Irfan Alibay

@marinegor I've not had time to look at this PR properly yet, but are you vendoring a bunch of code from elsewhere? If so then yes tests would be needed...

I see, I wouldn't necessarily be advocating for maintaining a fork of some other repo - that invites a lot more maintenance burden. 1) Is there any way to convince...

Thanks @marinegor, pinging @MDAnalysis/coredevs again - please weigh in on this.

@orbeckst just to check - by suggesting that we vendor the code, are you saying that we should be making einops a core dependency of MDAnalysis? Having vendored code with...

> If we don't want to use einops then we could keep @marinegor 's workaround. I am not saying that we _must_ keep pydssp_numpy.py bit-identical to the original, just as...

> Fair point, although pure Python/noarch is probably still the one that is easiest to catch up. (Maybe make a PR for their CI to include 3.12.) This would be...

> I would probably suggest not having pydssp as a dependency in pyproject.toml, but leave the code as is -- i.e. if the user already has pydssp, it'll be used....

Probably? I think we're still missing a set of wheels that are fully compatible with the current status of MDA though, might need to get that done first.

I think we're just missing the CF migration, I'll add it to my to-do list for later in the month.

:/ looks like the job running against nightly wheels is broken - might need to fix both at the same time.