How neurodesk/neurocontainers works
Main Components
• CI/CD Pipeline: Represent the GitHub workflows (located in .github/workflows) that trigger builds, tests, and issue management. These workflows (e.g., auto-build.yml, build-apps.yml) automate the container builds.
• Container Builder: The interactive_builder directory holds scripts (e.g., extract.py, run_interactive_builder.sh) that may be involved in processing or customizing container builds.
• Recipe Modules: The bulk of the repository is the “recipes” directory that contains subdirectories for each neuro-tool (e.g., afni, ants, fmriprep, freesurfer, etc.). Each recipe includes build scripts, test scripts, and documentation for its respective tool.
• Documentation: The docs folder (and recipe-specific READMEs) gives the requirements and guidelines for containerizing these applications.
• Repository Configuration: Files like .dockerignore, .gitignore, LICENSE, and other CI/CD related configuration files that guide the overall process.
• External Integrations: Connections to external documentation sites (neurodesk.org) and possibly artifact repositories (such as Docker Hub) where the built containers are pushed.
Component Relationships and Data Flow
• CI/CD then invokes the Container Builder and associated build scripts within each recipe directory.
• After the build process, successful container builds are pushed to the corresponding container registry or made available for Neurodesktop integration. Show this as an external system.
Architectural Patterns and Design Principles
• Automation: The CI/CD workflows encapsulate automated building and testing, emphasizing a “pipeline” or “chain of responsibility” pattern.
• Separation of Concerns: Distinguish between configuration, build scripts, tests, and documentation.
Relevant Technologies and External Dependencies
• GitHub Actions: Used for building, testing, and manually triggered processes.
• Docker/Container Technology: Since container building is core, include Docker as an underlying technology.
• Scripting Languages: The use of shell scripts (.sh), Python (extract.py), R (dependencies.R) and, in some cases, Julia, MATLAB scripts in some recipes.
• External Documentation: Link to Neurodesktop and Neurocontainers websites as part of the ecosystem.