This is needed for the next commit, which needs access to the `cx` from
within the `decorate` closure.
- Change `as_local_hir_id` to an associated function, since it only
needs a `TyCtxt`
- Change `source_span_for_markdown_range` to only take a `TyCtxt`
Upgrade to LLVM 12
This implements the necessary adjustments to make rustc work with LLVM 12. I didn't encounter any major issues so far.
r? `@cuviper`
Pass `CrateNum` by value instead of by reference
It's more idiomatic to pass a small Copy type by value and `CrateNum` is
half the size of `&CrateNum` on 64-bit systems. The memory use change is
almost certainly insignificant, but why not!
Revert `Vec::spare_capacity_mut` impl to prevent pointers invalidation
The implementation was changed in #79015.
Later it was [pointed out](https://github.com/rust-lang/rust/issues/81944#issuecomment-782849785) that the implementation invalidates pointers to the buffer (initialized elements) by creating a unique reference to the buffer. This PR reverts the implementation.
r? ```@RalfJung```
Improve page load performance in rustdoc
Add an explicit height to icons (which already had an explicit width) to allow browsers to lay out the page more accurately before the icons have been loaded. https://web.dev/optimize-cls/.
Add min-width: 115px to the crate search dropdown. When the HTML first loads, this dropdown includes only the text "All crates." Later, JS loads the items underneath it, some of which are wider. That causes the dropdown to get wider, causing a distracting reflow. This sets a min-width based on the size that the dropdown eventually becomes based on the crates on doc.rust-lang.org, reducing page movement during load.
Add font-display: swap. Per https://web.dev/font-display/, this prevents "flash of invisible text" during load by using a system font until the custom font is available. I've noticed this flash of invisible text occasionally when reading Rust docs. Note that users without cached fonts will see text, and then see it reflow. For `docs.rust-lang.org`, [setting caching headers will help a lot](https://github.com/rust-lang/simpleinfra/issues/62).
Generated output at https://jacob.hoffman-andrews.com/rust/flow-improvements/std/string/struct.String.html.
Pass callee expr to `confirm_builtin_call`. This removes a partial
pattern match in `confirm_builtin_call` and the `panic` in the `else`
branch. The diff is large because of indentation changes caused by
removing the if-let.
Add `-Z unpretty` flags for the AST
Implements rust-lang/compiler-team#408.
Builds on #82269, but if that PR is rejected or stalls out, I can implement this without #82269.
cc rust-lang/rustc-dev-guide#1062
It's more idiomatic to pass a small Copy type by value and `CrateNum` is
half the size of `&CrateNum` on 64-bit systems. The memory use change is
almost certainly insignificant, but why not!
On reflection on the issue in https://github.com/rust-lang/rust/pull/79540#discussion_r572572280, I think the bug was actually using the `compiler/` filter, not using `--author=bors`. 9a1d6174c9 has no CI artifacts because it was merged as part of a rollup:
```
$ curl -I https://ci-artifacts.rust-lang.org/rustc-builds/96e843ce6ae42e0aa519ba45e148269de347fd84/rust-std-nightly-x86_64-unknown-linux-gnu.tar.xz
HTTP/2 404
```
So 9a1d6174c9 is the correct commit to download, and that's what `--author=bors` does:
$ git log --author=bors 4aec8a5da5
commit 9a1d6174c9
Ideally it would look for "the most recent bors commit not followed by a change to `compiler/`", which would exclude things like documentation changes and avoid redownloading more than necessary, but
- Redownloading isn't the end of the world,
- That metric is hard to implement, and
- Documentation-only or library-only changes are very rare anyway since they're usually rolled up with changes to the compiler.
Move the check for potentially forgotten `return` in a tail expression
of arbitrary expressions into the coercion error branch to avoid
computing unncessary coercion checks on successful code.
Follow up to #81458.
We previously used -Os to disable this pass, but since
https://reviews.llvm.org/D91673 this no longer works. Explicitly
disable it using -Cllvm-args instead.
This works around a design defect in the LLVM 12 pass builder
implementation. In LLVM 13, the PreLink ThinLTO pipeline properly
respects the OptimizerLastEPCallbacks.
Update tracing to 0.1.25
* Update tracing from 0.1.18 to 0.1.25
* Update tracing-subscriber from 0.2.13 to 0.2.16
* Update tracing-tree from 0.1.6 to 0.1.8
* Add pin-project-lite to the list of allowed dependencies (it is now a direct dependency of tracing).