-
Notifications
You must be signed in to change notification settings - Fork 66
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
Update for changes to Arrow linkage in libcudf #2357
Update for changes to Arrow linkage in libcudf #2357
Conversation
Build |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
👍 LGTM!
Thanks for working on this, @vyasr. 👏
I've run a smattering of integration tests in spark-rapids
based on this change, including the parquet_pyarrow
tests. Things appear to be in order.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This PR will need to explicitly update the thirdparty/cudf submodule to a version that contains rapidsai/cudf#16640, e.g.: latest branch-24.10, so it can test with those changes. We periodically update the submodule to follow cudf changes, but those updates will only automatically occur if it builds and tests pass. Since it won't build as-is without these changes, we'll need this PR to pull in the cudf changes.
Build |
Thanks for confirming Jason. I've updated the branch and attempted to trigger a build although I suspect I won't have the power to do so. |
build |
With rapidsai/cudf#16640, the libcudf library no longer links to arrow. While that dependency has been removed, the Java native code has not been updated to avoid arrow and still needs it for interop with arrow buffers (cudf now uses nanoarrow). While that may be removed in the future, for now the spark-rapids-jni needs to have its build updated to support the new location of arrow libraries and headers since they are now within the libcudfjni build instead of the libcudf build.
This PR depends on rapidsai/cudf#16640, but will be required for spark-rapids-jni to build once that cudf PR is merged.