ccp-r

Lightweight Docker image definition for R development from VS Code or Orca.

This image is based on rocker/r-ver rather than rocker/rstudio, so it does not include RStudio Server. The goal is a smaller base image with the R, Quarto, and package set needed for psychology/statistics workflows.

Published image: ghcr.io/ykunisato/ccp-r

• R on rocker/r-ver
• Quarto CLI
• Japanese fonts (ipaexfont, noto-cjk)
• Toolchain and system libraries needed by the requested packages
• CRAN packages: languageserver, httpgd, renv, remotes, jsonlite, jtools, lme4, lmerTest, nlme, effectsize, effsize, pwr, psych, GPArotation, MASS, car, Hmisc, lavaan, semPlot, semTools, emmeans, multcomp, BayesFactor, brms, bayestestR, mice, qgraph, bootnet, networktools, IsingFit, IsingSampler, NetworkComparisonTest, EGAnet, psychonetrics, mgm, quarto
• GitHub packages: ykunisato/psyinfr, dstanley4/apaTables

docker build -t ghcr.io/ykunisato/ccp-r .

Start an interactive R session:

docker run --rm -it -v "$(pwd):/work" ghcr.io/ykunisato/ccp-r:latest

Open a shell in the container:

docker run --rm -it -v "$(pwd):/work" ghcr.io/ykunisato/ccp-r:latest bash

Check Quarto:

quarto --version

Check the VS Code integration packages from R:

library(languageserver)
library(httpgd)

• brms installs the R package and toolchain, but if you want to use CmdStan you should install it separately inside the container.
• The package installation script stops the build if any CRAN or GitHub package fails.
• The image intentionally avoids RStudio Server to keep the base image lighter.
• For GitHub Actions pushes to GHCR with GITHUB_TOKEN, the workflow or job needs packages: write, and the repository or organization settings must allow package writes.