paleolimbot commented on PR #692: URL: https://github.com/apache/arrow-nanoarrow/pull/692#issuecomment-2542831632
> This is nice. When running with numeric vectors it is even closer to zero-copy as it looks like as we retain the buffer: Yes, it's so close! If R did row-major matrices it would be. I need one more iteration here to get this down to a single `matrix()` call as the source of the copy but it's relatively close! > Do you foresee enhancing the C(++) side of things as well? Getting the nested offsets is still pretty tricky. I'd like to provide something like: ```c int ArrowArrayViewGetListElement(ArrowArrayView* view, int64_t offset, ArrowArrayView** view_out, int64_t* offset_out, int64_t* length_out); ``` ...which I think will be rather helpful for people to ingest the wild combination of Arrow storage that could possible give you a list as element of an array (e.g., listview, dictionary of list, run end encoded of list, etc.). -- 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]
