Brian McFee

Thanks for this! We've never had a security policy before because it didn't seem relevant - we're a purely python project, and all file or network IO is done through...

Ok, well let's leave this issue up as a placeholder in case anyone with security know-how wants to take it on. Alternatively, it might be useful to see what comparable...

Good catch - thanks for noting it. This came from #1766, and is an artifact of some of the array broadcast trickery we needed to pull in order to allow...

> At this point, I think we're nearing a 0.11 release, which gives us the opportunity to make some API-breaking decision choices, or codify the implementation as is and bring...

Actually, looking a bit more carefully at it now - the types and annotations are correctly described as a union of float / ndarray. The only thing that changed, which...

> So an easy workaround is to just set the clipping _after_ creating the annotation with `ann.set_clip_path(rect)` or `ann.set_clip_box(tbbox)` (it doesn't appear necessary to set both.) Thanks for this -...

Thanks for this, but please keep the diff minimal. The proposed changes here are primarily irrelevant cosmetic differences, and in some cases remove information. This kind of thing makes PRs...

Thanks for the thorough report. I'm pretty sure whatever problem is triggering this is upstream of us - probably a numba/numpy 2.1 interaction. That said, I noticed that the python...

Ok - would it be possible to run the code snippet provided in the issue template under the environment where the tests actually run (python 3.10), instead of the outer...

Nudging - @mcepl are you still interested in helping to diagnose this problem?