BUG: Allow unsigned integer indexing, fixes #1405 #1406
Conversation
Permit indexing with unsigned integers. This should fix pydata#1405.
|
Indeed, looks like this should do the trick, but we definitely need a regression test before merging this (also a bugfix note, see the checklist in your first comment). |
|
Perhaps I was too fast, I edited it directly in github. |
|
@gerritholl when files are edited on github it creates a branch on your fork (here gerritholl:patch-3) which you can clone locally and update with the usual workflow. |
There was a problem hiding this comment.
@gerritholl - thanks for the fix here. Do you think you'll come back around to finish this PR up? Otherwise, I'll mark as "Contributions Welcome".
| @@ -58,7 +58,7 @@ def canonicalize(indexer): | |||
| '1d arrays') | |||
| if indexer.dtype.kind == 'b': | |||
| indexer, = np.nonzero(indexer) | |||
| elif indexer.dtype.kind != 'i': | |||
| elif indexer.dtype.kind not in 'ui': | |||
There was a problem hiding this comment.
nit: this feels too cute. I suggest writing this as:
elif indexer.dtype.kind not in ['u', 'i']:
|
I may make time for it later |
|
I have an updated version of this PR up at #1544 |
|
I'm closing in favor of #1473 |
Permit indexing with unsigned integers. This should fix #1405.
git diff upstream/master | flake8 --diff