Migrate bootstrap to Clap-based argument parsing
Supercedes #108083
I chose to re-do the work rather than rebase the onto the large changes since the original PR. If it's preferred I can instead force-push the original PR to this version.
cc `@jyn514` `@albertlarsan68`
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
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.
Split `x setup` sub-actions to CLI arguments
Closes#107846
This adds a new `none` profile option which simply skips the `config.toml` step. It also adds `hook` and `vscode` subcommands, for installing the pre-push hook and getting `settings.json` respectively.
Pass arguments to `x` subcommands with `--`
Fixes#107375
Any arguments passed to `x` following `--` are removed and not parsed, instead passed down to subcommands (just the ones listed in the issue, not sure if more are needed). This does not remove the existing `--args` and `--test-args` behaviour, just extends it. It's also not documented in the help, not sure of the best way to format it.
r? `@jyn514`
Allow cleaning individual crates
As a bonus, this stops special casing `clean` in `Builder`.
## Motivation
Cleaning artifacts isn't strictly necessary to get cargo to rebuild; `touch compiler/rustc_driver/src/lib.rs` (for example) will also work. There's two reasons I thought making this part of bootstrap proper was a better approach:
1. `touch` does not *remove* artifacts, it just causes a rebuild. This is unhelpful for when you want to measure how long the compiler itself takes to build (e.g. for https://github.com/rust-lang/rust/issues/65031).
2. It seems a little more discoverable; and I want to extend it in the future to things like `x clean --stage 1 rustc`, which makes it easier to work around https://github.com/rust-lang/rust/issues/76720 without having to completely wipe all the stage 0 artifacts, or having to be intimately familiar with which directories to remove.
Before, the submodule handling was very jank and would update *between two interactive prompts*:
```
; x setup
Building rustbuild
Finished dev [unoptimized] target(s) in 0.05s
Welcome to the Rust project! What do you want to do with x.py?
a) library: Contribute to the standard library
Please choose one (a/b/c/d/e): a
Updating submodule library/backtrace
Submodule 'library/backtrace' (https://github.com/rust-lang/backtrace-rs.git) registered for path 'library/backtrace'
error: you asked `x.py` to setup a new config file, but one already exists at `config.toml`
Build completed unsuccessfully in 0:00:02
```
That's not a great user experience because you need to wait a long time between prompts.
It would be possible to move the submodule handling either before or after the prompt, but it seems
better to just not require submodules to be checked out at all, to minimize the time spend waiting
just to create a new configuration.
`--skip` is inconsistent with the rest of the interface and redundant with `--exclude`.
Fix --exclude to work properly for files and directories rather than having a separate flag.
If someone needs to use --skip for something other than compiletest,
they can use `--test-args --skip` instead.
Stage 1 check has been supported since https://github.com/rust-lang/rust/pull/81064.
#81064 changed the error message for this, but I don't think there's any reason we should prevent using it.
I tested locally and `keep-stage` works fine. Don't give a hard error when trying to use it.
Support `-A`, `-W`, `-D` and `-F` when running `./x.py clippy`
Resolves#97059
This PR adds support for `-A`, `-W`, `-D` and `-F` when running `./x.py clippy`.
implement detail_exit but I'm not sure it is right.
not create new file and write detail exit in lib.rs
replace std::process::exit to detail_exit
that is not related to code runnning.
remove pub
Print stderr consistently
Solves https://github.com/rust-lang/rust/issues/96712
I tried to follow what I perceived as the general consensus for error messages in boostrap i.e messages that were ..
* resulting from an Err(...) =>
* literally called as "Error: ...."
* by the end of the block scope forced to run a panic! or process::exit with a guaranteed non-zero error code.
libtest already supports a "--skip SUBSTRING" arg which excludes any
test names matching SUBSTRING.
This adds a "--skip" argument to compiletest and bootstrap which is
forwarded to libtest.
I didn't know that the `test::` syntax was valid before, and it doesn't
seem to be documented anywhere. Add a test so it doesn't regress accidentally,
and as executable documentation.
There's several places where the x.py command names are matched as
strings, leading to some inconsistencies and opportunities for cleanup.
* Add Format, Clean, and Setup variants to builder::Kind.
* Use Kind to parse the x.py subcommand name (including aliases)
* Match on the subcommand Kind rather than strings when handling
options and help text.
* Several subcommands don't display any paths when run with `-h -v` even
though the help text indicates that they should. Fix this and refactor
so that manually keeping matches in sync isn't necessary.
Fixes#95937
this also fixes a bug where bootstrap would try to use the fake `rustc` binary built by bootstrap -
cargo puts it in a different directory when using `cargo run` instead of x.py
The majority of the code is only used by either rustbuild or
rustc_llvm's build script. Rust_build is compiled once for rustbuild and
once for every stage. This means that the majority of the code in this
crate is needlessly compiled multiple times. By moving only the code
actually used by the respective crates to rustbuild and rustc_llvm's
build script, this needless duplicate compilation is avoided.
This shows up to 5% less instruction counts on multiple benchmarks, and up to
19% wins on the -j1 wall times for rustc self-compilation.
We can afford to spend the extra cycles building LLVM essentially once more for
the x86_64-unknown-linux-gnu CI build today. The builder finishes in around 50
minutes on average, and this adds just 10 more minutes. Given the sizeable
improvements in compiler performance, this is definitely worth it.
Now that Cargo deduplicates diagnostics from different targets, this doesn't flood the console with
duplicate errors.
Note that this doesn't add `--all-targets` in `Builder::cargo` directly because `impl Step for Std`
actually wants to omit `--all-targets` the first time while it's still building libtest.
When passed `--all-targets`, this warns that the option isn't needed, but still continues to compile.
