Installing R 4.0.3 on Raspberry Pi 4 with Ubuntu 20.10

[AuthorizedCustomer]

I am trying to install R 4.0.3 on a Raspberry Pi 4 running Ubuntu 20.10 64-bit. It is fully updated.

Before I did anything, I added the correct CRAN repository for this OS using sudo add-apt-repository 'deb https://cloud.r-project.org/bin/linux/ubuntu groovy-cran40/'.

When I run sudo apt install r-base, I get an error that r-base-core and r-recommended, both of which are specified to require version >= 4.0.3-1.2010.0, are "not going to be installed".

Checking them individually, it seems that r-base-core is the problem. If I try to install r-recommended by itself, I get an error that it needs "r-base-core (>= 4.0.3-1.2010.0) but 4.0.2-1build1 is to be installed".

If I check https://cloud.r-project.org/bin/linux/ubuntu/groovy-cran40/, I notice that r-base-core 4.0.3 is only available in an AMD64 variant, per the amd64 in the filenames. I observe that the r-base package's files do not have amd64 in their filenames. To me, that suggests that, currently, you're only running R 4.0.3 on Ubuntu 20.10 if you are on an Intel/AMD 64-bit platform. https://cloud.r-project.org/bin/linux/ubuntu/groovy-cran40/Packages seems to back that up.

Are there practical alternatives, other than to wait for an r-base-core 4.0.3 package to appear that is compatible with more platforms?

[andresrcs]

There are no recent binaries for R on Ubuntu repos, you have to compile from source, it is pretty simple, just use these commands.

sudo apt-get install -y gfortran libreadline6-dev libx11-dev libxt-dev \
                               libpng-dev libjpeg-dev libcairo2-dev xvfb \
                               libbz2-dev libzstd-dev liblzma-dev \
                               libcurl4-openssl-dev \
                               texinfo texlive texlive-fonts-extra \
                               screen wget libpcre2-dev
cd /usr/local/src
sudo wget https://cran.rstudio.com/src/base/R-4/R-4.0.3.tar.gz
sudo su
tar zxvf R-4.0.3.tar.gz
cd R-4.0.3
./configure --enable-R-shlib #--with-blas --with-lapack #optional
make
make install
cd ..
rm -rf R-4.0.3*
exit
cd

Edit: Sorry, I didn't take into account that you are using Ubuntu instead of Raspberry Pi OS (Debian based) so the system dependencies might be named different.

[AuthorizedCustomer]

Thank you. All the packages installed successfully, so your suggested packages appear to have been spot on. Ubuntu is also Debian-based, so I am not surprised it worked.

Running configure is failing with this:
checking whether zlib support suffices... configure: error: zlib library and headers are required

dpkg -s zlib1g says I have zlib 1:1.2.11.dfsg-2ubuntu4 installed.

Scratching my head and trying to figure this one out.

[andresrcs]

Sorry I can't be of help, last time I tested this was with Raspbian Stretch and I don't have a spare Raspberry Pi to test with newer OSs right now.

[AuthorizedCustomer]

I just figured it out. The fact that headers are needed suggests I need zlib1g-dev, too. I installed it and got past it. Finding I may need other packages, too. Working through it now, and will post update later.

[AuthorizedCustomer]

It works! All I needed were a few additional libraries. I posted the steps at https://www.arencambre.com/2020/11/29/installing-r-on-raspberry-pi-4-with-ubuntu-20-10-64-bit/.

If you would like to get StackOverflow points, you can answer my question at https://stackoverflow.com/questions/65056788/installing-r-4-0-3-on-ras. Otherwise, I'll put in basically what I have on my blog.

[andresrcs]

Nice!, in case you want a starting point to set up a fully functional data science server with R on a Raspberry Pi you can take a look at my installation guide, although, it is a little outdated since I haven't had the opportunity to completely update my setup.

Setting up your own Shiny and RStudio Server on a Raspberry Pi 3B+

I'm also planning to put together an Ansible Playbook to automate all these tedious processes but it is going to take a while.

[AuthorizedCustomer]

It seems like a container would be a good way to solve this?

[andresrcs]

I'm aware of that option, and for deployments, I think it is the best choice but for general use, I find a regular installation simpler to manage and easier to keep up to date, although, this is completely opinionated.