sagudev

Artifacts are already made in build job all the time no matter if WPT is requested or not. So we only need to "skip" build task if it is already...

or maybe limit concurrent number of docs building: ```yaml concurrency: group: ${{ github.workflow }}-${{ github.ref }} cancel-in-progress: true ```

Now that https://github.com/servo/servo/pull/31441 landed I plan to add `mach bootstrap-nightly-rust` for ASan needs.

@mrobinson could we push this forward?

Copy trick does not work on mac (due to otool and gstreamer), so I guess we need to do this in right, but more complicated way or introduce even more...

`RUST_LOG=script,webgpu=trace` gives: ``` ▶ CRASH [expected ERROR] /_webgpu/webgpu/cts.https.html?q=webgpu:shader,execution,expression,call,builtin,dot:f32_vec4:* │ │ [2024-02-21T17:10:43Z WARN script::timers] Resuming an already resumed timer. │ [2024-02-21T17:10:43Z INFO script::dom::bindings::refcounted] growing refcounted references by 3 │ [2024-02-21T17:10:43Z INFO...

This might be due to caching infra in WebGPU CTS for `webgpu:shader` tests.

Adding `error!("discovering: {}", self.url);` as the first line to: https://github.com/servo/servo/blob/4849ba901efab9304d71b316ec9e0d7e98e1993b/components/script/script_module.rs#L632-L637 makes crash not reproducible, so I think there might be some data-race happening here.

Extract from [full log](https://github.com/servo/servo/files/14388858/more_logs.txt): ``` ▶ CRASH [expected ERROR] /_webgpu/webgpu/cts.https.html?q=webgpu:shader,execution,expression,call,builtin,dot:f32_vec4:* │ │ [2024-02-23T07:05:17Z ERROR script::script_module] Going to advance and finish for: https://web-platform.test:8443/_webgpu/webgpu/common/framework/test_config.js │ [2024-02-23T07:05:17Z ERROR script::script_module] Going to advance and...

Using `std::thread::yield_now()` in `find_first_parse_error` mostly solves the issue, but this is not permanent solution.