CANFAR Science Containers for POSSUM/VLASS

This repository hosts Docker container configurations for the VLASS* and POSSUM pipelines, optimized for deployment on the CANFAR Science Platform but more broadly employable as well. These containers are designed to facilitate scientific computing tasks in a reproducible and scalable manner. The repository is based on the original repository found here.

For POSSUM there are currently two Dockerfiles, which have identical software, except that the 'notebook' version installs jupyterlab, while the 'headless' version does not.

Usage

This repository utilizes GitHub workflows to automate the building and pushing of new containers every time a significant commit is made (see the Actions tab). It follows the Semantic Versioning convention MAJOR.MINOR.PATCH to label the containers.

The workflow process is as follows:

1. Make Changes: Make necessary changes or additions to the Dockerfiles or any other relevant files within the repository.

2. Commit Changes: Commit your changes with a message following the Conventional Commit Message format:

   • For a feature (MINOR VERSION UPDATE): git commit -m "feat(optional scope): description"
   • For a bugfix (PATCH VERSION UPDATE): git commit -m "fix(optional scope): description"
   • For a breaking change (!) (MAJOR VERSION UPDATE): git commit -m "featorfix!(optional scope): description"

3. Push Changes: Push your changes to the repository: git push.

4. Check for PR: After pushing changes, check whether a pull request has been successfully created. In Continuous Deployment (CD), the releaseer action will release a pull request, while the builder action will fail.

5. Merge Pull Request: Go to the pull request on the GitHub website and merge it.

6. Verify Build: After merging the pull request, the builder action should complete successfully, and the new version will be pushed to CANFAR.

7. Update Locally: To sync the changes locally, pull the changed log and .toml file: git pull.

8. Label Images: Finally, go to images.canfar.net (harbour) and label your images with the appropriate labels (e.g., headless, notebook) to ensure they can be accessed by users.

TODO

• Add functionality to automatically add labels once CANFAR supports this feature.
• 2026-01: I have heard through the grapevine that labels are optional

Contributing

Contributions to this repository are welcome! Please feel free to submit issues or pull requests for bug fixes, new features, or improvements.

Additional resources

The official CANFAR User Documentation can be found at: https://www.opencadc.org/science-containers.git/

Platform Infrastructure

The repository for the platform service infrastructure and deployment configuration is at: https://github.com/opencadc/science-platform

Additional version notes

Version 1.15.2+ have the newest version of possum2caom2 (as of Aug 20, 2025)

Version 1.14 and up have a downgraded version of astroquery to allow headless login

Version 1.13 and above will contain singularity/apptainer and screen

Version 1.7.5 Contains possum2caom2 as of 2024-08-20, which has the ingest set-up for the 3D pipeline.

Version 1.7 now contains fitsverify, part of libcfitsio-bin. Version 1.7.4 should contain a new version of possum2caom2 (caom2 to version 2.6.2 and caom2utils to version 1.7.3, 0.1.6 of possum2caom2) which (hopefully) fixes ingesting 3D FDFs and RMSF into CADC.

Version 1.6 (2024-07-16) now contains java (jre) and cadctap pyvo and vos as well. Note that pyfftw does not work with py3.12 as of 2024-07-23 but there is a separate container available on CANFAR which runs py3.11. Additionally there is also a container available that contains a 2022 version of WSclean (both not maintained on this GitHub page though).

Version 1.0 (2024-05-16) is now here! The POSSUM polarimetry processing is now in full-swing and working (so far), so it is time for version 1.0.