Erik Welch

To determine the partitions, I would have few reservations about creating a 1-D array of dtype `object` of tuples and then using the existing `partition_quantiles` method. I realize this isn't...

Thanks for the reply. Then we should pivot this issue, because there is handling of categoricals in `cudf` that _does_ assume `drop_duplicates` preserves order: https://github.com/rapidsai/cudf/blob/6b20f2a6f338685ca65572b4ac5b15046fdf1f49/python/cudf/cudf/core/column/categorical.py#L1477-L1482 https://github.com/rapidsai/cudf/blob/6b20f2a6f338685ca65572b4ac5b15046fdf1f49/python/cudf/cudf/core/dataframe.py#L7259-L7264 Note that this makes...

Here's an example that fails for me: ```python import cudf import dask_cudf transactions = [ ["SRC", "DST"], [(89021, 11), (89216, 4), (78634, 16), (32431, 4), ] ] relationships = [...

Yes, as a "please defer to other categorical ordered property during `astype`, but this is weird so users probably shouldn't use `ordered=None` explicitly and we wanted to deprecate this" kind...

> It's the latter behavior (ordered property getting changed when it shouldn't) that is affecting my code. Let me highlight this from my examples above: ```python In [1]: import cudf...

Wow, nice! Yup, I expect to start work on this today or Monday.

We should hold off on MG until the two `cudf` issues above are fixed. I can push MG changes to a different branch while we wait. This PR also opens...

Using categoricals for the type changes the behavior of `renumber_edges_by_type` and `renumber_vertices_by_type`. Previously, we would sort using on the _name_ of the type. Now, we sort on the categorical value...

As @seunghwak, `nx.triangles` doesn't support directed graphs, but `nx.clustering` _does_ support directed (and weighted graphs), so triangle counting on directed graphs _is_ needed. For the clustering coefficient on directed graphs,...

Yeah, for SG. I'm going to put MG into a different WIP branch, because of issues I'm having with `dask_cudf`. I still want to add one thing: the ability to...