Matt Topol

@jorisvandenbossche I may have missed it in an earlier comment, but what is the current status of having a `__arrow_c_device_array__` and `__arrow_c_device_stream__` in pyarrow? @vyasr I'm currently traveling for a...

I'm back from my conference travels, so I'm slowly working through my backlog and should get to this within the next day or so to review and respond to the...

> > I may have missed it in an earlier comment, but what is the current status of having a __arrow_c_device_array__ and __arrow_c_device_stream__ in pyarrow? > That discussion lives in...

@vyasr Has work started to use the new types to expose the `to_arrow_device` / `from_arrow_device` methods in pycudf yet?

There might be benefit for a wider scope and potential if you target [ADBC](https://pkg.go.dev/github.com/apache/arrow-adbc/go/adbc) and then just use the FlightSQL driver for it. That way it can be easily adapted...

The open source Dremio docker image is likely the easiest integration test I can think of. Have a look at https://github.com/apache/arrow-adbc/blob/715fe2459a836527602383c50ecd40b8d520bb41/.github/workflows/integration.yml#L106 which is the CI workflow used for the flightsql...

Thanks for filing this, looks like you're correct. It's coming from `NewStructArrayWithFields`, which and the `ToRequestedSchema` looks like it doesn't properly carry over the null bitmap in that scenario either...

Yes it's a little circular because the builder tracks what updates are performed to the metadata. This is required to be able to specify the exact updates that are made...

I'd have to think about it a bit as far as recommending a path. But I would definitely see this as within the scope of this implementation if we can...

Hey @alexandre-normand sorry for missing this! My next personal priority in this library is going to be adding the variant data type, but if you want to contribute I'll happily...