R version pin issue

[GeorgianaElena]

Originally reported in: #107 (comment) by @yuvipanda

The version of RStudio we were deploying is too old for
R 4.1's graphics engine, causing this error:

R graphics engine version 14 is not supported by this version of RStudio. The Plots tab will be disabled until a newer version of RStudio is installed.`

So we pin R to 4.0 until we can resolve that. The underlying
issue is tracked in jupyterhub/jupyter-rsession-proxy#97

Solution

We tried to address this in the following PRs:

• Pin R to 4.0 #107
• Bump rsession version #109
• Fix RStudio install #108

Problem

I can still see the R, RStudio version incompatibility warning and the newer R version 👀

[yuvipanda]

It's back to R 4.1. I think this is why I removed r-recommended in the berkeley image - it might be coming from a different apt repository. Can you try interactively doing this without that package?

[yuvipanda]

I tried this #111, but did not work - with E: Unable to correct problems, you have held broken packages.

[yuvipanda]

It's failing with:

Step 21/47 : RUN apt-get update -qq --yes > /dev/null &&     apt-get install --yes -qq     r-base=${R_VERSION}     r-base-dev=${R_VERSION}     r-cran-littler=0.3.11-1.2004.0     nodejs     npm > /dev/null
 ---> Running in 18b2b7e45878
E: Unable to correct problems, you have held broken packages.

So I'm going to debug this by deleting all lines in the Dockerfile from that line, so the image itself builds up until that point. Then I can do a docker run into it, and then try things manually to iterate

[yuvipanda]

#112 is the result, the PR there has more description of the process I undertook

[choldgraf]

Is #112 deployed to the staging hub? I am still seeing R v4.1 on there.

[choldgraf]

this is now deployed in production as well (#113), so I'm going to close this :-)

thanks @GeorgianaElena and @yuvipanda for iterating so much to figure this one out!