MAINT run latest CI on Python 3.8 #15637
Conversation
|
Yes, that should do it, thanks! There are a few test failures due to warnings, some in neighbours and some in matplotlib. |
|
gimme a sec :) |
|
Seems like a bug in numpy? import numpy as np
np.ones(3).astype(np.float, copy=np.zeros(3).max() > 0)gives a warning. |
|
numpy/numpy#14397 https://bugs.python.org/issue37980 Should we work around it? |
| seed = 45 | ||
| # going to Python3.8 this fails even worse! Changing seed as temporary | ||
| # work-around | ||
| seed = 41 |
There was a problem hiding this comment.
Yes, this test seems unstable. Earlier today I got a failure in a clean environment with py37 for it as well.
There was a problem hiding this comment.
maybe it's the specific scipy release? I used 1.3.2
There was a problem hiding this comment.
should we xfail that test?
I'm gonna be mostly offline after this message for about a week so feel free to hack around ;)
There was a problem hiding this comment.
Actually this has been reported in #15420
|
Leaning towards xfail if someone wants to do that. |
|
On the call, I was speaking of MacPython/scikit-learn-wheels#35 which is having issues with |
|
This is ready to merge. For now, the warnings from matplotlib are explicitly ignored in the plotting tests. |
|
I am working on the windows |
Fixes #15618