Andrew Lamb

I general, other than the fact that this proposal sounds like a lot of work, I think it sounds wonderful 🏆 I did have a question about the proposed trait...

> Sure, but it reduces dyn dispatch by a lot (once per batch instead once per group), removes the take kernel and the duplication can be hidden by careful macros/generics....

@iajoiner -- here is the main ticket that is tracking the row format progress. I think there are many PRs and other docs linked from here.

> It seems like a table function should produce RecordBatchs and effectively compile down to an ExecutionPlan. I agree it should definitely produce `RecordBatch`

> @alamb @thinkharderdev @doki23 i met the same problem in https://github.com/apache/arrow-datafusion/issues/2343 I left some thoughts in https://github.com/apache/arrow-datafusion/issues/2343#issuecomment-1110222756

I plan to give this a more careful review tomorrow

@gandronchik thank you for the explanation in this PR's description. It helps though I will admit I still don't fully understand what is going o. I agree with @doki23 --...

I think adding UDTFs (aka user defined table functions) that produce a 2 dimensional table output (aka `Vec` or a `SendableRecordBatchStream`) would be a valuable addition to DataFusion. I think...

marking as draft until we figure out what to do with this

Thanks @gandronchik -- I will try and find time to re-review this PR over the next few days in light of the information above.