-
Notifications
You must be signed in to change notification settings - Fork 232
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Use Spark's HybridRowQueue to avoid MemoryConsumer API shim #2000
Conversation
Signed-off-by: Jason Lowe <jlowe@nvidia.com>
Full disclosure: I was not able to verify this fixed the failing |
build |
@firestarman were you able to validate? |
I'm going to merge this. The nightly build is still broken, and this is highly likely to fix it. I wasn't able to get the cudf_udf integration test to run properly, but I can guarantee that if the integration test still fails after this is merged that it will fail differently. I'll kick the snapshot build and integration test build early to see if there are any subsequent failures. |
Sorry guys, I missed this. I will pay more attention to review requests. Thanks all for this. |
#640 added a copy of Spark's
HybridRowQueue
and supporting classes, but this breaks on Spark 3.2.0 because theMemoryConsumer
API changed. The original reasons for adding a copy was listed as visibility problems, but this can be worked around by placing our code in a package with visibiility, and such a package already exists.