28 +0000)]
Auto merge of #104376 - compiler-errors:thin-metadata-implies-thin-ptr, r=wesleywiser

layout_of: `T: Thin` implies `sizeof(&T) == sizeof(usize)`

Use the `<T as Pointee>::Metadata` associated type to calculate the layout of a pointee's metadata, instead of hard-coding rules about certain types.

Maybe this approach is overkill -- we could instead hard-code this approach as a fallback, with the matching on `Slice`/`Dynamic`/etc. happening first

Fixes this issue here https://github.com/rust-lang/rust/pull/104338#issuecomment-1312595844 .. But is also useful with transmutes, for example, given the UI test I added below.

17 months agoAdd test case to click popover child link
Ardis Lu [Wed, 4 Jan 2023 04:38:07 +0000 (20:38 -0800)]
Add test case to click popover child link

17 months agoAuto merge of #106432 - compiler-errors:rollup-lzj0lnp, r=compiler-errors
bors [Wed, 4 Jan 2023 03:37:21 +0000 (03:37 +0000)]
Auto merge of #106432 - compiler-errors:rollup-lzj0lnp, r=compiler-errors

Rollup of 8 pull requests

Successful merges:

 - #104748 (Ensure `lld` is supported with `download-ci-llvm`)
 - #105541 (Simplify some iterator combinators)
 - #106045 (default OOM handler: use non-unwinding panic, to match std handler)
 - #106157 (Don't trim path for `unsafe_op_in_unsafe_fn` lints)
 - #106353 (Reduce spans for `unsafe impl` errors)
 - #106381 (Jsondoclint: Add `--verbose` and `--json-output` options)
 - #106411 (rustdoc: remove legacy font-feature-settings CSS)
 - #106414 (Add cuviper to the review rotation for libs)

Failed merges:

r? `@ghost`
`@rustbot` modify labels: rollup

17 months agorustdoc: fix buggy JS check for absolute URL
Michael Howell [Wed, 4 Jan 2023 02:46:40 +0000 (19:46 -0700)]
rustdoc: fix buggy JS check for absolute URL

The old code did the wrong thing when faced with a crate named "http".

17 months agoEnable query_impl doctests
CarsonV [Wed, 4 Jan 2023 02:12:09 +0000 (20:12 -0600)]
Enable query_impl doctests

17 months agoRollup merge of #106414 - cuviper:libs-review, r=Mark-Simulacrum
Michael Goulet [Wed, 4 Jan 2023 01:19:28 +0000 (17:19 -0800)]
Rollup merge of #106414 - cuviper:libs-review, r=Mark-Simulacrum

Add cuviper to the review rotation for libs

17 months agoRollup merge of #106411 - notriddle:notriddle/font-feature-settings, r=GuillaumeGomez
Michael Goulet [Wed, 4 Jan 2023 01:19:28 +0000 (17:19 -0800)]
Rollup merge of #106411 - notriddle:notriddle/font-feature-settings, r=GuillaumeGomez

rustdoc: remove legacy font-feature-settings CSS

According to [caniuse], these vendor prefixes aren't needed in any supported web browsers as defined in [RFC 1985].

* The last version of Chrome that required a vendor prefix was version 47. The current version is 108.
* Firefox 33 is the last version that required a vendor prefix. The [current Firefox ESR] is version 102.
* The last version of Safari that required a vendor prefix was version 9.1. The current version is 16.
* The last version of Safari/iOS that required a vendor prefix was version 9.3. The current version is 16.
* Edge never required vendor prefixes.
* UCAndroid never required vendor prefixes.

[caniuse]: https://caniuse.com/?search=font-feature-settings
[RFC 1985]: https://rust-lang.github.io/rfcs/1985-tiered-browser-support.html
[current Firefox ESR]: https://wiki.mozilla.org/Releases

17 months agoRollup merge of #106381 - aDotInTheVoid:jsondoclint-more-options, r=notriddle
Michael Goulet [Wed, 4 Jan 2023 01:19:27 +0000 (17:19 -0800)]
Rollup merge of #106381 - aDotInTheVoid:jsondoclint-more-options, r=notriddle

Jsondoclint: Add `--verbose` and `--json-output` options

There quite helpful for manually using jsondoclint as a debugging tool, especially on large files (these were written to look into core.json).

r? rustdoc

17 months agoRollup merge of #106353 - lukas-code:reduce-red-lines-in-my-ide, r=wesleywiser
Michael Goulet [Wed, 4 Jan 2023 01:19:27 +0000 (17:19 -0800)]
Rollup merge of #106353 - lukas-code:reduce-red-lines-in-my-ide, r=wesleywiser

Reduce spans for `unsafe impl` errors

Because huge spans aren't great for IDEs.

Prior art: https://github.com/rust-lang/rust/pull/103749

17 months agoRollup merge of #106157 - LeSeulArtichaut:106126-thir-unsafeck-good-path-bug-2, r...
Michael Goulet [Wed, 4 Jan 2023 01:19:26 +0000 (17:19 -0800)]
Rollup merge of #106157 - LeSeulArtichaut:106126-thir-unsafeck-good-path-bug-2, r=compiler-errors

Don't trim path for `unsafe_op_in_unsafe_fn` lints

Fixes #106126, alternative to #106127.
r? `@ghost` for now.

17 months agoRollup merge of #106045 - RalfJung:oom-nounwind-panic, r=Amanieu
Michael Goulet [Wed, 4 Jan 2023 01:19:26 +0000 (17:19 -0800)]
Rollup merge of #106045 - RalfJung:oom-nounwind-panic, r=Amanieu

default OOM handler: use non-unwinding panic, to match std handler

The OOM handler in std will by default abort. This adjusts the default in liballoc to do the same, using the `can_unwind` flag on the panic info to indicate a non-unwinding panic.

In practice this probably makes little difference since the liballoc default will only come into play in no-std situations where people write a custom panic handler, which most likely will not implement unwinding. But still, this seems more consistent.

Cc `@rust-lang/wg-allocators,` https://github.com/rust-lang/rust/issues/66741