-
Notifications
You must be signed in to change notification settings - Fork 7
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Added CI jobs that use conda compilers #52
Conversation
Codecov ReportBase: 46.04% // Head: 76.33% // Increases project coverage by
Additional details and impacted files@@ Coverage Diff @@
## main #52 +/- ##
===========================================
+ Coverage 46.04% 76.33% +30.29%
===========================================
Files 18 17 -1
Lines 2235 1348 -887
===========================================
Hits 1029 1029
+ Misses 1206 319 -887
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. ☔ View full report at Codecov. |
Interestingly the compilation fails with the conda clang compiler:
which I don't really understand as I thought stdio.h was pretty standard and is being included by Python header files. But this does seem similar to the error in e.g. #23 (although there it was stdarg.h). Let's see what happens with the conda gcc build. |
The conda-forge gcc works fine, so it seems it's clang from the default anaconda channel which has issues. I've found other issues complaining of broken conda compilers (e.g. davisking/dlib#1525) so maybe this is beyond our control and not due to anything specifically in casa-formats-io. With the v0.2 wheels, we should see fewer installation issues since fewer users will be installing from source, but users running into installation problems with the source build and using conda might need to switch what compiler they are using and report the issues to continuum? I'll see if I can report the current clang failure here to the continuum folk. |
I've emailed the anaconda maintainers with a link to this PR. |
There is nothing wrong with In order to fix this, you can explicitly point to the Of course, you'd have to point it to wherever your CI system installs the SDK for OSX. I verified that adding this line makes your error go away. |
9e945ab
to
0a261aa
Compare
@sumit0190 - thank you for the detailed explanation! I'll try and wrap up the PR along the lines you suggest. |
95cb86c
to
d61d69c
Compare
d61d69c
to
8a5cd27
Compare
Well this hasn't revealed any issues with our extension though no harm in merging it to safeguard against future issues |
This re-uses an approach we developed in extension-helpers - I wonder whether issues such as #48 are due to our C code not compiling properly with conda compilers - and in any case it's not a bad idea to have these jobs.
I do want to try and switch over the rest of the tests to the OA workflow but that's not urgent.