Question: How can we store description / information about the container in a way that is easy for users to discover?
While container technology such as Singularity/AppTainer provides a nice way to aid reproducibility of research computations, it is not clear whether it already solves the problem of metadata on the container itself.
From user’s perspective, they may want to ask questions such as:
- Which software/programs/tools are available in this container?
- What are these tools doing?
- Which science domain(s) are these tools for?
- Do the tools in this container meet what my needs to carry out my research?
- How do I run these tools in the current cluster? Are there templates, sample scripts, etc?
- How do I learn more about these software tools?
- What to do to get help with the software tools, such as support, troubleshooting, tutorials and examples?
More technical people may be interested in things such as:
- Which versions are the software tools packaged in the container?
- How were the tools configured/built?
What have been available to suit these needs / support these inquiries? I am aware of container labels in Singularity, though not competent in using it.
If there is nothing else out there, we could get started by defining a simple YML or JSON file in the root dir of the container, perhaps?