alamb commented on issue #7424: URL: https://github.com/apache/arrow-rs/issues/7424#issuecomment-2835994322
> Thanks [@alamb](https://github.com/alamb) for the great writeup! I’ve sketched out a rough implementation and had two related questions: > > * Reusing metadata: Do we want to support metadata reuse during the value creation phase within the same builder, or is the goal to allow sharing metadata across multiple Variant values after they are already created and in Arrow memory? (My current version only supports reuse during value creation within a single builder.) I think we should start with reuse within a single builder. > * Key ordering: Since the spec allows for sorted_strings in the metadata, should the VariantBuilder take an option to control key sorting?(This refers to sorting keys in the metadata, and is separate from object-level sorted dictionaries.) Yes, I think having an option on the builder makes the most sense. I haven't fully thought through the interplay between creating sorted metadata and trying to reuse the metadata -- it seems like once the metadata fields have been added, we can't then add new field names without distrurbing existing values (the metadata field indexes would have changed) I think the design principles of the arrow-rs crate are to provide high performance primitives and reasonable defaults, and allow it to the user to specify / control things at a lower level of performance -- 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
