Evgeni Burovski

`dev.py test ... -- extra_pytest_arguments` is indeed supported; IIUC spin supports this, too, and we need this. And UX wise, `$ python dev.py test path/to/file` was never supported, it is...

The current `dev.py test` behavior is to ignore `-t` if `-s` is given (cf a demo under the fold). While this is not reasonable, we've never had a complaint or...

FYI @czgdp1807 : `smoke-docs` and related refguide-check commands can be taken from numpy: https://github.com/numpy/numpy/blob/main/.spin/cmds.py#L291 IIRC there were a couple of small hiccups due to spin idiosyncrasies, so you can just...

One rather annoying example I tripped over in scipy is ``` >>> n = 5 >>> x = xp.arange(n, dtype=xp.float64) >>> xp.exp(1j * xp.pi * x / n) ... TypeError:...

> when to use xp and when to use np https://github.com/scipy/scipy/pull/22014#discussion_r1885087007 To expand that comment: - in tests, having an `xp` fixture make it loop over backends indeed. - in...

``` ____________________ TestFiltFilt.test_complex_coeff[torch] ____________________ scipy/signal/tests/test_signaltools.py:2787: in test_complex_coeff z *= xp.exp(2j * xp.pi * fcentre/fs) E TypeError: exp(): argument 'input' (position 1) must be Tensor, not complex fcentre = 50 fs...

There seems to be an ongoing discussion between @roryyorke and @DietBru , and it's unlikely the discussion resolves before branching 1.16. Am bumping the milestone. Would be great to arrive...

We bumped the milestone on this once, and the next release is approaching (https://discuss.scientific-python.org/t/proposed-release-schedule-for-scipy-1-17-0/2146/1). Skimming the history, it looks like both _owners_ are broadly in favor: https://github.com/scipy/scipy/pull/22014#discussion_r1887301973 and https://github.com/scipy/scipy/pull/22014#issuecomment-2543155213. I'm...

cross-ref https://github.com/data-apis/array-api/pull/900/ which aims to actually standardize it.

Looks like gh-900 covered this request, and we can close it as completed, I'd think.