DEV: assorted Pixi workspace improvements - #175
Conversation
|
GPU tests failed. |
It looks like SciPy got brought in as a dependency for the CuPy tests despite only being an optional dependency of CuPy. I'm not sure why. Update: Oh, it's because |
cool, let's return to this PR after that and scipy/xsref#15 |
|
Oh sorry, I guess this isn't just going to work out of the box, because the |
|
could you check that the CI logs look to be picking up the xsref tables correctly @steppi ? |
Everything looks fine with the tables. The Windows failure is unrelated and is something that we had run into before and fixed. I'm not sure what retriggered it. |
| } | ||
|
|
||
| inline std::complex<float> cyl_bessel_je(float v, std::complex<float> x) { | ||
| return static_cast<std::complex<float>>(cyl_bessel_je(static_cast<double>(v), static_cast<std::complex<double>>(x)) |
There was a problem hiding this comment.
It's kind of annoying that clang-format changes its mind on how to format things almost every time we update clang. Can we keep it pinned to the previous version in this PR, and put clang updates in separate PRs, doing the .git-blame-ignore-revs thing for the style change commits?
There was a problem hiding this comment.
this can be saved for a future PR which runs pixi update.
0e4608c to
9830a0b
Compare
There was a problem hiding this comment.
this has been generalised from requiring the xsref source to be cloned into the root directory, to having xsref installed in a Python virtual environment. Requires scipy/xsref#17 to merge first.
|
|
||
| [environments.lint] | ||
| features = ["clang-format"] | ||
| no-default-feature = true |
There was a problem hiding this comment.
xsref is added to the default feature, but we don't need xsref (or python) for the lint env
|
this PR is now ready from my side! Reminder to review commits individually. |
lucascolley
added
the
DX
steppi
left a comment
steppi
left a comment
There was a problem hiding this comment.
Looks great! Thanks @lucascolley, this is a major improvement
3 participants
@fbourgey another one! I think after this PR I can rebase gh-41. Probably easiest to review commits individually :)
depends on scipy/xsref#17