pracpac: Practical R Packaging with Docker
The goal of pracpac is to provide a usethis-like interface to create Docker images from R packages under development. The pracpac package uses renv by default, ensuring reproducibility by building dependency packages inside the container image mirroring those installed on the developer’s system. The pracpac package can be used to containerize any R package and deploy with other domain-specific non-R tools, Shiny applications, or entire data analysis pipelines.
Installation
Install pracpac from CRAN with:
install.packages("pracpac")
Install the development version of pracpac from GitHub with:
# install.packages("devtools")
devtools::install_github("signaturescience/pracpac", build_vignettes = TRUE)
Example
Detailed documentation is available in the basic usage and use cases vignettes:
In the most simple example, running use_docker()
inside of a package directory will (1) capture all the package dependencies installed on the developers system using renv, (2) build the package source tar.gz, (3) create a Dockerfile which builds an image with the package and its entire dependency chain, and (4) optionally builds a Docker image with tags set using the version in the package DESCRIPTION
. The Dockerfile, renv.lock
file, and the package source tar.gz file will all be placed into a docker/
subdirectory of the package, which is added to the package’s .Rbuildignore
. The workflow is shown in the figure below.
For example, running use_docker()
in the example package included in pracpac at inst/hellow will produce a Dockerfile with the following contents:
FROM rocker/r-ver:latest
## copy the renv.lock into the image
COPY renv.lock /renv.lock
## install renv and biocmanager
RUN Rscript -e 'install.packages(c("renv","BiocManager"), repos="https://cloud.r-project.org")'
## set the renv path var to the renv lib
ENV RENV_PATHS_LIBRARY renv/library
## restore packages from renv.lock
RUN Rscript -e 'renv::restore(lockfile = "/renv.lock", repos = NULL)'
## copy in built R package
COPY hellow_0.1.0.tar.gz /hellow_0.1.0.tar.gz
## run script to install built R package from source
RUN Rscript -e "install.packages('/hellow_0.1.0.tar.gz', type='source', repos=NULL)"
And an renv.lock
with the dependencies of hellow
(in this case just the praise
package):
{
"R": {
"Version": "4.0.2",
"Repositories": [
{
"Name": "CRAN",
"URL": "https://cran.rstudio.com"
}
]
},
"Packages": {
"praise": {
"Package": "praise",
"Version": "1.0.0",
"Source": "Repository",
"Repository": "CRAN",
"Hash": "a555924add98c99d2f411e37e7d25e9f",
"Requirements": []
}
}
}
By default, use_docker()
does not actually build the image. You can build the image with build_image()
after running use_docker()
, or in one step using use_docker(build=TRUE)
. This two-step procedure is useful because other use cases may require edits to the Dockerfile to install system libraries, or copy in Shiny app or pipeline-specific code. See the help page for ?use_docker
and the use cases vignette (vignette("use-cases", package="pracpac")
) for details.