containerit

containerit packages R script/session/workspace and all dependencies as a Docker container by automagically generating a suitable Dockerfile. The package’s website is https://o2r.info/containerit/.

Prerequisites

• containerit only fully works if you have Docker installed and is only tested with Docker Engine - Community (previously called Docker Community Edition or Docker CE).
• R (>= 3.5.0) is needed so that some dependencies (e.g. BiocManager) are available; older versions of R predate the development of the package and were never tested.

Quick start

Try out containerit in a container

You can spin up a Docker container with containerit pre-installed if you want to try out the package. The default of the containerit images on Docker Hub is to start plain R, but you can also start an with RStudio session in a browser.

Note: The geospatial variant is used so that examples from containerit vignettes are likely to work. For a minimal Dockerfile to run containerit, see inst/docker/minimal/Dockerfile.

docker run --rm -it -e PASSWORD=o2r -p 8787:8787 o2rproject/containerit:geospatial /init

Now go to http://localhost:8787 and log in with the user rstudio and password o2r. Continue in section Use.

Install

Installation is only possible from GitHub:

# install.packages("remotes")
remotes::install_github("o2r-project/containerit")

Use

containerit can create Dockerfile objects in R and render them as Dockerfile instructions based on session information objects or runnable R files (.R, .Rmd).

suppressPackageStartupMessages(library("containerit"))
my_dockerfile <- containerit::dockerfile(from = utils::sessionInfo())
#> INFO [2019-08-20 10:10:52] Going online? TRUE  ... to retrieve system dependencies (sysreq-api)
#> INFO [2019-08-20 10:10:52] Trying to determine system requirements for the package(s) 'assertthat,backports,crayon,curl,desc,digest,evaluate,formatR,fs,futile.logger,futile.options,htmltools,httpuv,jsonlite,knitr,lambda.r,later,magrittr,mime,miniUI,pillar,pkgconfig,promises,R6,Rcpp,rlang,rmarkdown,rprojroot,semver,shiny,shinyFiles,stevedore,stringi,stringr,tibble,versions,xfun,xtable,yaml' from sysreqs online DB
#> INFO [2019-08-20 10:11:25] Adding CRAN packages: assertthat, backports, crayon, curl, desc, digest, evaluate, formatR, fs, futile.logger, futile.options, htmltools, httpuv, jsonlite, knitr, lambda.r, later, magrittr, mime, miniUI, pillar, pkgconfig, promises, R6, Rcpp, rlang, rmarkdown, rprojroot, semver, shiny, shinyFiles, stevedore, stringi, stringr, tibble, versions, xfun, xtable, yaml
#> INFO [2019-08-20 10:11:25] Created Dockerfile-Object based on sessionInfo

print(my_dockerfile)
#> FROM rocker/r-ver:3.6.1
#> LABEL maintainer="daniel"
#> RUN ["install2.r", "assertthat", "backports", "crayon", "curl", "desc", "digest", "evaluate", "formatR", "fs", "futile.logger", "futile.options", "htmltools", "httpuv", "jsonlite", "knitr", "lambda.r", "later", "magrittr", "mime", "miniUI", "pillar", "pkgconfig", "promises", "R6", "Rcpp", "rlang", "rmarkdown", "rprojroot", "semver", "shiny", "shinyFiles", "stevedore", "stringi", "stringr", "tibble", "versions", "xfun", "xtable", "yaml"]
#> WORKDIR /payload/
#> CMD ["R"]

You can disable logging:

futile.logger::flog.threshold(futile.logger::ERROR)

Now we create a Dockerfile for a specific R version and R Markdown file and do not add any packages already available in the base image:

rmd_dockerfile <- containerit::dockerfile(from = "inst/demo.Rmd",
                                          image = "rocker/verse:3.5.2",
                                          maintainer = "o2r",
                                          filter_baseimage_pkgs = TRUE)
#> Detected API version '1.40' is above max version '1.39'; downgrading
#> Detected API version '1.40' is above max version '1.39'; downgrading
print(rmd_dockerfile)
#> FROM rocker/verse:3.5.2
#> LABEL maintainer="o2r"
#> # CRAN packages skipped because they are in the base image: assertthat, backports, cli, crayon, curl, desc, digest, evaluate, formatR, fs, htmltools, httpuv, jsonlite, knitr, later, magrittr, mime, miniUI, pillar, pkgconfig, promises, R6, Rcpp, rlang, rmarkdown, rprojroot, rstudioapi, sessioninfo, shiny, stringi, stringr, tibble, withr, xfun, xtable, yaml
#> RUN ["install2.r", "fortunes", "futile.logger", "futile.options", "lambda.r", "semver", "shinyFiles", "stevedore", "versions"]
#> WORKDIR /payload/
#> CMD ["R"]

For extended instructions, see the vignettes at in the directory vignettes/, which are readable online at https://o2r.info/containerit/articles/.

Images

Images are available starting from different base images. All images are also available with version tags.

The Dockerfiles are available in the directory inst/docker/Dockerfile.

verse

docker inspect o2rproject/containerit

Base image: rocker/verse:3.5.3

geospatial

docker inspect o2rproject/containerit:geospatial

Base image: rocker/geospatial:3.5.3

RStudio Add-in

RStudio Addins allow to create interactive user interfaces for the RStudio development environment. Courtesy of a great contribution by a group of enthusiasts at the ROpenSci OZ Unconference 2018, there are several forms to quickly create Dockefiles from different use cases, e.g. the current session, a vector of expressions, or a script file.

Contribute

All help is welcome: asking questions, providing documentation, testing, or even development.

Please note that this project is released with a Contributor Code of Conduct. By participating in this project you agree to abide by its terms.

See CONTRIBUTING.md for details.

Development

r-hub builder is great for running checks, e.g. before submitting to CRAN and on other operating systems.

library("rhub")
rhub::check_for_cran()
rhub::check_on_windows()

You can build the pkgdown site with

pkgdown::build_site()

You can build the Docker images locally with the current development version using the following commands.

docker build --tag containerit:dev --file inst/docker/Dockerfile.local .
docker build --tag containerit:geospatial-dev --file inst/docker/geospatial/Dockerfile.local .

You can use pre-commit hooks to avoid some mistakes.

A codemeta file, codemeta.json, with metadata about the package and its dependencies is generated automatically when this document is compiled.

codemetar::write_codemeta("containerit")

License

containerit is licensed under GNU General Public License, version 3, see file LICENSE.

Copyright (C) 2019 - o2r project.