Fail to install on Windows about gcn4r HOT 11 OPEN

Hi Eric,
The software was designed with R v3.6, I am not sure if incompatibilities exist for Rv4 but am happy to patch issues that arise. Can you verify ref = "master"?

If there are issues with master, I can also make another branch.

from gcn4r.

Hi Eric, The software was designed with R v3.6, I am not sure if incompatibilities exist for Rv4 but am happy to patch issues that arise. Can you verify ref = "master"?

If there are issues with master, I can also make another branch.

bad luck for me, still not able to install.
I'd tried git clone, download zip, and not having good luck.
Will try other version, and see if I can solve this issue.

from gcn4r.

Updated
Here is the issue I encountered when installing on Ubuntu 1804 + Python 3.6.13

Downloading mock-3.0.5-py2.py3-none-any.whl (25 kB)
INFO: pip is looking at multiple versions of <Python from Requires-Python> to determine which version is compatible with other requirements. This could take a while.
INFO: pip is looking at multiple versions of captum to determine which version is compatible with other requirements. This could take a while.
INFO: pip is looking at multiple versions of gcn4r to determine which version is compatible with other requirements. This could take a while.
ERROR: Cannot install gcn4r and gcn4r==0.1 because these package versions have conflicting dependencies.

The conflict is caused by:
    gcn4r 0.1 depends on networkx==2.4
    cdlib 0.1.8 depends on networkx<2.4 and >=2.2

To fix this you could try to:
1. loosen the range of package versions you've specified
2. remove package versions to allow pip attempt to solve the dependency conflict

ERROR: ResolutionImpossible: for help visit https://pip.pypa.io/en/latest/user_guide/#fixing-conflicting-dependencies

from gcn4r.

Ok, I just updated the package dependencies. Does the install work now?

Also, you may need to reinstall torch geometric with the following installation instructions with the set version: https://pytorch-geometric.readthedocs.io/en/latest/notes/installation.html

from gcn4r.

Hi Joshua,
Here is a short stiory.
first, I setup a virtualenv with python 3.6, with cuda 10.2 etc, on ubuntu 1804.
then, pip3 install git+https://github.com/jlevy44/GCN4R

At this moment, err msg says, "no such module as torch", so I installed torch independedly as following,

  pip install torch==1.4.0 torchvision==0.5.0
  pip3 install git+https://github.com/jlevy44/GCN4R

Still wrong, Failed to build torch-cluster torch-scatter torch-sparse torch-spline-conv

and pip3 install torch-cluster==1.5.4 torch-scatter==2.0.4 torch-sparse==0.6.1 torch_spline_conv==1.2.0 doesn't work, even I removed the version number.

Hope the information might be useful for you.
Thanks,
Eric

from gcn4r.

Do you have access to anaconda? If so, I would recommend setting up that environment.

I've also updated the readme to contain new installation instructions, including running install_gcn4r_dependencies, which should set up those difficult to install packages with the appropriate wheels. Torch should automatically be installed with gcn4r. pip and pip3, in certain virtual environments can refer to different versions of python. You may want to run the following:

which pip
which pip3

To ensure they are pointing to the correct environment.

from gcn4r.

hi Levy,
Have installed on MRO 4.0.2 + Rstudio Server + Ubuntu 4.02. Quite complicated but doable.
Just FYI,
thanks.
Eric

from gcn4r.

Thanks Eric,
Sorry to hear you went through so much trouble. I developed this package using a Mac, and unfortunately did not test the installation on Windows. Installing packages on Windows using devtools can be a pain. I will have to give optimal software packaging another thought. Potentially I could work on a Docker container, though this requires users to have Docker installed.

I'll leave this issue open until I can fully troubleshoot a Windows install.

Cheers,
Joshua

from gcn4r.

You are more than welcome. It's interesting that github returns nothing back when I'm working with the package.

Basically the idea is to clone the package, and generate a compression file, then make it in R environment. The complexity comes from the missing libraries which we have to install manually.

I think, it's nothing to do with your package but my doubtful configuration. So thank you for your generous support.

Many thanks and happy new Chinese year of Tiger,
Eric

from gcn4r.

Typically the devtools install should be sufficient, though I will continue to work on this.

from gcn4r.

Happy new year :)

from gcn4r.