Right now, we use "schema: metaschema" as a special indicator in parameter resources to specify that the parameter schema is to be read directly from the metaschema, and an instance-based resource schema will not be generated on algorithm execution.
I'm not sure we need to specify this in the resource at all. My initial thought with this was that we needed to be explicit about where we're reading a schema from, so a user looking at the parameter resource knows where to look. However maybe we shouldn't specify a schema at all in the params case? It would lead to a slightly cleaner resource, however we would lose the ability for a naive user to easily understand what's going on.
Another overlapping issue is that I think there is potential for some confusion about how and when schema instances get generated from metaschemas for tabular data resources on execution.
Right now, we use "schema: metaschema" as a special indicator in parameter resources to specify that the parameter schema is to be read directly from the metaschema, and an instance-based resource schema will not be generated on algorithm execution.
I'm not sure we need to specify this in the resource at all. My initial thought with this was that we needed to be explicit about where we're reading a schema from, so a user looking at the parameter resource knows where to look. However maybe we shouldn't specify a schema at all in the params case? It would lead to a slightly cleaner resource, however we would lose the ability for a naive user to easily understand what's going on.
Another overlapping issue is that I think there is potential for some confusion about how and when schema instances get generated from metaschemas for tabular data resources on execution.