The crate hash is needed:
- if debug assertions are enabled, or
- if incr. comp. is enabled, or
- if metadata is being generated, or
- if `-C instrumentation-coverage` is enabled.
This commit avoids computing the crate hash when these conditions are
all false, such as when doing a release build of a binary crate.
It uses `Option` to store the hashes when needed, rather than
computing them on demand, because some of them are needed in multiple
places and computing them on demand would make compilation slower.
The commit also removes `Owner::hash_without_bodies`. There is no
benefit to pre-computing that one, it can just be done in the normal
fashion.
Use `partial_cmp` to implement tuple `lt`/`le`/`ge`/`gt`
In today's implementation, `(A, B)::gt` contains calls to *both* `A::eq` *and* `A::gt`.
That's fine for primitives, but for things like `String`s it's kinda weird -- `(String, usize)::gt` has a call to both `bcmp` and `memcmp` (<https://rust.godbolt.org/z/7jbbPMesf>) because when `bcmp` says the `String`s aren't equal, it turns around and calls `memcmp` to find out which one's bigger.
This PR changes the implementation to instead implement `(A, …, C, Z)::gt` using `A::partial_cmp`, `…::partial_cmp`, `C::partial_cmp`, and `Z::gt`. (And analogously for `lt`, `le`, and `ge`.) That way expensive comparisons don't need to be repeated.
Technically this is an observable change on stable, so I've marked it `needs-fcp` + `T-libs-api` and will
r? rust-lang/libs-api
I'm hoping that this will be non-controversial, however, since it's very similar to the observable changes that were made to the derives (#81384#98655) -- like those, this only changes behaviour if a type overrode behaviour in a way inconsistent with the rules for the various traits involved.
(The first commit here is #108156, adding the codegen test, which I used to make sure this doesn't regress behaviour for primitives.)
Zulip conversation about this change: <https://rust-lang.zulipchat.com/#narrow/stream/219381-t-libs/topic/.60.3E.60.20on.20Tuples/near/328392927>.
Rollup of 7 pull requests
Successful merges:
- #106440 (Ignore files in .gitignore in tidy)
- #108613 (Remove `llvm.skip-rebuild` option)
- #108616 (Sync codegen defaults with compiler defaults and add a ping message so they stay in sync)
- #108618 (Rename `src/etc/vscode_settings.json` to `rust_analyzer_settings.json`)
- #108626 (rustdoc-json: switch from HashMap to FxHashMap to fix non-determinism)
- #108744 (Don't ICE when encountering bound var in builtin copy/clone bounds)
- #108749 (Clean up rustdoc-js tester.js file)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
rustdoc-json: switch from HashMap to FxHashMap to fix non-determinism
Using `HashMap` in `rustdoc_json_types::Crate` were causing creating randomly ordered objects in the json doc files. Which might cause problems to people who are doing comparison on those files specially in CI pipelines. See https://github.com/rust-lang/rust/issues/103785#issuecomment-1307425590
This PR fixes that issue and extends the coverage of `tests/run-make/rustdoc-verify-output-files` testing ability.
Remove `llvm.skip-rebuild` option
This was added to in 2019 to speed up rebuild times when LLVM was modified. Now that download-ci-llvm exists, I don't think it makes sense to support an unsound option like this that can lead to miscompiles; and the code cleanup is nice too.
r? `@Mark-Simulacrum` cc `@varkor` #65612
Ignore files in .gitignore in tidy
- Switch from `walkdir` to `ignore`. This required various changes to make `skip` thread-safe.
- Ignore `build` anywhere in the source tree, not just at the top-level. We support this in bootstrap, we should support it in tidy too.
As a nice side benefit, this also makes tidy a bit faster.
Before:
```
; hyperfine -i '"/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0-tools-bin/rust-tidy" "/home/gh-jyn514/rust2" "/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0/bin/cargo" "/home/gh-jyn514/rust2/build" "32"'
Benchmark 1: "/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0-tools-bin/rust-tidy" "/home/gh-jyn514/rust2" "/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0/bin/cargo" "/home/gh-jyn514/rust2/build" "32"
Time (mean ± σ): 1.080 s ± 0.008 s [User: 2.616 s, System: 3.243 s]
Range (min … max): 1.069 s … 1.099 s 10 runs
```
After:
```
; hyperfine '"/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0-tools-bin/rust-tidy" "/home/gh-jyn514/rust2" "/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0/bin/cargo" "/home/gh-jyn514/rust2/build" "32"'
Benchmark 1: "/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0-tools-bin/rust-tidy" "/home/gh-jyn514/rust2" "/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0/bin/cargo" "/home/gh-jyn514/rust2/build" "32"
Time (mean ± σ): 705.0 ms ± 1.4 ms [User: 3179.1 ms, System: 1517.5 ms]
Range (min … max): 702.3 ms … 706.9 ms 10 runs
```
r? `@the8472`
`WalkBuilder` handles top-level paths differently than `fn walk` used
to: it doesn't run the `skip` function to determine if it should be
skipped, instead assuming the top-level function is always included.
This is a reasonable assumption; adapt our code so it doesn't make
pointless calls to `walk`.
- Switch from `walkdir` to `ignore`. This required various changes to
make `skip` thread-safe.
- Ignore `build` anywhere in the source tree, not just at the top-level.
We support this in bootstrap, we should support it in tidy too.
As a nice side benefit, this also makes tidy a bit faster.
Before:
```
; hyperfine -i '"/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0-tools-bin/rust-tidy" "/home/gh-jyn514/rust2" "/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0/bin/cargo" "/home/gh-jyn514/rust2/build" "32"'
Benchmark 1: "/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0-tools-bin/rust-tidy" "/home/gh-jyn514/rust2" "/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0/bin/cargo" "/home/gh-jyn514/rust2/build" "32"
Time (mean ± σ): 1.080 s ± 0.008 s [User: 2.616 s, System: 3.243 s]
Range (min … max): 1.069 s … 1.099 s 10 runs
```
After:
```
; hyperfine '"/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0-tools-bin/rust-tidy" "/home/gh-jyn514/rust2" "/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0/bin/cargo" "/home/gh-jyn514/rust2/build" "32"'
Benchmark 1: "/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0-tools-bin/rust-tidy" "/home/gh-jyn514/rust2" "/home/gh-jyn514/rust2/build/aarch64-unknown-linux-gnu/stage0/bin/cargo" "/home/gh-jyn514/rust2/build" "32"
Time (mean ± σ): 705.0 ms ± 1.4 ms [User: 3179.1 ms, System: 1517.5 ms]
Range (min … max): 702.3 ms … 706.9 ms 10 runs
```
rustc_middle: Remove trait `DefIdTree`
This trait was a way to generalize over both `TyCtxt` and `Resolver`, but now `Resolver` has access to `TyCtxt`, so this trait is no longer necessary.
Desugaring of drop and replace at MIR build
This commit desugars the drop and replace deriving from an
assignment at MIR build, avoiding the construction of the
`DropAndReplace` terminator (which will be removed in a following PR).
In order to retain the same error messages for replaces a new
`DesugaringKind::Replace` variant is introduced.
The changes in the borrowck are also useful for future work in moving drop elaboration
before borrowck, as no `DropAndReplace` would be present there anymore.
Notes on test diffs:
* `tests/ui/borrowck/issue-58776-borrowck-scans-children`: the assignment deriving from the desugaring kills the borrow.
* `tests/ui/async-await/async-fn-size-uninit-locals.rs`, `tests/mir-opt/issue_41110.test.ElaborateDrops.after.mir`, `tests/mir-opt/issue_41888.main.ElaborateDrops.after.mir`: drop elaboration generates (or reads from) a useless drop flag due to an issue with the dataflow analysis. Will be fixed independently by https://github.com/rust-lang/rust/pull/106430.
See https://github.com/rust-lang/rust/pull/104488 for more context
Apply BOLT optimizations without rebuilding LLVM
This PR adds an explicit BOLT bootstrap step which applies BOLT on the fly when LLVM artifacts are copied to a sysroot (it only does this once per bootstrap invocation, the result is cached). This avoids one LLVM rebuild in the Linux CI dist build.
r? `@jyn514`
Make compressed rmeta contain compressed data length after header
Fixes#90056, which is caused by link.exe introducing padding to the `.rustc` section, since it assumes this will have no effect besides allowing it to possibly use the extra space in future links.
Rollup of 7 pull requests
Successful merges:
- #108627 (Properly colorize multi-part suggestions in the same line)
- #108632 (Omit unchanged options from config.toml in `configure.py`)
- #108715 (Remove unclosed_delims from parser)
- #108723 (rustdoc: function signature search with traits in `where` clause)
- #108724 (field is not used outside the crate)
- #108734 (rustdoc: Note in a type's layout/size if it is uninhabited)
- #108736 (Remove `allow(potential_query_instability)` from `ast_passes`)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Omit unchanged options from config.toml in `configure.py`
Leaves section tags, but removes options that are unchanged.
Change in `config.toml.example` is to prevent comments from sneaking in by being directly after a section tag
closes#108612
Don't call `temporary_scope` twice.
`mirror_expr_inner` calls `temporary_scope`. It then calls `make_mirror_unadjusted` which makes an identical call to `temporary_scope`.
This commit changes the `mirror_expr_inner` to get the `temp_lifetime` out of the expression produced by `make_mirror_unadjusted`, similar to how it currently gets the type.
r? `@cjgillot`
rustc_expand: make proc-macro derive error translatable
kept this tiny so as to point to it as an example in rustc-dev-guide
`@rustbot` label +A-translation
