Patrick Hoefler

Results 345 comments of Patrick Hoefler

Yes, If we get a test for it

> Are you familiar with https://docs.dask.org/en/stable/10-minutes-to-dask.html ? Yes, but that's was too much (and unnecessary) information for someone who wants to start playing around with. It's not really a quick...

This isn't ready yet and definitely needs more content, but serving 5 different use-cases at the same time is one of the issues with the other page. Someone who hasn't...

Hi, thanks for your report. This is definitely a bug, this should behave the same as on the previous version. I'll put up a solution

Thanks for the report. We rewrote indexing because the previous implementation was terrible in most cases. Is there a sparse equivalent to concat these arrays?

A sparse equivalent for np.concatenate that does not raise for sparse arrays with 0 elements

We could overload yes. I think (haven't looked deeply) that argsort could also be problematic. The main change is: We are now concatenating and then calling argsort on it

Happy to review a pr that addresses this, haven’t looked much into what needs to change though

I think it's in ``concatenate_arrays`` in ``array._shuffle.py``. Just special Casing would be fine for an initial version

I think the only direction we would consider here is to make Dask Array raise as well. We are moving away from implicit computes since they can be ridiculously expensive....