Kev Wang

> @kevinzwang It is true that this needs to be implemented in the new UDF later. However, the current new UDF seems to block many behaviors of underlying resources. I...

Thanks for raising this up! This has been something we have been thinking about adding as well.

> @kevinzwang OK. I'm quite curious why we need to reset the number of detectable GPUs here? It seems that if this is the expectation, the current implementation can't achieve...

> If it is intended that when it equals None, it should be 0, then the environment variable CUDA_VISIBLE_DEVICES would be modified, but this is not currently observed. In that...

> However, I have another question: are the current thread's udf and the process's udf executed by whether the concurrency parameter is set? Why is there a distinction between the...

@Jay-ju to answer your previous question, you are correct in that the ability to configure if your UDF runs on a thread vs process is largely for performance reasons. In...

@srilman I'm honestly fine with only allowing UDFs to return serializable values. With our v0.6 version bump this might be a good time to start enforcing this, instead of building...

Hi @kszlim, thanks for submitting this issue. We are actually looking into streaming data sources/incremental ingestion, but I am not sure what the timeline is. Pinging @desmondcheongzx here if you...

It looks like we currently ignore the given `io_config` if you pass in a unity catalog table, and [we do not construct an S3Config with a region](https://github.com/Eventual-Inc/Daft/blob/main/daft/unity_catalog/unity_catalog.py#L102-L106) in our `UnityCatalog.load_table`....

@lukaskratoch what version of deltalake are you using? As for a solution to the io config thing without having to modify the library, you could perhaps extract the table_uri and...