pchintar commented on PR #9778: URL: https://github.com/apache/arrow-rs/pull/9778#issuecomment-4316938535
Yes @alamb , that’s a fair point. The only new copy introduced by this fix should be limited to dense union offsets, and only when the Vec-backed body buffer is not aligned for `i32`. So the tradeoff is: - main: aligned body allocation upfront, no union-offset repair copy - this PR: faster Vec read path generally, but possible small copy only for dense union offsets This is not covered by the current `ipc_reader` benchmark, since that benchmark does not exercise union-heavy data. -- 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]
