Apply groupby patch based on rapidsai/cudf#17985 to branch-24.12 #2920
Conversation
pmattione-nvidia
force-pushed
the
patch_2412
branch
2 times, most recently
from
4428aef to
fe8b95e
Compare
Signed-off-by: Paul Mattione <[email protected]>
pmattione-nvidia
force-pushed
the
patch_2412
branch
from
fe8b95e to
462144a
Compare
|
Since we are applying to 24.12, I'm not sure how to merge this. Maybe it needs ops team to work on it. |
The plan at the moment is to not merge it. Perhaps that will change, but we won't do that for now. |
|
We are going to try for a 24.12.2 patch release with this PR. Please comment as needed, or approve if it looks ok. . |
|
build |
gerashegalov
changed the title
There was a problem hiding this comment.
I missed in the first review that this file should have a name starting with 1_ in order to make sure that this patch is applied after the 0_ patch.
However, the patches have no conflict. The order is still established given 'f' > '0'. But we need to be careful if more patches are required.
This PR applies a patch to 24.12 that contains the core fix to the groupby algorithm that was merged into cuDF 25.02.