Can we start R and Jupyter from the same repo ideally using a conda env for both?
How should I make the distinction?
Can we start R and Jupyter from the same repo ideally using a conda env for both?
How should I make the distinction?
Have you looked at the R + Python Binder Example?
You can easily install both Python and R in the same conda environment. If it’s not working for you please can you post a link to your repo?
Thank you for your quick answers. This community is very active!
So I combined R and python in the same conda env. I followed the instructions to create specific rstudio and jupyter lab links in the R + Python Binder Example.
Here is the repo. I work on the ‘tr_rmd’ branch
For now, the jupyter builds sometimes but the Rstudio takes very long. Which doesn’t make so much sense to me, because it should be the same docker, isn’t it?.
I thought about speeding up the building with a separate environment repo. Should I then create two environment repos. One for R and one for python?
The binder loding hangs with
nothing provides requested r-base 3.6.1-3bionic**
You’ve got r-base=4
in your environment.yml
But you’ve also added runtime.txt
To follow-up on what @manics said, see here and my next reply below in that discussion for an idea of why adding runtime.txt
hijacks the R installation to make it incompatible with what is in your environment.yml
. The first example I gave you didn’t refer to the import caveats discussed here about not mixing the approaches. There specifically addresses install.R
; however, I think runtime.txt
triggers that same process and so it is also incompatible, I believe.
Thank you @fomightez and @manics
I removed the runtime.txt and the binder could be built! Great.
I need to figure out how to specify a different path. If you would have some advice on this it would be welcome.
I tried to use ?filepath=scripts?urlpath=rstudio
and ?urlpath=rstudio/tree/scripts to start Rstudio in the scripts directory but it didn’t worked.
I don’t think it’s possible to start rstudio under a custom directory. I might be wrong though.