CI: Disable the failed macOS Builds

[lupyuen]

Summary

This PR disables the failed macOS Builds from CI Checks sim-02 and sim-03, so that the macOS CI Checks will complete successfully in the NuttX Mirror Repo:

• sim:libcxxtest: macOS doesn't support cxxabi.h

• sim:lua: macOS doesn't support pipe2()

• sim:note: macOS doesn't support ld --wrap

• sim:quickjs: macOS fails to patch quickjs-libc.c

• sim:rpproxy_virtio: macOS fails at uintmax_t unsigned long long vs unsigned long

• sim:rpserver_virtio: macOS fails at uintmax_t unsigned long long vs unsigned long

Impact

With this PR, macOS CI Checks will complete successfully in the NuttX Mirror Repo.

Testing

The macOS CI Checks completed successfully in our Test Repo:
https://github.com/lupyuen5/label-nuttx/actions/runs/11828738112

[nuttxpr]

[Experimental Bot, please feedback here]

Yes, this PR appears to meet the NuttX requirements, though some sections could be more explicitly filled out. Here's a breakdown:

Strengths:

• Clear Summary: The summary explains the "why" (failed macOS builds), "what" (disabling specific tests), and "how" (by excluding them from the CI).
• Impact is Addressed: The impact on the user, build process, hardware, documentation, security, and compatibility is adequately addressed, albeit briefly. The main impact is a positive one: fixing the broken CI.
• Testing Provided: A link to a successful test run is provided, which demonstrates the effectiveness of the change.

Areas for Improvement (While Minor, Worth Considering):

• Explicitly State "No Impact" Where Applicable: Instead of just implying no impact, explicitly state "NO" for each impact category where there's no change. For example: "Impact on user: NO". This improves readability and clarity.
• More Detailed Testing Logs (If Feasible): While the link to the successful run is good, including snippets of relevant log output directly in the PR would further strengthen it. Show a "before" (failing builds) and "after" (successful builds) example, even if brief.
• Mention Specific CI Config Files Changed: If possible, mention the specific CI configuration files that were modified to disable the tests. This adds transparency.
• Consider Upstream Solutions (Long-Term): While disabling the tests is a practical short-term fix, mention (if applicable) whether upstream fixes for these macOS issues are being explored. This shows that the disabled tests are not intended to be a permanent solution.

Despite the minor suggestions, the PR provides sufficient information to understand the change, its impact, and the verification process. It generally meets the requirements.

[xiaoxiang781216]

should we fix this problem instead disabling it?

[lupyuen]

Hmmm can anyone help? Our macOS builds have been failing in our NuttX Mirror for a couple of weeks, kinda distressing to see it failing the NuttX Mirror Build every day 😬

[xiaoxiang781216]

could you attach the warning message?

[lupyuen]

I have created the issue:

• [BUG] macOS CI: sim:rpproxy_virtio fails at uintmax_t: 'unsigned long long' vs 'unsigned long' #14775

[xiaoxiang781216]

but pipe2 should work in Darwin

[lupyuen]

Something special about the Lua build I think, since the other calls to pipe2 seem OK?

[lupyuen]

I have created the issue:

• [BUG] macOS CI: sim:lua doesn't link with pipe2() #14773

[xiaoxiang781216]

https://github.com/apache/nuttx/tree/master/libs/libxx/libcxxabi could provide cxxabi.h, can we switch to it?

[lupyuen]

Sorry not quite sure how to switch it?

[lupyuen]

I'll post the above as NuttX Issues, since I'm not able to resolve them. Thanks!

[lupyuen]

I have created the issue

• [BUG] macOS CI: sim:libcxxtest can't find cxxabi.h #14774