nealrichardson commented on issue #45438: URL: https://github.com/apache/arrow/issues/45438#issuecomment-2675043068
I would try UDFs, which ultimately do [add functions to the registry](https://github.com/apache/arrow/blob/main/r/src/compute.cpp#L787), and they provide a nicer interface around all of the C++ machinery for registering functions. If they work, great. If not, or they're not performant enough, then you can go deeper. > UDFs are probably not what you want since that operates row by row so is not particularly efficient. I'm not sure about this: technically they operate on the same chunks of data as any scalar function does. There is something funky about the R main thread that happens, I think, because UDFs could allocate R memory in theory (though yours would not have to), which may have negative effects. But IMO it's worth trying. Registering functions in DataFusion is an interesting idea, but then there's no R DataFusion package. You may also want to ask this on the d...@arrow.apache.org mailing list. You may find more people who have done this kind of thing with Acero, since it was designed to be extensible in that way. -- 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: github-unsubscr...@arrow.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
