Skip to content

Updated Dockerfile #209

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed

Conversation

benediktbrandt
Copy link

@benediktbrandt benediktbrandt commented May 10, 2018

NVIDIA switched their repos from http to https, see the following commit in the nvidia/cuda docker file

https://gitlab.com/nvidia/cuda/commit/f31ece4bc9195a446f1dd6b9c8479c71fb295dfa

As a result we now need to install apt-transport-https in order to be able to use the NVIDIA repos.

The present version of the Dockerfile has other issues too. There are several libraries that were recently updated and whose installation now fails. I am currently in the process of updating the respecive lines in the Dockerfile and will submit additions to this pull request once I have tested the changes.

@benediktbrandt
Copy link
Author

benediktbrandt commented May 14, 2018

To get the Dockerfile to successfully build the docker image additional changes were needed. Here are the details:

  • The conda version of spacy currently has issues with some pinned dependencies, that cause the installation to fail. Installation through pip works fine, so I substituted the conda install with a pip install.

  • Theano 1.0 has recently been released. However hep-ml 0.5.0 has requirement theano==0.8.2, which is why I am forcing the installation of theano 0.8.2.

  • Basemap installed from source has an issue with mpl_toolkits and causes the import of basemap to fail. Installing it through conda forge instead works fine (same version)

  • kmeans-smote is not compatible with the nightly built of scikit learn, so I chose to install scikit learn through pip

  • I updated to pytorch 0.4 since the latest version of pytorch-audio requires it

  • The latest version of pymc3 requires theano>=1.0. However hep-ml requires theano==0.8.2. The most reasonable thing to do seems to be to install pymc3.0 which is compatible with theano 0.8.2

  • We need to have the vega package installed first for jupyter nbextension install --py vega to succeed.

After making these changes I was able to build the docker image and to successfully run all the tests.

Please let me know if I should change something in this pull request, or if some of the conflicts should be resolved differently. I am grateful for all feedback.

@benediktbrandt benediktbrandt changed the title added apt-transport-https due to update of nvidia repos Updated Dockerfile May 14, 2018
@rosbo rosbo closed this Dec 21, 2018
@rosbo
Copy link
Contributor

rosbo commented Dec 21, 2018

Hi,

We added apt-transport-https a little while ago and vega3.

If you are still having issues with our image, let me know.

Thank you

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants