fix: proposal for #2947

[maxgalli]

Test against np.integer as well in ArrayModuleNumpyLike.shape_item_as_index such that it does not break if ak.from_buffers is built with a numpy integer instead of an integer.
I have the feeling that this might be an "easy fix" and that maybe the fix should be implemented at a deeper level - please let me know if this is the case.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 82.21%. Comparing base (b749e49) to head (65fd4eb).
Report is 129 commits behind head on main.

Additional details and impacted files

Files	Coverage Δ
src/awkward/operations/ak_from_buffers.py	94.19% <100.00%> (+0.07%)	⬆️

... and 100 files with indirect coverage changes

[agoose77]

@maxgalli thanks for this PR! I think you're right that we'd prefer to tackle this slightly differently; presently, our internal representation of shape items is int or our unknown length types. So, we'd prefer to cast the NumPy types in the entry-points to our internals, e.g. ak.from_buffers.

[maxgalli]

@maxgalli thanks for this PR! I think you're right that we'd prefer to tackle this slightly differently; presently, our internal representation of shape items is int or our unknown length types. So, we'd prefer to cast the NumPy types in the entry-points to our internals, e.g. ak.from_buffers.

thanks @agoose77, I changed the PR accordingly to perform the casr in ak.from_buffers. Are there other functions in which this might be needed?

[agoose77]

At this stage, any function that accepts a length e.g. ak.unflatten, ak.from_buffers, and maybe some of the functions like ak.pad_none etc. I think it's reasonable to wait for bug reports there, rather than guessing :)

[agoose77]

This LGTM, and I feel confident in merging despite being out of the regular development cycle. @jpivarski I'm notifying you that I'm merging despite feeling comfortable to do so!

[agoose77]

@maxgalli thanks for this useful PR! <3

[jpivarski]

@all-contributors please add @maxgalli for code

[allcontributors]

@jpivarski

I've put up a pull request to add @maxgalli! 🎉