Run CI on a big endian system?

Development
1

So, is it worth a thought to eventually run CI/tests on a system with big endian? I have a feeling our code has never been tested on big endian, to be honest. At least reftek tests would have failed then (see comments in https://github.com/obspy/obspy/pull/2678).

Or is it just not worth it to put effort in it…? Has anybody ever worked on a big endian machine?

It seems that github does not offer any big endian runners for github actions, so it might be some work to find some big endian runner hosted elsewhere…

CC @d-chambers @dsentinel

2

I think Sun Sparcs might have been big-endian. However, I doubt there are many of those systems still running.

3

IMHO our time would be better spent elsewhere. I would hate to see another CI service added now that we have unified all of them on github actions.

4

Sure, definitely only worth a thought if it would be possible to grab an existing image in the github actions workflow.

5

Instead of running on a big endian system it might be more useful to try to get CI running on Power9 and ARM architectures. At least the latter will become relevant before too long because Apple will change its chips to it.

But I’d say this should be dealt with if it becomes a problem - otherwise its just extra work with little benefit.

6

I may be able to set up something in a Fedora copr; I haven’t enabled it, but it now has arm (32-bit and 64-bit) and s390x (a big endian machine). I am not sure how useful they are as they might be emulated.

It could be setup as CI, or we could just check it around release candidate time.