Julia Signell
Julia Signell
I think there are a couple different things going on here. In the first example, Dask is trying to be a bit too clever. It sees that the name that...
For the first issue. We recently added a warning, but maybe that was the wrong approach. Maybe we should stop bailing early in that case or maybe we should raise...
> But also, it must still be doing _something_, right? Or the metadata error wouldn't happen at all. That's a good point, but it's not. You get the same metadata...
Hmm. There must be something wrong with the meta in shuffle for the case where a dd.Index object is passed. ```python import pandas as pd import dask.dataframe as dd df...
Thanks for opening this @TomNicholas! I have a few thoughts: 1. The mypy failure definitely feels like a hypothesis issue. It seems like it would not be too hard to...
> > I am not convinced that this belongs in the `dask/array` part of the codebase. This is something the people would only use in tests right so probably it...
This is another example of per-partition metadata (like length) that if-tracked could provide useful fast paths. This idea has been discussed before and kind of fizzled out in high-level expressions...
Hmm this is an interesting one. Thanks for writing it up. It looks like the issue is in the aggregate method. I can't quite figure out what's going wrong though....
Thanks for keeping on top of this Charles!
Yeah we definitely don't want to bump the minimum. If it's easy to bring in some compatibility code that seems like a good approach.