Dragonfly docker files

There are some dragonfly dockers maintained in this repository, all based on ubuntu (switch branches to switch your base ubuntu version). Dockers built here are available from https://docker.dragonfly.co.nz. The dockers are:

1. dragonfly-base - Based on ubuntu, with New Zealand localisation and a deployhub user
2. dragonfly-tidyverse - A basic R docker containing the recommended R packages and R packages from the tidyverse
3. dragonfly-reports - The tidyverse, latex, and enough R packages to process basic Sweave files
4. dragonverse - A fat desktop sized docker with Potsgres, GIS tools, and a range of R packages

The docker files have the ubuntu version in their name, e.g., dragonfly-reports-20.04 and are tagged with the date that they were pushed to the registry. To list the tags of the available images, you can visit the registry URL for each repository: e.g., https://docker.dragonfly.co.nz/v2/dragonfly-reports-20.04/tags/list

To use these images in Gorbachev, put something like the following in your gorbachev.yaml file:

docker: dragonverse-20.04:2021-07-28

Your project will then be built with that image every time it runs on Gorbachev.

Running in a docker context

To use a docker to run a particular project, you can use the following command:

docker run -it --rm --net=host --user=$$(id -u):$$(id -g) -v$$(pwd):/work -w /work DOCKER CMD

where DOCKER is the name of the docker that you want to run, and CMD is the command that you want to be executed when the docker starts. This mounts the current directory into a directory called work and the command will be run with the same user and group as in the host. Make sure that you write any files that you want to keep somewhere under the work directory, otherwise you wll lose them when the docker finishes.

Updating the docker images

1. Change the date in the base image (the first line) of each Dockerfile to today's date
2. Update the Rprofile.site file in the dragonfly-tidyverse Dockerfile, changing the date so that R packages are from a recent version of MRAN
3. Change date in README.md (this file) to latest date.
4. Dockers will build on gorbachev and get deployed to docker.dragonfly.co.nz; deployment to docker hub is manual for now.

Rolling your own docker

The dockerfiles are examples of how to build your own docker images. In this case, R is built from the default version of the ubuntu repository. The R packages are all pinned at particular versions, in order to minimise surprises. You can build a dockerfile by running a command like the following:

docker build -t docker.dragonfly.co.nz/NAME:TAG PATH

Where NAME is the name you want to give your repository, the optional TAG allows you to specify the version, and PATH is the path to the directory of the Dockerfile (often times just .). Once you have happily built the docker, push it to the dragonfly repository with:

docker push docker.dragonfly.co.nz/NAME:TAG 

The dragonverse

The latest version is docker.dragonfly.co.nz/dragonverse-18.04:2020-06-04

This docker contains the following R packages:

Packages in dragonfly-tidyverse and dragonfly-reports

• tidyverse
• devtools
• dplyr
• ggplot2
• data.table
• xtable
• rmarkdown
• knitr
• pander
• kable
• kableExtra

Talking to databases and other formats

• foreign
• RPostgreSQL'
• rjson
• dbplyr
• data.tree
• ape

GIS

• shapefiles
• maptools
• rgeos
• mapproj
• rgdal
• sf
• rgrass7
• raster
• lwgeom
• mapview
• leaflet

Plotting and utilities

• RColorBrewer
• viridis
• gridExtra
• gridSVG
• ggrepel
• directlabels
• plotrix
• cowplot
• ggsn
• plotmapbox
• Rmdformats
• here
• furrr
• drake
• prettydoc
• paletteer

Bayes & Co

• rjags
• rstan
• brms
• coda
• ggmcmc
• bayesplot
• loo
• MASS
• lme4
• mgcv
• cplm
• tables
• testthat
• tidybayes
• tidymodels
• modelr