How best to update / upgrade RStudio inside or underneath Anaconda Navigator?

[lysakowski]

This is cross-posted from # How best to update / upgrade RStudio inside or underneath Anaconda Navigator?. This is a different issue than what was posted by someone else on this forum. My delivery vehicle to the end user must be Jupyter Notebook or JupyterLab with Python and R kernels installed by Anaconda Navigator. RStudio is not the user's endpoint here.

The version of R Studio that is bundled with the latest release of the Anaconda distribution is over a year old, specifically version rstudio-1.1.456. If I check the Rstudio website current RStudio download, it says the latest version is RStudio-1.2.5033.exe.

I want to use the current RStudio release. Anaconda Navigator package manager doesn't give me an option to install any later versions. Also, Anaconda put my R-dedicated virtual environment (which I named Rdev') and installed R in a wierd place too: C:\Users\rlysakow.conda\envs\Rdev\Lib\R\bin`.

My judgement is that R and RStudio are basically UNIX programs that have been ported to Windows, and have all kinds of "leftover" Unix-izms and quirks in the way they reference files, directories, and scripts. Plus, Anaconda makes a lot of choices without always asking the user about where to put things.

It looks like to get a good RStudio update, I must update RStudio from within RStudio, but this will likely break a lot of Anaconda R functionality. RStudio doesn't use Anaconda Navigator's conda program to do the upgrade. The Anaconda Navigator's conda package index will likely not reflect the changes that RStudio makes.

Before I make a likely irreversible and breaking change to Anaconda, can anyone recommend a better way to do this Rstudio update and have Anaconda's package index correctly reflect changes that RStudio will make with executables and packages that it updates and installs?

Please do not recommend something like "ditch Anaconda and just use RStudio for all your R work", because that doesn't allow easy integration with Jupyter Notebook and the conda package manager inside Anaconda. I could install Jupyter Notebook separately, but then I have to manager that outside of Anaconda too. The end users of my application want to use Jupyter Notebook for their R work.

[technocrat]

Hi, and welcome!

I really like Anaconda. It's a great Python environment and package manager.

The interactions between RStudio and Anaconda are frequently mentioned here. I can't claim to have seen all of them, but the resolution is

Choose your poison

Let's focus on the end goal: Jupyter integration. After all, R package maintenance is almost trivial

update.packages(repos='http://cran.rstudio.com/', ask=FALSE, checkBuilt=TRUE)

(If one is so fortunate, this can even be a cron job.)

RStudio <=> Juypter is the goal.

RStudio => Juypter can be done through the rsconnect package. Now, whether or not that is satisfactory for that half is your call, of course.

Juypter => RStudio is available commercially through the RStudio Connect product for enterprise use at a price. Same with RStudio Pro Server

Are there open-source alternatives? Not so simple. I haven't test driven it, but it should be possible with retriculate with some bothersome overhang of saving to csv and reading into Python. Probably not insurmountable as a scripting problem, but a kludge.

I wish I had a better approach to offer. Although I've used both Python and R for over 10 years, I haven't spent much effort into trying to integrate them, so my take may have holes. I hope others will weigh in.