Conversation
Pull Request Test Coverage Report for Build 20348695864Details
💛 - Coveralls |
|
Ok. Found out the culprit: swig>=4.4 (numpy/numpy#30128). For now, I have asked clang to raise a warning, instead of an error, as seems to be doing gcc. |
I tested this loaclly on macOS arm64. the build succeeds, but the underlying numpy/SWIG warning (import_array() in an int-returning function) is still present and would still fail if treated as an error. This PR seems to mitigate CI behavior but i am not sure if this really fixes the root cause. perhpas check numpy init macro in the swig wrapper so the warning cannot occur on macOS? I am not sure if this makes any sense at all here is some output |
|
The root error has been fixed in numpy (in the PR i linked). At some point it will be available through pypi. In any case, this modification makes clang and gcc behaviour to agree (at least in this regard). |
Got it, thanks — in that case this looks fine to me and should resolve the CI issue you are a superhero rockstar!!! |
nikosarcevic
left a comment
There was a problem hiding this comment.
Looks good to me — thanks for the clarification. This should resolve the CI issue.
|
Late to the party, but just to say, amazing detective work @carlosggarcia |
I believe the problem is that this warning in ubuntu
becomes this error in macos
I think the problem is that we are using(Actually the numpy version is not the issue. See #1271)numpy<2for compatibility withdarkemulator.