rustdoc DocFragment rework
Kind of a follow-up of #80119.
A few things are happening in this PR. I'm not sure about the impact on perf though so I'm very interested about that too (if the perf is worse, then we can just close this PR).
The idea here is mostly about reducing the memory usage by relying even more on `Symbol` instead of `String`. The only issue is that `DocFragment` has 3 modifications performed on it:
1. Unindenting
2. Collapsing similar comments into one
3. "Beautifying" (weird JS-like comments handling).
To do so, I saved the information about unindent and the "collapse" is now on-demand (which is why I'm not sure the perf will be better, it has to be run multiple times...).
r? `@jyn514`
Fix search section position on small devices
Fixes#79526.
This is exactly the same issue fixed in 9c36491538 (in https://github.com/rust-lang/rust/pull/79936) but applied to the search section. When the width becomes too small, the search input goes on its own line to get more space, making it go "under" the section following (so either "main" or "search"). The fix is to simply make the section go more under so that it doesn't go over the search input.
r? `@jyn514`
Cleanup markdown span handling, take 2
This PR includes the cleanups made in #80244 except for the removal of `locate()`.
While the biggest conceptual part in #80244 was the removal of `locate()`, it introduced a diagnostic regression.
Additional cleanup:
- Use `RefCell` to avoid building two separate vectors for the links
Work to do:
- [ ] Decide if `locate()` can be simplified by assuming `s` is always in `md`
- [ ] Should probably add some tests that still provide the undesired diagnostics causing #80381
cc `@jyn514` This is the best I can do without patching Pulldown to provide multiple ranges for reference-style links. Also, since `locate` is probably more efficient than `rfind` (at least it's constant time), I decided to not check the link type and just cover every &str as it was before.
Enable ASan, TSan, UBSan for aarch64-apple-darwin.
I confirmed ASan, TSan, UBSan all work for me locally with `clang` on my new Macbook Air.
~This requires https://github.com/rust-lang/llvm-project/pull/86~
Use Array.prototype.filter instead of open-coding
Part of #79052, originally suggested in https://github.com/rust-lang/rust/pull/79052#discussion_r523468743 by `@jyn514`
Besides making main.js smaller (always a plus), this also performs better by using the optimized filter implementation in your browser's JavaScript engine (according to `@GuillaumeGomez,` an 84% performance improvement).
Fix tests that incorrectly used `!@has` instead of `@!has`
The command is ``@!has`,` not `!`@has`.` I don't think these checks were
doing anything before! Ideally we would accept `!`@has`` as well, or at
least fail tests that use `!`@has`.` The current behavior seems to be
silently ignoring the check, which is very confusing.
r? `@GuillaumeGomez`
In which we start to parse const generics defaults
As discussed in this [zulip topic](https://rust-lang.zulipchat.com/#narrow/stream/260443-project-const-generics/topic/const.20generic.20defaults), this PR extracts the parsing parts from `@JulianKnodt's` PR #75384 for a better user-experience using the newly stabilized `min_const_generics` (albeit temporary) as shown in #80507: trying to use default values on const generics currently results in parse errors, as if the user didn't use the correct syntax (which is somewhat true but also misleading).
This PR extracts (and slightly modifies in a couple places) `@JulianKnodt's` parsing code (with attribution if I've done everything correctly), AST and HIR changes, and feature gate setup.
This feature is now marked as "incomplete" and thus will also print out the expected "const generics default values are unstable" error instead of a syntax error. Note that, as I've only extracted the parsing part, the actual feature will not work at all if enabled. There will be ICEs, and inference errors on the const generics default values themselves.
Fixes#80507.
Once this merges, I'll:
- modify the const generics tracking issue to refer to the `const_generics_defaults` gate rather than the older temporary name it uses there.
- create the GH `F-const_generics_defaults` label
r? `@varkor`
- Adds optional default values to const generic parameters in the AST
and HIR
- Parses these optional default values
- Adds a `const_generics_defaults` feature gate
Add edition 2021.
🎆 Happy new ~~year~~ Rust. 🍾
This adds --edition=2021, and updates suggestions about 2018 to say "2018 *or later*".
Related Cargo PR: https://github.com/rust-lang/cargo/pull/8922
---
Edit: This adds the new edition as *unstable*. Without `-Z unstable-options`, `--edition=2021` results in:
```
$ rustc --edition=2021
error: edition 2021 is unstable and only available with -Z unstable-options.
```
Update LLVM
- [GlobalISel][IRTranslator] Fix a crash when the use of an extractvalue is a non-dominated metadata use.
- [asan] Use dynamic shadow memory position on Apple Silicon macOS
r? ``@cuviper``
Fix broken ./x.py install
During my tarball refactorings in https://github.com/rust-lang/rust/pull/79788 I changed the directory layout used by the tarball generation code, and that broke the other parts of rustbuild which hardcoded the paths of those directories. Namely, `./x.py install` relied on the uncompressed copy of the tarball left behind by `fabricate`/`rust-installer`, causing https://github.com/rust-lang/rust/issues/80494.
While the easy fix for https://github.com/rust-lang/rust/issues/80494 would've been to just update the hardcoded paths to match the new structure, that fix would leave us in the same situation if we were to change the directory layout again in the future. Instead I refactored the code to return a `GeneratedTarball` struct as the output of all the dist steps, and I put all the paths the rest of rustbuild needs to care about in its fields. That way, future changes to `src/bootstrap/tarball.rs` will not break other stuff.
This PR is best reviewed commit-by-commit.
r? `@Mark-Simulacrum`
`@rustbot` modify labels: beta-nominated beta-accepted T-release
