Mehmet Ozan Kabak

> It appears there is some work afoot to support this in [synnada-ai#1](https://github.com/synnada-ai/arrow-datafusion/pull/1) Yes, we plan to submit a PR draft early next week that makes progress on this issue....

@alamb, mentioning you since you showed interest in this. Once this gets some feedback, we will perform any fixes if necessary and then add `GROUPS` functionality so that this work...

Seems like we forgot the license header in one file 🙂 CI checks are passing except that. We will fix the license thing along with any other changes based on...

Thanks for the reviews, everybody! We are working on changes that address your reviews and plan to send them your way in a day or two.

Hello again! We resolved almost all the points you've mentioned. Additionally, we identified some challenging corner cases regarding NULL treatment and updated our code to handle these cases, along with...

Thanks @alamb! In our first follow-up PR, we will tweak the test you noted to increase coverage and also enrich the comments to explain subtle points such as the one...

In your example, the algorithm I'm proposing would do this: 1. Build (a1.mpp, a2.mpp, ...) in the order given by the compiler, parallelize only to the extent allowed by that...

> It is not yet possible to support separate parallelization of .mpp and .cpp between multiple targets, they can only be compiled together in parallel. > And even if separate...

We are interested in this too.

We should definitely be careful when doing this, doing this gung-ho style has the potential to break many things. @mustafasrepo please watch any work on this carefully and discuss possible...