[JOSS] Docker installation

[edeno]

It would be great to document what a user should do after running the docker image. For example, I tried:

root@05b17dd067c8:/# ipython
Python 3.9.14 (main, Oct  5 2022, 15:22:07) 
Type 'copyright', 'credits' or 'license' for more information
IPython 8.11.0 -- An enhanced Interactive Python. Type '?' for help.

In [1]: import deepof
---------------------------------------------------------------------------
ModuleNotFoundError                       Traceback (most recent call last)
Cell In[1], line 1
----> 1 import deepof

ModuleNotFoundError: No module named 'deepof'

xref: openjournals/joss-reviews#5394

[lucasmiranda42]

Thank you for trying all installation methods! Indeed, one must install DeepOF after loading the docker image (either using pip or from the repo). Would you recommend editing the documentation, or adding DeepOF to the Docker image directly?

[edeno]

I would add DeepOF to the docker image directly. Might as well cut down on an additional step for the user.

[lucasmiranda42]

Perfect. Building already. Should be up in an hour or so

[lucasmiranda42]

Dear @edeno,

I just pushed an updated Docker image with DeepOF loaded onto it, which should address the problem presented above. I'll close this issue for now, but feel free to reopen if needed, of course!

[edeno]

Okay I was able to import deepof successfully in this docker container. One question for you is would it also be beneficial to have deeplabcut in the container as well since the two libraries are dependent or is that out of scope in your mind?

[lucasmiranda42]

Glad it worked! Thank you for the suggestion. While I'm happy to discuss it further if you think it's necessary, I'm not a big fan of the idea, mainly because there are many different versions of DeepLabCut that are compatible with DeepOF, and users may prefer one over the other for some particular reason. Besides, DLC provides its own set of Docker images already, so setting up two independent environments would be the most versatile. Hope this helps!