Bradley Dice

I worked a bit with @tommy-waltmann on this problem. From my understanding, decorators and classes don't play nicely together when it comes to inheritance. This StackOverflow post seems related: https://stackoverflow.com/questions/2542747/class-decorators-inheritance-super-and-maximum-recursion

I wanted to post a couple nice links about Dremel encoding, since this is a core part of this PR. Summary of the encoding with nice diagrams: https://blog.twitter.com/engineering/en_us/a/2013/dremel-made-simple-with-parquet Original paper:...

@shwina That's very helpful, I did not consider `explode()`/`agg(list)`. To simplify and match this example: ```python def list_add(s1, s2): return (s1.explode() - s2.explode()).groupby(level=0).agg(list) ``` I'm guessing that `explode()` returns a...

@shwina Interesting. Would you consider exposing both `list.values` and `list.leaves`? It seems important to have a way to un-nest one level at a time (like with `list.offsets`).

I would expect eval’s behavior with + to match the + operator’s behavior, but we stated in a previous conversation (last week’s standup, I think?) that we explicitly do not...

> Right, which is why I'm suggesting a distinct `DataFrame.list.eval` API (note the namespace). I missed that namespace, thanks for the pointer. I have a lot of questions about how...

@shaswat-indian Thanks for working on this! We're entering code freeze for the 22.08 release tomorrow, so I bumped this PR to target `branch-22.10`. 👍

I think this is ready to merge! Thanks @shaswat-indian!

@gpucibot merge

Yes, we should do this. It was on my list of things to investigate after I upgrade RAPIDS to Thrust 1.17. (Also I think it’s an outer namespace like `CUSTOM::thrust::reduce`.)