Hello All, I'd like to discuss the potential addition of an asynchronous-oriented version of the C Data Stream interface.
This idea was originally brought up working with ADBC [1] as the current ADBC interface is inherently synchronous and requires any attempts to be asynchronous be created/managed/implemented by consumers rather than providing an asynchronous API itself. As any such API would require needing to handle interactions with the Arrow C Data Interface, it is better to upstream such concepts to the Arrow format than create something specific to ADBC. As such, I've created a PR [2] based on the discussions in the ADBC issue, as a jumping off point for more discussion. I didn't go as far as creating tests, examples, etc. for it yet as the core design of it might change based on feedback in this thread and I figured I would wait until I get more consensus from the community before I invest more time into it. Anyways, I'm looking for any and all feedback on the design, structure, concept, etc. in the hopes of getting this officially added to the Arrow spec. Thanks everyone! --Matt [1]: https://github.com/apache/arrow-adbc/issues/811 [2]: https://github.com/apache/arrow/pull/43632
