Currently in the skeleton code, the metadata for the data descriptor is just a dictionary. This needs to be more formalized into a more formal schema and/or representation of metadata. On the ROSE workflow side, this could be a full package that handles metadata if that level of complexity is desirable, but on the client side this should be instantiable via basic C++ primitives. At a minimum, the user should be able to define the metadata by using C++ methods on the object. This pathway should not close off the pathway for alternate methods (e.g. a JSON or YAML reader).
Currently in the skeleton code, the metadata for the data descriptor is just a dictionary. This needs to be more formalized into a more formal schema and/or representation of metadata. On the ROSE workflow side, this could be a full package that handles metadata if that level of complexity is desirable, but on the client side this should be instantiable via basic C++ primitives. At a minimum, the user should be able to define the metadata by using C++ methods on the object. This pathway should not close off the pathway for alternate methods (e.g. a JSON or YAML reader).