endrebak
Thanks for the investigation. I much appreciate it! I'm starting work soon but I'll look at this afterwards.
I was busy with a more critical bug. I’ll look at this today. Thanks for the investigation again. It was brave of you to look at init - it must...
pr.subtract - strandedness=None does not compare PyRanges on the same strand (even if both stranded)
Thanks for reporting. Can you come up with an example where the behavior is different? Based on the examples below it seems like they act the same way as subtract...
pr.subtract - strandedness=None does not compare PyRanges on the same strand (even if both stranded)
I might have written the wrong thing in the docstring though and copy-pasted it to all the functions. If the API had changed the usage for strandedness=None I'm sure someone...
I should not have written the issue like that, sorry! I have not actually looked at what breaks multiprocessing, so I should not have blamed ray. Also, I thought no...
I am wondering if just `gr.apply(lambda df: df.groupby...)` is good enough and that a custom function would only be confusing.
I'll think more about it, but this is functionality that is unlikely to see progress within the next few months :/
I guess the latter can be done if `from_dataframe` has a cache of column names it has seen. But that might cause quirky behavior in multicpu scenarios.
Do you have the pandas version? I wasnt able to reproduce yesterday :)
Thanks for the sleuthing. I am not able to reproduce. ``` In [1]: import pyranges as pr pr._ In [2]: pr.__version__ Out[2]: '0.0.87' In [3]: import pandas as pd In...