Did this end up working with multithreading? The homebrew version is there because if you install clang via homebrew, it will come with OpenMP support. The system clang compiler does not have openmp support.

Did this end up working with multithreading?

Yes, it works for me!

Well if so, I think you should feel free to add a note that this is how you use the system's clang. However, the previous instructions show how to install clang via homebrew, so if the user followed the instructions above they end up not using the compiler that it used.

What do you think is the best way to describe this? I think if you were to just not set the CC and CXX variables, the path you set is what gets found.

Ah. And I think this relates to whether sudo was used when running homebrew, maybe. Did you use sudo? That would explain the difference if so.

@sohhae -- can you double check that you were able to run on multiple threads with apple clang, and not just that the code ran cleanly? You can double check that you are seeing something like:

                   | The OpenMC Monte Carlo Code
         Copyright | 2011-2021 MIT and OpenMC contributors
           License | https://docs.openmc.org/en/latest/license.html
           Version | 0.13.0-dev
          Git SHA1 | 333137dbae939cbfc06eef6237ae756a1d8083e7
         Date/Time | 2024-09-10 15:17:25
    OpenMP Threads | 10

at the top when OpenMC is run (note the OpenMP Threads line). Is it reporting more than 1? If the OpenMP Threads line is not showing up, then this means OpenMC was not compiled with threading support. If you are getting multiple threads, did you use any other homebrew install commands and/or sudo? The system default apple llvm/clang does not have OpenMP support, as @gridley said.

@gridley This is the clang installed by homebrew. I have confirmed this by uninstalling llvm, and verifying that the clang executable is no longer present, then reinstalling and confirming that the executable is present again. The current path specified in the documentation seems to be incorrect, export CXX=/opt/homebrew/opt/llvm/bin/clang++ does not exist on my system even after installing llvm with homebrew. It is possible this comes down to configuration differences, or homebrew/MacOS versions, so it might be worth some other people verifying, but the paths I've given worked on my M2 Mac Pro, running macOS Sonoma 14.6.1 with brew version 4.3.19.

@jtramm thanks for the explanation. Please see below.

                 | The OpenMC Monte Carlo Code
       Copyright | 2011-2024 MIT, UChicago Argonne LLC, and contributors
         License | https://docs.openmc.org/en/latest/license.html
         Version | 0.15.1-dev
        Git SHA1 | 57816e6b8cf23ed0e9b020b72752ed6aeb9501dd
       Date/Time | 2024-09-10 10:27:34
  OpenMP Threads | 12

 Reading settings XML file...
 Reading materials XML file...
 Reading geometry XML file...
 Reading tallies XML file...

Wow, I wonder what the difference is! Have you seen what happens if you just don't set the CC and CXX files?

Thanks for clarifying @sohhae! On my system, homebrew installs llvm to /opt/homebrew/opt/llvm/bin/clang++ similar to @gridley. One option for the docs might be to have the discussion tell the user to ensure they add clang++ to their path, and that probable locations if installed via homebrew are at:

/opt/homebrew/opt/llvm/bin/clang++

OR

/usr/local/opt/llvm/bin/clang++

From what I can tell in some of the secondary comments on this stack overflow post, the paths that homebrew makes binary simlinks is different depending on M1 vs. M2 or newer (for reasons unknown!). To future proof things, we might also advise the user to double check where homebrew is installing things via:

brew info llvm

Thank you for getting to the bottom of this John! Yes, we should definitely just note that it may be one or the other. Hopefully we can eventually write a more detailed install guide (not on this page) noting that homebrew isn't necessarily the only route as you've noted before John.

Thank you both @gridley and @jtramm. This was my first ever openmc pull request, so I'm still learning the ropes on how all this works. I will try to be more descriptive in the future! :)