fix: Fixes uproot.dask bug with empty branches #700
Conversation
for more information, see https://pre-commit.ci
| common_keys, entry_start=first_basket_start, entry_stop=first_basket_stop | ||
| ) | ||
| meta = dask_awkward.core.typetracer_array(first_basket) | ||
| empty_arr = hasbranches[0].arrays(common_keys, entry_start=0, entry_stop=0) |
There was a problem hiding this comment.
This change had to be made so we support empty branches. Earlier we were checking the entry_stop for the first basket.
I wrongly assumed that in empty branches the first TBasket would still be present albeit empty.
There was a problem hiding this comment.
Yeah, that makes sense: it could have been implemented that way, but it wasn't.
There was a problem hiding this comment.
It looks to me like this solves the issue! The test exactly reproduces @masonproffitt's reproducer, and furthermore you verify that the Dask-computed output has the same form as eager reading. The code update appears to be fairly minimal, too.
You didn't explicitly request a review. Are you done with the PR? If so, you can squash-and-merge it yourself.
| common_keys, entry_start=first_basket_start, entry_stop=first_basket_stop | ||
| ) | ||
| meta = dask_awkward.core.typetracer_array(first_basket) | ||
| empty_arr = hasbranches[0].arrays(common_keys, entry_start=0, entry_stop=0) |
There was a problem hiding this comment.
Yeah, that makes sense: it could have been implemented that way, but it wasn't.
|
Yes, I'm done with the PR and was waiting for your review. Thanks for the review! |
Fixes #697