R via Docker — A fresh start

Instructions below include

1. Cleanup of existing docker
2. Installation of a fresh RStudio Server via rocker/geospatial
3. Ensuring the addition of needed packages
4. Installation of SAGA-GIS in the new container
5. Some visual tweaks

Clean-up/Clear-out Docker

On my windows machine docker ended up taking up way too much space (e.g. well over 80Gb) and was crowding my hard disk space. I have obviously assumed you know how to get docker running on windows, If you need to figure that out see Docker.com and I found this quick video on YouTube with simple and clear instructions.

To rectify this I had to reset Docker’s virtual disk as per M. Roussy.

1. Open Docker GUI
2. Bug tab
3. Clean/Purge data (select them all)

These steps reduced the massive footprint that Docker was holding. Now I need to restore my go-to image.

RStudio Server with geospatial packages

Dockerhub is the place to find most any image that you can imagine. In my case I use Rstudio by posit with R and numerous geospatial packages for the R environment. I find this image has most of what I need:

rocker/geospatial

1. Download the docker image.
   In this example, I did not choose the lastest. As I want install additional packages going back a version-point may be more beneficial (e.g. to reduce occurrences of packages not being available for my version of R.)

   $docker pull rocker/geospatial:4.1.1
   

2. Create a new container

   
   docker run --name rockergeo ^               ## Name the container 
              -d                               ## detached mode
              -v windows/folder:/home/rstudio  ## share a file location with docker
              -p 8787:8787                     ## port to view Rstudio Server
              -e PASSWORD=somepassword         ## create the password
              -e TZ="America/Vancouver"        ## set the timezone 
              rocker/geospatial:4.1.1          ## docker image to use 
   

   • -v can be used numerous time to link various locations.
   • -e TZ I find this useful especially when generating reports. So the time generated is actually my time. A list of timezones is linked here

Your docker container is now running. To use it open your web-browser and navigate to: http://localhost:8787.

Add additional packages

Most likely not every package that you ever use is available in the default image, below I install what might be missing – based on my previous use of R.

Hopefully from another installation, or your past experience with R you have a list of packages that you use. To generate one:

## Save current list of packages
mypackages <- as.data.frame(installed.packages()[,c(1,3:4)])
write.csv(mypackages, "somewhere/mypackages.csv")

In the new docker session.

pks <- read.csv("~/somewhere/mypackages.csv") ## list from above

## installed already
hav <- as.data.frame(installed.packages()[,c(1,3:4)]) 

## to be installed
intl <- setdiff(hav$Package, pks$Package) 

install.packages(intl)

Rmarkdown tweaks

Some updating will be needed, especially if generating PDF or TEX based documents run:

tinytex::install_tinytex()

Also, remember that the initial pdf builds will likely take a little extra time as TinyTex will install the needed TEX packages.

Install of SAGA-GIS

I often use a number of SAGA-GIS modules. To make these available I need to:

1. From host machine open a terminal/console:

   docker exec -ti rockergeo bash
   

2. Update repositories and install SAGA ... (I should really sort out how to install the latest version of SAGA ...)

   sudo apt update
   sudo apt install saga
   

   Extra: Rstudio Themes

   I really like Garrick Buie’s rsthemes

   devtools::install_github("gadenbuie/rsthemes")
   
   rsthemes::install_rsthemes(include_base16 = TRUE)
   

   To make quick changes to the theme I add the following to my global profile. Note that on a fresh install this may be a blank file.

   ## to open the profile file
   usethis::edit_r_profile()  
   

   add and save the following. Then restart R (CTRL + SHIFT + F10)

   ## rsthemes from https://www.garrickadenbuie.com/project/rsthemes/
   if (interactive()) {
     rsthemes::set_theme_light("Solarized Light {rsthemes}")
     rsthemes::set_theme_dark("a11y-dark {rsthemes}")
     rsthemes::set_theme_favorite(
       c("Fairyfloss {rsthemes}",
         #"Oceanic Plus {rsthemes}",
         #"Nord Polar Night Aurora {rsthemes}",
         "a11y-dark {rsthemes}",
         "One Dark {rsthemes}",
         "Solarized Dark {rsthemes}",
         "base16 Unikitty Dark {rsthemes}",
         "base16 Apathy {rsthemes}",
         "base16 Flat {rsthemes}",
         "Solarized Light {rsthemes}",         ### start light
         "base16 Atelier Lakeside Light {rsthemes}",
         "base16 Atelier Seaside Light {rsthemes}"
       )
   
     )
   }
   

   Finally, As suggested on the website add a keyboard shortcut to switch themes. From the Rstudio menu: Tools >> Modify Keyboard Shortcuts .... Search for Next Favourite Theme and add a shortcut. I use Ctrl + Alt + N.