YAML files will need something for Dataset with the minimum props for a crate (none in process_run.yaml)

[ptsefton]

This is an issue for all profiles -- do they aim to specify enough to make a useful data package (I think they should) see this issue ResearchObject/workflow-run-crate#88

I would expect to see something like this with some help-text to encourage users to fill out contextual info like author/creator

classes:
  Dataset:
    comments:
    - The entry point to a workflow RO-Crate
    is_a: CreativeWork
    class_uri: wfrun:Dataset
    slots:
    - name
    - description
    - datePublished
    - license
    - mainEntity
    - author
    - publisher

[multimeric]

You're referring to the root data entity here, correct?

I suppose the best way to do this in LinkML is define this as a new class, because it doesn't make sense to re-define Dataset globally, which needs to be re-used for other directories. So I envisage something like:

classes:
  ProcessRunDataset:
    comments:
    - The entry point to a workflow RO-Crate
    is_a: CreativeWork
    class_uri: wfrun:Dataset
    slots:
      [...]
    - datePublished
   slot_usage:
     datePublished:
       required: true

Then they will need to declare their root data entity as a:

"@type": ["Dataset", "ProcessRunDataset"]

[ptsefton]

I think it would be better to treat this an new class that's RO-Crate specific - not tied to the particular profile. This profile does not need its own wfrun: prefix as it only uses terms from published schemas. So something like ROCrateDataset -- that your tools treat as a "magic" definition when generating validation shapes, modes, etc

[multimeric]

Good point. In this case it will be resolved when I'm able to make a schema for RO-Crate itself, which can define an implicit ,RootDataEntity class. Then profiles will be able to customize it.