Rollup of 6 pull requests
Successful merges:
- #105465 (Improve top-level docs)
- #105872 (Suggest remove last method call when type coerce with expected type)
- #106032 (std: only use LFS function on glibc)
- #106078 (Provide more context on FileCheck failures)
- #106100 (Codegen test for derived `<` on trivial newtype [TEST ONLY])
- #106109 (rustdoc: make line number CSS for doc comment and scraped the same)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Codegen test for derived `<` on trivial newtype [TEST ONLY]
I originally wrote this for #106065, but the libcore changes there aren't necessarily a win.
So I pulled out this test to be its own PR since it's important (see https://github.com/rust-lang/rust/pull/105840#discussion_r1056030324) and well-intentioned changes to core or the derive could accidentally break it without that being obvious (other than by massive unexplained perf changes).
Provide more context on FileCheck failures
FileCheck provides 5 lines of context by default. This is often insufficient to analyze failures that happen in CI. Increase the amount of context to 100 lines.
Improve top-level docs
See a detailed explanation in the commit messages. This is a companion PR to https://github.com/rust-lang/rustc-dev-guide/pull/1528.
* Link to other resources instead of inlining their information
* Remove ancient and outdated reference to `config.mk`
* Suggest `profile = "user"` in the README
* Add detail about dependencies from the dev-guide
* Link to CONTRIBUTING.md instead of rustc-dev-guide in the main readme
* Link to `std-dev-guide` in CONTRIBUTING.md
char: µoptimise UTF-16 surrogates decoding
According to Godbolt¹, on x86_64 using binary and produces slightly
better code than using subtraction. Readability of both is pretty
much equivalent so might just as well use the shorter option.
¹ https://rust.godbolt.org/z/9jM3ejbMx
The other places are more accurate and up-to-date.
- Link to `std-dev-guide` in CONTRIBUTING.md
Thom and Mara said the guide is in reasonably good shape, and it's tailored more closely to people working on the standard library.
- Link to CONTRIBUTING.md instead of rustc-dev-guide in the main readme
CONTRIBUTING.md has more information and also links the std-dev-guide.
- Link to forge for the list of tested platforms; the one in the readme
was hopelessly out of date.
rustdoc: remove no-op CSS from `.docblock-short`
The rules `overflow:hidden` and `text-overflow:ellipses` only have an effect if overflow occurs, which cannot happen because it will just line wrap instead.
These rules definitely became obsolete by https://github.com/rust-lang/rust/pull/77699, when the stylesheet was decidedly changed to have line wrapping in short docblocks, but given the bug it was fixing, this probably got broken earlier.
fix vec::IntoIter::drop on high-alignment ZST
This fixes a soundness bug: IntoIter would call `drop_in_place` on an insufficiently aligned pointer. So if a ZST with alignment greater 1 had drop glue, that would be called with an unaligned reference. Since https://github.com/rust-lang/rust/pull/103957, Miri checks alignment even if the type does not have drop glue, which is how this bug was found.
r? ``@thomcc``
rustdoc: use a more evocative name for CSS/JS `#titles`
This renames the ID, which is only used in search results, to `#search-tabs`. Also changes the `.count` to a span, so it doesn't need its display mode to be overridden.
Update cargo
7 commits in c994a4a638370bc7e0ffcbb0e2865afdfa7d4415..2381cbdb4e9b07090f552d34a44a529b6e620e44 2022-12-18 21:50:58 +0000 to 2022-12-23 12:19:27 +0000
- fix: deduplicate dependencies by artifact target (rust-lang/cargo#11478)
- Add warning if potentially-scrapable examples are skipped due to dev-dependencies (rust-lang/cargo#11503)
- Don't scrape examples from library targets by default (rust-lang/cargo#11499)
- Stabilize terminal-width (rust-lang/cargo#11494)
- Make sure that hash of `SourceId` is stable (rust-lang/cargo#11501)
- Use workspace lockfile when running `cargo package` and `cargo publish` (rust-lang/cargo#11477)
- Show `--help` if there is no man page for subcommand (rust-lang/cargo#11473)
r? `@ghost`
Rollup of 6 pull requests
Successful merges:
- #105661 (implement the skeleton of the updated trait solver)
- #105853 (Make the pre-push script work on directories with spaces)
- #106043 (Move tests)
- #106048 (Run `tidy` in its own job in PR CI)
- #106055 (Check arg expressions properly on error in `confirm_builtin_call`)
- #106067 (A few metadata nits)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
The rules `overflow:hidden` and `text-overflow:ellipses` only have an
effect if overflow occurs, which cannot happen because it will just line
wrap instead.
These rules definitely became obsolete by
https://github.com/rust-lang/rust/pull/77699, when the stylesheet was
decidedly changed to have line wrapping in short docblocks, but given the
bug it was fixing, this probably got broken earlier.
7 commits in c994a4a638370bc7e0ffcbb0e2865afdfa7d4415..2381cbdb4e9b07090f552d34a44a529b6e620e44
2022-12-18 21:50:58 +0000 to 2022-12-23 12:19:27 +0000
- fix: deduplicate dependencies by artifact target (rust-lang/cargo#11478)
- Add warning if potentially-scrapable examples are skipped due to dev-dependencies (rust-lang/cargo#11503)
- Don't scrape examples from library targets by default (rust-lang/cargo#11499)
- Stabilize terminal-width (rust-lang/cargo#11494)
- Make sure that hash of `SourceId` is stable (rust-lang/cargo#11501)
- Use workspace lockfile when running `cargo package` and `cargo publish` (rust-lang/cargo#11477)
- Show `--help` if there is no man page for subcommand (rust-lang/cargo#11473)
Run `tidy` in its own job in PR CI
This duplicates mingw-check into two jobs where one job runs `tidy` only while the other job does not. The tidy job will not cancel other jobs on failure.
Make the pre-push script work on directories with spaces
As a secondary benefit, it's also a lot simpler.
cc `@caass` - would love to have a review if you have time :)
implement the skeleton of the updated trait solver
cc ```@rust-lang/initiative-trait-system-refactor```
This is mostly following the architecture discussed in the types team meetup.
After discussing the desired changes for the trait solver, we encountered cyclic dependencies between them. Most notably between changing evaluate to be canonical and returning inference constraints. We cannot canonicalize evaluate without returning inference constraints due to coinductive cycles. However, caching inference constraints also relies on canonicalization. Implementing both of these changes at once in-place is not feasible.
This somewhat closely mirrors the current `evaluate` implementation with the following notable differences:
- it moves `project` into the core solver, allowing us to correctly deal with coinductive projections (will be required for implied bounds, perfect derive)
- it changes trait solver overflow to be non-fatal (required to backcompat breakage from changes to the iteration order of nested goals, deferred projection equality, generally very useful)
- it returns inference constraints and canonicalizes inputs and outputs (required for a lot things, most notably merging fulfill and evaluate, and deferred projection equality)
- it is implemented to work with lazy normalization
A lot of things aren't yet implemented, but the remaining FIXMEs should all be fairly self-contained and parallelizable. If the architecture looks correct and is what we want here, I would like to quickly merge this and then split the work.
r? ```@compiler-errors``` / ```@rust-lang/types``` :3
Take revision into account in non-incremental-mode `// incremental` tests
A UI test I added in #105983 confusingly [failed](https://github.com/rust-lang/rust/pull/106031#issuecomment-1362558067) in a merge because two different revisions raced with each other for the same incremental directory for a (non-incremental-mode, i.e. `src/test/ui`) UI test.
Let's take the revision name into account when generating an incremental directory so that other UI tests that combine `// incremental` and `// revisions` won't race and cause possible flakiness in CI.
According to Godbolt¹, on x86_64 using binary and produces slightly
better code than using subtraction. Readability of both is pretty
much equivalent so might just as well use the shorter option.
¹ https://rust.godbolt.org/z/9jM3ejbMx
Remove old version of `cfg-if` by bumping `packed_simd_2` and `getrandom v0.2` versions
```console
> cargo update --package packed_simd_2 --package getrandom@0.2.0
Updating crates.io index
Removing cfg-if v0.1.10
Updating getrandom v0.2.0 -> v0.2.8
Updating packed_simd_2 v0.3.4 -> v0.3.8
```
`packed_simd_2` is only used as a dependency of `bytecount` which in turn is only used by `rustfmt`. I can't see any issue with the minor version bump.
The bigger jump is `getrandom@0.2.0` which is used by a number of things, but 0.2.8 is still semver compatible and there doesn't seem to be any worrying changes (see the [changelog](https://github.com/rust-random/getrandom/blob/master/CHANGELOG.md)). The only breaking change are the removal of XP, stdweb and CloudAbi support but these are not host targets and rustc lacks support for them in any case (stdweb development was [abandonded](https://rustsec.org/advisories/RUSTSEC-2020-0056.html), XP is [tier 3 and `no_std`](https://doc.rust-lang.org/beta/rustc/platform-support.html#tier-3) and CloubAbi is [no longer supported](https://github.com/rust-lang/rust/pull/78439)).
Note that this doesn't affect `getrandom@0.1.16` which is what std depends on and which is already using the latest version of `cfg-if` (besides, there are already plans to remove that entirely).
FileCheck provides 5 lines of context by default. This is often
insufficient to analyze failures that happen in CI. Increase the
amount of context to 100 lines.
