search

MFlowCode / MFC

Exascale simulation of multiphase/physics fluid dynamics
https://mflowcode.github.io
MIT License
132 stars 56 forks source link

Revert MacOS runner - new one unstable? #443

Closed sbryngelson closed 3 weeks ago

sbryngelson commented 1 month ago

This is a mock PR to test CI. Related to https://github.com/MFlowCode/MFC/issues/444 and my humble attempt to work around it for now.

sbryngelson commented 1 month ago

I've tried the macos-12,13,14 runners and they all grab gcc13.3.0 and fail, which as far as I can tell is the only relevant change to these runners in the past few days. Strangely, gcc14 works fine with fftw on my machine, though I don't know if I ever checked gcc13.3.0 (perhaps only an older minor version).

henryleberre commented 1 month ago

@sbryngelson I also tried to fix this but couldn't a) reproduce it locally or b) understand why this error occurs. I have a suspicion it's related to it picking the wrong system headers but I am not quite sure. This might be related to Github updating their runners, but I'm not quite sure. It did start failing when brew updated GCC from v13.2 to v13.3 but that makes no difference on my mac. I could only find one question online about someone having this error (not while building fftw) and it wasn't very helpful.

sbryngelson commented 1 month ago

@sbryngelson I also tried to fix this but couldn't a) reproduce it locally or b) understand why this error occurs. I have a suspicion it's related to it picking the wrong system headers but I am not quite sure. This might be related to Github updating their runners, but I'm not quite sure. It did start failing when brew updated GCC from v13.2 to v13.3 but that makes no difference on my mac. I could only find one question online about someone having this error (not while building fftw) and it wasn't very helpful.

I've had the same experience you describe. Something else must have changed.

sbryngelson commented 3 weeks ago

Closing this PR in favor of https://github.com/MFlowCode/MFC/pull/469