Installation instructions incomplete

Dear all,
There are different ways of installing renku mentioned at https://renku-python.readthedocs.io/en/latest/installation.html. None of the descriptions include git-lfs or papermill. Git-lfs is needed in any case and papermill is needed for the tutorial, so it would be great if this could be added to the instructions.

Thank you for contacting us, that is a good point!

I’ve created an issue to update those docs with gitlfs info in https://github.com/SwissDataScienceCenter/renku-python/issues/1750

As for papermill, since the tutorial is for working on renkulab and our default python images already include papermill, I don’t think we should include it in the installation instructions for the commandline client, as it’s not needed by the client. Maybe we could update the tutorial in case someone uses a different image? @rrrrrok

1 Like

Wow, thank you, I was not aware of this requirements.txt template. How do I use it? If I create a new project on renkulab.io, I get an empty requirements.txt.

@schymans yes, that one is already a part of the base image that your project is initialized with. Whatever you put in requirements.txt is installed in addition to those packages.

I see, so it is not really visible when someone clones the project.
Actually, is there an easy way to reproduce the whole environment defined in a renkulab project on an external machine? For example, I clone the same project onto a lab machine to collect data, push to it, then I pull it again on my office machine, do analysis and push back, then I would like anyone creating an environment on renkulab or deciding to work with the repo locally to have everything needed available to them. Could such a workflow (incl. example docker commands) be included in the docu?

yes, definitely this is possible and not currently well documented. Is docker an option? If not, there are other possibilities with virtual environments, but would need something beyond documentation at this point.

Yes, docker would be fine, whatever is easiest. We don’t have docker installed on the lab computers and I will have ask IT support to do it, but it would be the most robust solution. At the moment, I use conda and pip.

Ok sounds good - I’ll put together some instructions around this and we can then figure out how to potentially automate some part of the process.

1 Like

Awesome, thank you! Could you tag me when they are ready to try out?