Julia Signell

Hmmm. I am wondering if somehow the cumsum is being called on the individual partitions before they are coerced to pandas objects. If you wrap the output of the map_partitions...

Hmmm that's an interesting issue! Raising an error seems like the best bet. So then the question is where to raise the error from. I think `compute` should probably be...

@jrbourbeau pointed me to https://github.com/dask/dask/issues/7082 as a broader idea that would help with this specific issue. It is probably overkill if this is the main improvement, but keeping track of...

Feel free to just open a pull request @ppsbs as soon as you start work.

Once you are understand what is going on it might make sense, but it is not the behavior that a user would naively expect. This is not likely to be...

I think raising an exception or even just a warning is the right path.

I agree in principle with @jorloplaz that this is an odd behavior and that duplicate column names are never a good idea. However since they are still possible to create,...

I am going to start taking a look as these.

Thanks for writing this up @ncclementi! Please feel free to pick up either of those PRs, or we can work on them together if you like.

I totally agree that if there is a simple way to deduce shape that is desirable. Especially if it doesn't add too much complexity to the codebase.