schema_salad icon indicating copy to clipboard operation
schema_salad copied to clipboard

Allowing type duplications is not documented in the spec

Open tom-tan opened this issue 2 years ago • 3 comments

The spec of Apache Avro says:

Unions may not contain more than one schema with the same type, except for the named types record, fixed and enum.

As described in the spec of SALAD, SALAD is based on Apache Avro.

Salad builds on JSON-LD and the Apache Avro data serialization system

Also, there are few descriptions of union types in the spec of SALAD.

Therefore, we expect that union types behave as same as Apache Avro. That is, SALAD does not allow type duplications in union types. However, schema-salad-tool allows type duplications.

For example:

$graph:
- name: Foo
  type: record
  fields:
    - name: field1
      type: [string, string]
$ schema-salad-tool schema.yml 
/home/vscode/.local/bin/schema-salad-tool Current version: 8.4.20230606143604
Schema 'schema.yml' is valid

Is it an intended behavior? If intended, it would be nice if it is clarified in the spec of SALAD.

tom-tan avatar Jul 10 '23 06:07 tom-tan

Allowing duplicated types is not specifically intended behavior, but nothing in the current code specifically checks for uniqueness, so it isn't disallowed either. It has a minimal impact on correctness. If we want to explicitly disallow it, we can.

tetron avatar Jul 10 '23 13:07 tetron

I'm happy to accept a PR to formally allow type duplicates in union types.

mr-c avatar Jul 10 '23 15:07 mr-c

OK, I will send a request to clarify it.

tom-tan avatar Jul 11 '23 16:07 tom-tan