Follow-up cleaning & fixes for compatibility with NumPy 1 & 2 #7326
Conversation
These aren't actually necessary, as problems with precision were caused by another test `test_moments_weighted_central` that was changing the float representation globally with `np.set_printoptions`.
In `test_overrange_tolerance_float`: The previous test actually checked what happens if a Python float that is out of range for the image type is used as the tolerance. Not casting with `max_value.item()` would result in `np.float32(inf)` which is a different test. In `flood`: Furthermore, we need to turn `min_value` and `max_value` into Python scalars with `.item()` as well. Otherwise `max` and `min` seem to trigger an overflow error during the comparison.
|
@stefanv, let me know if I shall revert the deprecation of |
|
Is there a reason to deprecate |
In #7288 (comment) it was mentioned that there might be some issues. But I agree that's probably not a reason enough to change the status quo. I'll revert the deprecation an we can deal with that down the road. 👍 |
add a basic test for it, and make sure its docstring appears in our HTMl documentation (wasn't the case previously).
| We provide a ``lookfor`` function to search API functions:: | ||
|
|
||
| import skimage as ski | ||
| ski.util.lookfor('eigenvalue') |
There was a problem hiding this comment.
Note the change from ski.lookfor to ski.util.lookfor. It's still available as the former, but not advertised. Now, ski.util.lookfor is also included in the HTML API doc.
lagru
force-pushed
the
maintenance/follow-up-numpy2
branch
from
31bdad2
to
ecfb0a0
Compare
Description
A follow-up to #7288 addressing a few review comments that were left pending.
Checklist
./doc/examplesfor new featuresRelease note
Summarize the introduced changes in the code block below in one or a few sentences. The
summary will be included in the next release notes automatically: