search

jupyter / nbviewer

nbconvert as a web service: Render Jupyter Notebooks as static web pages
https://nbviewer.jupyter.org
Other
2.22k stars 551 forks source link

dockerhub images are not tagged #741

Open MarkusTeufelberger opened 6 years ago

MarkusTeufelberger commented 6 years ago

https://hub.docker.com/r/jupyter/nbviewer/tags/ only shows the "latest" image, it would be nice to have some indication about which nbviewer version this container actually contains. It seems like there are some automated builds on dockerhub, but it is not even clear to me which version of the code is used in there.

parente commented 6 years ago

A bit of background: We have Docker Hub configured to build images on every merge to master. We've only ever tagged 2 versioned releases in order to push a package to PyPI. We probably won't tag additional versions unless nbviewer undergoes some major changes.

I see two options for associating a meaningful tag with every image build on Docker Hub.

  1. We could switch to building on Docker Cloud and use the versioneer generated version based on the last tag, the number of commits since that tag, and the git sha, as a basis for the docker tag.
  2. We could switch to build on Docker Cloud and use the git sha alone as the docker tag like in jupyter/docker-stacks.

Both provide some link between what's on Docker Hub and what's in GitHub.

skritch commented 5 years ago

Hi - wanted to give a +1 for version-tagged Docker images. I'm setting up an internal notebook gallery with some extensions on top of NbViewer, and would like to be able to start from a version-pinned image off Dockerhub to ensure my extensions don't randomly break on a future update.