CoW: Remove todos that aren't necessary #56169
Conversation
phofl
changed the title
| msg = "'float' object has no attribute 'startswith'" | ||
| with pytest.raises(AttributeError, match=msg): | ||
| data.apply(transform, axis=1) | ||
| with tm.assert_cow_warning(warn_copy_on_write): | ||
| data.apply(transform, axis=1) |
There was a problem hiding this comment.
I think the reason I thought this didn't need to warn is that row should not be a view, given the dataframe has mixed dtypes.
So after CoW is enabled, nothing will change here?
There was a problem hiding this comment.
Yeah sorry, you are correct
The implementation is dirty... I'll mess around with it later today
Ah, yes, that ugly bug .. (I was assuming we didn't track some ref correctly in case of Index because of it being immutable in the non-CoW mode)
Yes, but it's still a "false positive" warning for the user. Just one we decide we can't solve? (or is not worth our time to fix) |
yeah sorry should have been more precise. This is a known limitation. I have no hope in coming up with a fix before 2.2 is out if ever |
This reverts commit b09a160
xref #56019
This should warn, row is a Series and the setitem is inplace then
The Index case is a bug at the moment, creating the df is zero copy, updating the df afterwards will propagate to the index, but will be fixed with CoW
test_del_framethe parent is a single block, we are viewing the same block in df2 even though we deleted column "b"