Open tombentley opened 2 months ago
The output schema defines error
because errors are only strings. It doesn't define a subschema for annotation
because annotations can be of any value.
Wouldn't it be clearer to use annotation: true
to define the property in that case? At least that way annotation
is present in the schema. When I read the text I found it confusing that something was mentioned in the text (not in an especially clear way, imho) but not even present in the schema.
I see your point. Intentionality is worth something.
I'll go ahead and reopen this, but it'll be for an errata correction. I'm not sure how to make that change since the file has already moved on to the next version of the output. This might need to be done in the website repo.
I'm not sure how to make that change since the file has already moved on to the next version of the output.
There's branch for each release. You can make the change on top of that branch.
The 2020-12 spec says this about successful validations:
A later sentence draws an explicit contrast for the case of nested results :
But the output schema only defines a property for the plural
annotations
and not the singularannotation
.