Pretty well supported with github.com/jupyterhub/jupyter-rsession-proxy/, although you need to be running inside a container (or something with network namespace isolation) for it to work.

PAWS now supports rstudio: https://hub.paws.wmcloud.org/hub/user-redirect/rstudio

Code at https://github.com/toolforge/paws/blob/6719585734beab915db727338850597ba2842060/images/singleuser/Dockerfile#L84

This is awesome, thank you @yuvipanda! Couple of issues encountered so far.

There's an issue with the latest stable release of RStudio and R 4.1 (https://twitter.com/kevin_ushey/status/1395054942088400905)

PSA: If you're an RStudio user planning to upgrade to R 4.1.0, you may want to install the preview release of RStudio (https://rstudio.com/products/rstudio/download/preview/) -- otherwise, you may see RStudio crash when you try to use ggplot2.

Indeed, when you open RStudio it shows the warning message: "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 just a heads-up :)

Separately – and please let me know if I should spin this off into its own Phab task – I'm trying to connect to the wiki replicas but I got the following error (both in RStudio and in a Jupyter Notebook with R kernel):

System has not been booted with systemd as init system (PID 1). Can't operate.
Failed to create bus connection: Host is down
Warning message:
In system("timedatectl", intern = TRUE) :
  running command 'timedatectl' had status 1

The R code (using DBI and RMariaDB driver) was:

library(DBI) # install.packages(c("RMariaDB", "DBI"))
library(glue)

# see https://wikitech.wikimedia.org/wiki/News/Wiki_Replicas_2020_Redesign
#     https://wikitech.wikimedia.org/wiki/Help:Toolforge/Database
# 

wiki <- "enwiki" # English Wikipedia

con <- dbConnect(
  RMariaDB::MariaDB(),
  group = "client", # defined in ~/.my.cnf
  host = glue("{wiki}.analytics.db.svc.wikimedia.cloud"),
  dbname = glue("{wiki}_p")
)

dbListTables(con)

Maybe it's just me, though?

@mpopov is this requirement still valid, as far as you are concerned?

For context, I'm reviewing the current Jupyter offering and I am investigating whether or not support for RStudio would still be useful. Thanks.