It looks like this job was intending to run all of the `needs-matching-clang`
tests (since they don't run without `RUSTBUILD_FORCE_CLANG_BASED_TESTS`), but
over time developed two problems:
- The tests it cares about were moved from run-make-fulldeps to run-make.
- Some of the relevant tests don't actually have "clang" in their name.
Switching to run-make solves the first problem, but we still don't run the
tests without "clang" in their name, because some of them are currently broken.
Add a custom Command wrapper to `run-make-support`
This should make it easier to make sure that we check process exit codes, and it should also make checking of stdout/stderr less verbose and more explicit in run-make tests. I prefer the `run()/run_fail().assert(...)` style to something like `run_fail_assert_exit_code`, because the former is more composable.
Regarding https://github.com/rust-lang/rust/issues/125747, I'm not sure if we really need a custom trait, I think that we can get far enough with just `Deref` on the `Cc/Clang/Rustc/Rustdoc/...` structs. But now that these structs don't even need `command_output` anymore, I think that they are fine-ish as they are with the macro.
Related issues: https://github.com/rust-lang/rust/issues/125617, https://github.com/rust-lang/rust/issues/125747
Fixes: https://github.com/rust-lang/rust/issues/125617 (because `command_output` is no longer a public method)
r? `@jieyouxu`
portable-simd: add test for non-power-of-2 bitmask
`@calebzulawski` is that the intended behavior? Specifically for arrays, the bitmask `[1, 0, 0, 1, 0, 0, 1, 0, 1, 0]` becomes
- `[0b01001001, 0b01]` on little endian
- `[0b10010010, 0b10]` on big endian
Change how runmake v2 tests are executed
This PR makes execution of v2 runmake tests more sane, by executing each test in a temporary directory by default, rather than running it inside `tests/run-make`. This will have.. a lot of conflicts.
Fixes: https://github.com/rust-lang/rust/issues/126080
Closes https://github.com/rust-lang/rust/issues/125726, because it removes `tmp_dir`, lol.
r? `@jieyouxu`
try-job: x86_64-msvc
fix: Fix generated markers not being patchable in package.json
I think the newlines are tripping up CI, so I hope clearing the property to leave an empty object won't make VSCode made here.
Port `tests/run-make-fulldeps/hotplug_codegen_backend` to ui-fulldeps
This is the last remaining run-make-fulldeps test, which means I actually had to leave behind a dummy README file to prevent compiletest from complaining about a missing directory.
(Removing the run-make-fulldeps suite entirely is non-trivial, so I intend to do so in a separate PR after this one.)
---
I wasn't sure about adding a new kind of aux build just for this one test, so I also tried to just port this test from Makefile to [rmake](https://github.com/rust-lang/rust/issues/121876) instead.
But I found that I couldn't get rmake to fully work for a run-make-fulldeps test, which convinced me that getting rid of run-make-fulldeps is worthwhile.
r? `@jieyouxu`
Add debugging utils and comments to Fuchsia scripts
This should help when debugging a failure in the Fuchsia build in CI.
I plan to follow up with a PR to the testing section of the dev guide with more details, along with more improvements happening in the Fuchsia repo itself.
try-job: x86_64-gnu-integration
Make html rendered by rustdoc allow searching non-English identifier / alias
Fix alias search result showing `undefined` description.
Inspired by https://github.com/rust-lang/mdBook/issues/2393 .
Not sure if it's worth it adding full-text search functionality to rustdoc rendered html.
