Closed ddeka2910 closed 3 years ago
@gerrith3
I'm not opposed to it, but why is it needed? Hovercraft is pure Python, using a different architecture should not make a difference?
Hi @regebro you are right that python should be completely architecture independent. However, we do wind up patching up products that have managed to use a library or function that is pulled/downloaded that is an intel executable, causing the whole application to fail. Also, there are certain versions of python that were never ported to ppc64le (i.e. predated the release of little endian power hardware). Regression testing ensures that you immediately catch any places where the code calls out to some architecture dependent component (hopefully rare in a case like this, although my team is finding some supposedly "pure python" that isn't) and also helps document and verify the versions of python that are available, as well as ensuring no marginal failures because of tests on platform architecture. Ultimately we leave it to your call, but it really simplifies "porting" if we know without a doubt that the package has been consciously regression tested on ppc64le.
OK, thanks for the explanation.
OK, implemented. It fails on 3.9-dev at the moment, though. I'll expect that will work once travis supports 3.9 properly.
@ddeka2910 can you check with Siddesh to see how travis includes python 3.9-dev for intel and see if we can do the same on power? This looks like a gap based on what I see in the .travis.yml.
Hi @regebro, the permission issue for python 3.9-dev has been resolved on travis. If you re-trigger the build it should now pass for python 3.9-dev.
Thank you for the code. Add support for architecture ppc64le.
This is part of the Ubuntu distribution for ppc64le. This helps us simplify testing later when distributions are re-building and re-releasing. For more info tag @gerrith3