mbutrovich commented on PR #3930: URL: https://github.com/apache/datafusion-comet/pull/3930#issuecomment-4237253731
Before I dive too deep into the review, I think the PR description is a bit lacking in high level design and assumptions. For example, "Introduces a "batch stash" optimization that passes batches as opaque u64 handles through the JVM instead of doing full Arrow FFI export/import": so to steal some C++ lingo, we're basically stashing pointers into u64 and reinterpret casting as a pointer, and we can do this because we know more things about the life cycle of these batches than the Arrow FFI interface is designed for? The FFI interface provides safety guarantees and call back support, managing life cycle. We know things about the life cycle since we control the application on both sides of the FFI boundary, so we can bypass some of the protections that Arrow FFI otherwise affords? Is that the high level context? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
