Docker configuration out of date?

[anselanza]

I am trying to get HyperGAN running via a Docker image (since a Dockerfile is provided), as I would much rather have a contained, pre-configured environment in which to use it (rather than fiddling with Python versions all over my machine).

Has anyone actually used the Docker setup successfully with this project? I notice that while the Dockerfile specifies a Python 3.4 environment, the syntax in the project itself only works using 3.5. This also means that the Tensorflow version needs to be -cp35 instead of -cp34.

The ADD command also makes no sense, referring to a directories that do not (or no longer?) exist: ADD build *.py shared /. I changed this to a simple COPY . . to just copy everything.

Finally, the ENTRYPOINT also makes no sense: ENTRYPOINT ["python3", "handler.py"] - what is handler.py? This file also does not appear to exist anywhere.

I changed the entry point to ENTRYPOINT ["python3", "hypergan", "test"] but now I get the following error (from the Docker container):

/usr/local/bin/python3: can't find '__main__' module in 'hypergan' 

Any idea where to go from here?

[mikkel]

Yes docker support is in a broken state at the moment. I was trying to get it running on SONM to get more cheap compute.

[anselanza]

Any idea where to go from here, then? The hard part just seems to be getting files in the correct place?

[martyn]

The Dockerfile will not work until it is updated for 1.0. Will be after release.