Introduce infrastructure for generating target docs

[Nilstrieb]

See #120745

It's highly unlikely that the format is optimal, but it's okay at least and can always be improved. Mostly posting this to get something working so we can continue.

Use TARGET_CHECK_ONLY=0 to actually build the book instead of just checking the new docs.

r? @davidtwco

[rustbot]

These commits modify the Cargo.lock file. Unintentional changes to Cargo.lock can be introduced when switching branches and rebasing PRs.

If this was unintentional then you should revert the changes before this PR is merged.
Otherwise, you can ignore this comment.

[Nilstrieb]

this is a very funny failure, sadly I cannot reproduce it locally.

[ehuss]

@Nilstrieb The issue is when it runs --print cfg --target csky-unknown-linux-gnuabiv2 it fails because LLVM 16 does not support that target.

It tries to print the error "No available targets are compatible with triple "csky-unknown-linux-gnuabiv2"".

However, one of the arguments is a SmallCStr which is created from a CStr. I believe there is a bug here where SmallCStr is created without the trailing null byte. It then later calls as_c_str which does from_bytes_with_nul_unchecked, but there is no null.

I can maybe try to post a fix for the panic later tonight. I don't know how best to handle the problem with LLVM 16 not supporting the target. Perhaps target-docs could check stderr for the "no available targets" message, and just skip that target?

[ehuss]

Posted #121125 so it at least doesn't ICE.

[davidtwco]

I really like this approach. I think the only feedback I've got is about where some of the specific bits of information go.

This approach shines with the overlapping documentation based with patterns, but if you need to have some information for each target (like tier, maintainers, descriptive name, supports std, etc.) and have to make documentation for every target anyway, then you lose some of that benefit.

I think I'd like to see us experiment with putting some of that information into the target specification, and then have a way of dumping that from the compiler (like you do with the cfgs now).

[Nilstrieb]

I like that. We could even print the short description with --print target-list --verbose :). I will take a look.

[Nilstrieb]

this is blocked on work on the tracking issue

[rustbot]

These commits modify compiler targets.
(See the Target Tier Policy.)

[workingjubilee]

#123010 (comment) @Nilstrieb I think this was yours, unfortunately.

@bors r-

[workingjubilee]

@rollup=iffy

[dpaoliello]

#123010 (comment) @Nilstrieb I think this was yours, unfortunately.

Checking this change out on Windows was failing - * is not a valid character for file names.

[Nilstrieb]

I see, Windows exists. Good point. I changed the filename to use _ instead of * for wildcards. Since that collides with _ present in some targets (like that one obscure architecture), I instead moved the pattern to the frontmatter, where asterisks are allowed (supposedly). target-docs enforces that the file name is consistent with the pattern.
I didn't rebase, so the force push compare page is accurate.
@rustbot ready

[onur-ozkan]

Changes in the last push seems solid. Can we add unit test in parse/tests.rs for the logic here?

rust/src/tools/target-docs/src/parse.rs

Lines 87 to 93 in 1219d99

	let expected_file_name = frontmatter.pattern.replace('*', "_");
	if expected_file_name != file_name {
	bail!(
	"`target pattern does not match file name. file name must be pattern with `*` replaced with `_`.\n\
	Expected file name `{expected_file_name}.md`"
	);
	}

[klensy]

Well, why glob-match crate? There already glob and globset in workspace, and current usage should be doable via https://docs.rs/glob/latest/glob/struct.Pattern.html#method.matches.

[davidtwco]

r=me with this comment resolved and if it isn't hard to use glob or globset instead, then that would be good too.

[Nilstrieb]

@onur-ozkan lol, thanks for mentioning the tests, I completely forgot about them and had to fix them

Another thing: I had to add this check:
https://github.com/rust-lang/rust/pull/121051/files#diff-b6fc213859d4da4176033b362381724914fea88a6c04122c0048ab5b76382343R1158

It was already kinda broken, lint-docs emitted warnings about things not quite matching up, so I don't think this should break anything and if it does it's good that it's broken.

[rust-log-analyzer]

The job x86_64-gnu-llvm-17 failed! Check out the build log: (web) (plain)

Click to see the possible cause of the failure (guessed by this bot)

# be missing.
ENV IS_NOT_LATEST_LLVM 1


# Ubuntu LLVM 17 does not have support for experimental targets like csky.
ENV TARGET_DOCS_SKIP_TARGETS "csky-unknown-linux-gnuabiv2,csky-unknown-linux-gnuabiv2hf"
# Using llvm-link-shared due to libffi issues -- see #34486
ENV RUST_CONFIGURE_ARGS \
      --build=x86_64-unknown-linux-gnu \
      --llvm-root=/usr/lib/llvm-17 \
---
------
 > importing cache manifest from ghcr.io/rust-lang-ci/rust-ci-cache:1735714b6b19ef0a4e8a4e6747a678dadee568ebdeb9719db617faccdb946b2eba73802683684b5fe46c7560fac125406846a9bd7e9383f3421f89057efbc7ae:
------
##[endgroup]
Setting extra environment values for docker:  --env ENABLE_GCC_CODEGEN=1 --env GCC_EXEC_PREFIX=/usr/lib/gcc/
[CI_JOB_NAME=x86_64-gnu-llvm-17]
---
sccache: Starting the server...
##[group]Configure the build
configure: processing command line
configure: 
configure: build.configure-args := ['--build=x86_64-unknown-linux-gnu', '--llvm-root=/usr/lib/llvm-17', '--enable-llvm-link-shared', '--set', 'rust.thin-lto-import-instr-limit=10', '--set', 'change-id=99999999', '--enable-verbose-configure', '--enable-sccache', '--disable-manage-submodules', '--enable-locked-deps', '--enable-cargo-native-static', '--set', 'rust.codegen-units-std=1', '--set', 'dist.compression-profile=balanced', '--dist-compression-formats=xz', '--disable-dist-src', '--release-channel=nightly', '--enable-debug-assertions', '--enable-overflow-checks', '--enable-llvm-assertions', '--set', 'rust.verify-llvm-ir', '--set', 'rust.codegen-backends=llvm,cranelift,gcc', '--set', 'llvm.static-libstdcpp', '--enable-new-symbol-mangling']
configure: target.x86_64-unknown-linux-gnu.llvm-config := /usr/lib/llvm-17/bin/llvm-config
configure: llvm.link-shared     := True
configure: rust.thin-lto-import-instr-limit := 10
configure: change-id            := 99999999
---
  Downloaded boml v0.3.1
   Compiling boml v0.3.1
   Compiling y v0.1.0 (/checkout/compiler/rustc_codegen_gcc/build_system)
    Finished `release` profile [optimized] target(s) in 3.57s
     Running `/checkout/obj/build/x86_64-unknown-linux-gnu/stage1-codegen/x86_64-unknown-linux-gnu/release/y test --use-system-gcc --use-backend gcc --out-dir /checkout/obj/build/x86_64-unknown-linux-gnu/stage1-tools/cg_gcc --release --mini-tests --std-tests`
Using system GCC
[BUILD] example
[AOT] mini_core_hello_world
/checkout/obj/build/x86_64-unknown-linux-gnu/stage1-tools/cg_gcc/mini_core_hello_world
abc
---
running 46 tests
error: test failed, to rerun pass `--lib`

Caused by:
  process didn't exit successfully: `/checkout/obj/build/bootstrap/debug/deps/bootstrap-de0d1d488e5b0274 --test-threads=1 --quiet -Z unstable-options --format json` (exit status: 1)
note: test exited abnormally; to see the full output pass --nocapture to the harness.
.....  local time: Thu Apr  4 19:35:20 UTC 2024
  network time: Thu, 04 Apr 2024 19:35:20 GMT
##[error]Process completed with exit code 1.
Post job cleanup.