Each dev-env is defined in the corresponding Dockerfile and gets automatically built by Github Actions. The resulting Docker images get pushed to the ghcr.io repository and tagged with the build date, as well as latest. The environments can be pulled directly from the registry or further customized in downstream projects. For example, R projects will likely want to do the following:

FROM ghcr.io/artemsokolov/devenv:rstudio-latest

# Install additional packages
RUN sudo R -e 'install.packages(c("qs", "microbenchmark"))'

The list of currently available dev-envs is below.

Darkroom is a Python environment with a noVNC virtual desktop and preinstalled Napari and basic image processing libraries. The environment also includes a client for interfacing with Synapse for data management. The virtual desktop is served on port 6080. The suggested .gitpod.yml for this environment is

image: ghcr.io/artemsokolov/devenv:darkroom-latest
ports:
  - port: 6080
    onOpen: open-preview
  - port: 5900
    onOpen: ignore

The RStudio environment is effectively a replica of the popular rocker/tidyverse, with tidyverse and RStudio server installed on top of the default gitpod/workspace-base. This provides access to all standard Gitpod tools (including pyenv) on top of RStudio/tidyverse, thus enabling integration between R and Python with reticulate and rpy2. RStudio server is available on port 8787. The suggested .gitpod.yml for this environment is

image: ghcr.io/artemsokolov/devenv:rstudio-latest
ports:
  - port: 8787
    onOpen: open-preview
tasks:
  - command: sudo /usr/lib/rstudio-server/bin/rstudio-server start