Can ASDF validation handle newer JSON schema versions as ASDF schemas?

[ketozhang]

ASDF schemas are basically JSON schemas, but converted to YAML...

These schemas, though expressed in YAML, are structured according to the JSON Schema Draft 4 specification.
— https://asdf.readthedocs.io/en/latest/asdf/extending/schemas.html

I am interested JSON Schema Draft 2020-12 largely because Pydantic can generate them.

[braingram]

The current ASDF schemas mostly use the http://stsci.edu/schemas/yaml-schema/draft-01 metaschema which is largely https://json-schema.org/draft-04/schema with some additions for YAML and ASDF features.

There is not currently support for other jsonschema drafts. It looks like pydantic allows you to customize the schema generation (to allow you to generate schemas of other dialects). Have you tried this for generating either draft-04 or ASDF schemas?

[ketozhang]

Yes, I've started experimenting with this. The task is given a jsonschema 2020-12 schema, coerce it into the one you want. It's not a trivial task, and I need handle all the differences.

Would asdf.schema.check_schema() alone be enough to validate (i.e., check against $schema field)? I have yet failed to pass this check.

[braingram]

asdf.schema.check_schema is a start but given the allowable options for the schemas and how they relate to asdf (through the converters and tags) it's not sufficient to fully evaluate a schema. One reason is that the degree of schema flexibility is an option to the user (as is the case for the base jsonschema).

Would you share an example of the draft 2020-12 schema that you're hoping to convert? Also are you able to provide some more details about what you're hoping to do? Seeing the code would be ideal.