deepcopy error when running workflow in python3 about ephypype

[davidmeunier79]

I do not know if it is related to python 3.6 instead of 3.5 that we have tested with Annalisa, but I would not think so.

However, we have had issues with latest version of nipype 0.14, you should try to force the install of the version 0.13 through pip install, or my version of nipype https://github.com/davidmeunier79/nipype

[dmalt]

I've tested both python3.5 and python3.6 and nipype0.12, nipype0.13 and nipype0.14. The issue
remains.
I've found a similar problem in one of the nipype's issues. The guy reporting the problem says that it probably was about a conflict between python2 and python3 in anaconda env. It doesn't seem to be the case though since I've tried several anaconda virtualenv configurations --- still no success. I've posted a question to the related issue thread in nipype so may be they could help.
There's also another thread that should be related to ours. It suggests that it's not a packages configuration issue.
It seems to me that the problem is either in a way that I assemble the pipeline or in the data I use it on. It works on python2.7 though.

[dmalt]

@davidmeunier79: Maybe I can send a code snippet that runs the pipeline together with the chunk of data, so we can see if the problem is in my setup or it's something else?

[dmalt]

Oh, I just figured it out.
In my pipeline I used dictionary keys as an iterable to the pipeline (it's part of cli code to handle paths to input files in parallel). Apparently in python3 you have to wrap dict.keys() in list() --- otherwise the deepcopy issue appears.

[ostanley]

@dmalt Can you elaborate on your solution for this a bit? I think I have the same issue. I am initializing nodes in a for loop and storing them as keys in a dictionary, and getting this error as well.