CurtHagenlocher commented on issue #811: URL: https://github.com/apache/arrow-adbc/issues/811#issuecomment-2123678458
I gather that `ArrowDeviceArray` has different allocation semantics than `ArrowArray`? I guess I need to take a closer look at it. Any contract around releasing the `ArrowDeviceArray`s and `ArrowSchema`? What if the consumer hasn't released them yet when the producer releases the `AsyncArrowStream`? > `rows_affected` should be populated if available before the call to `on_schema` I think this would be a bit of a pain to manage. Why not add `rows_affected` to the `on_schema` call? > IMO this gets annoying, as you have to handle every error in two locations What if the receiver e.g. can't allocate memory for an async task? I don't see how you get out of having to check the error in two locations in general, and it feels nicer for things like parameter validation failures to fail immediately. > How will this integrate with the rest of the Arrow ecosystem? Realistically, I think we'd need to push this API into Arrow itself :/. -- 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]
