search

uwhackweek / jupyterhub-deploy

template repository for creating a jupyterhub
MIT License
0 stars 1 forks source link

Potential R support #9

Open lsetiawan opened 3 years ago

lsetiawan commented 3 years ago

@emiliom mentioned today during the eScience sprint that oceanhackweek previously had support for also R users. I think it would be good for this deployment to have that option also to make sure that hackweeks are supported with this capability if they request to do so.

emiliom commented 3 years ago

The support for RStudio in the OceanHackWeek (OHW) Hub might not be found within the OHW Terraform stuff or the hub conda env alone. Other people figured it out literally the night before OHW started, and then it worked perfectly. If there is interest I can dig out that information. We also had an R conda kernel available, and had some R tutorials that used Jupyter and others that used RStudio. https://oceanhackweek.github.io/ohw-resources/schedule/

lsetiawan commented 3 years ago

Thank you @emiliom I think it would be good to have those info here. What do you think @aaarendt? I can see some hackweeks potentially needing that RStudio capability. @scottyhq what is your familiarity with this setup?

scottyhq commented 3 years ago

Just wanted to link to https://github.com/pangeo-data/pangeo-docker-images/issues/163 for related discussion and examples. I suspect this could actually be dealt with in the Docker image rather than hub and terraform config. The ESS data science course happening this quarter expressed interest in a Julia image. Our hackweeks have been Python-heavy in the past, but there is no reason R or Julia couldn't be used, but i'm tempted to postpone working on it until there is a definite need.

lsetiawan commented 3 years ago

Ah okay. Makes sense. Thanks @scottyhq for linking that resource. From my past experience, I think you're right, it's simply a conda environment thing I think, one simply need r-base and jupyter-rsession-proxy.

emiliom commented 3 years ago

Having a conda R kernel for Jupyter is totally straightforward. The trickier part was having RStudio play nice with JupyterHub.

As we get going with OceanHackWeek '21 (August), we'll be revisiting what we did with Python + R to do it again this year. You know where to find me.