westonpace commented on pull request #10376: URL: https://github.com/apache/arrow/pull/10376#issuecomment-875932074
Unfortunately, I think I agree with closing. The "common" path for IPC is hot-in-memory reads so the actual "read" calls should be more or less a no-op. S3/IPC is also probably a pretty uncommon path. For non-S3 the kernel should be doing its own coalescing. Maybe the reason there isn't much benefit is because IPC is always doing a full read of the file so there are no real gaps. Though if we want to start optimizing for partial reads I think there is more we'd have to do than just coalesce. So maybe a "close, but reopen and investigate further if there is a compelling use case". -- 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]
