Lawrence Mitchell

This is just for auto-completion, right? `pd.RangeIndex(2).str` is a property that exists and raises a "custom" `AttributeError` (since `RangeIndex` does inherit from `Index`). This happens because although we devolve to...

wrt the mask argument to the column factories, what is the use case in the python API for passing in a mask buffer? This speaks as well to @vyasr's question...

> Theoretically we could pass this to Dask, but I'm generally -1 on that because we would need to create a new Dask config for every possible UCX configuration. I...

Generic UCX environment setting via dask config is dask/distributed#7164

/ok to test

Merged trunk to pick up #15301 so the tests should run.

> Hello @wence- , now that #15028 is merged, would you please let me know if cuDF-python is still reading parquet row group metadata using pyarrow? Or is that step...

> As far as I can tell this is what they do in tsfc ([firedrakeproject/tsfc#158](https://github.com/firedrakeproject/tsfc/pull/158)) TSFC implements `CellAvg(expr)*dx` (and `FacetAvg`) as `cellwise_integral(expr/CellVolume *dx) * dx` as @michalhabera is advocating for

This is a bug in cudf's treatment of merge when both `left_on` and `right_on` are provided: ```python import cudf df = cudf.DataFrame({'a': [1, 2, 1, 2], 'b': [2, 3, 3,...

I think anything we do here will need to be in tandem with proposed serialisation changes in dask/distributed that are being contemplated. So I'll revisit this then.