Christopher J. 'CJ' Wright

Is the returned rdd the actual data or just a reference to the data? If it is a reference can we pickle it and push it over the wire?

Can you elaborate on what you think would need to be written?

As I was reading through the code there seem to be two separate dask schedulers when accessing data on a remote server. There is a client cluster that sends a...

Potentially related, there is a [conda-forge recipe/build for this](https://github.com/conda-forge/pyxdameraulevenshtein-feedstock) so you should be able to `conda install` this across platforms.

Feel free to PR/issue against it. I went from the PyPi release, which is missing the tests and licence.

What is the plan to bring failing feedstocks into compliance?

I think it would be best in a central location so that we can change providers across feedstocks. It could still be in smithy if we feel ok cutting smithy...

Right, the feedstock says `default` and then we translate that into an actual provider, whatever is most convenient at the time (as declared by this config thing).

Attn: @jakirkham @isuruf

Although I can't speak for @dougalsutherland the benefits I see are related to automatic re-rendering and migration. For instance with the CircleCI 1->2 migration this kind of data would make...