Record vs RecordSet

Both BeerXML 1.0 and the preliminary BeerXML 2.0.1 make a difference between a Record and a RecordSet.

I don’t quite understand the json schema files, but, by looking at the examples in the test directory, I get the impression that everything is treated as a RecordSet in beerjson. Many filenames seem to indicate a single record but they include the plural mentioning the type of record, followed by a list object (although with a single element defining the specific record). Am I reading it wrong or is it really what you want?

For example, instead of the file HopWithRequiredFieldsOnly.json reading

  "beer_xml": {
    "version": 2.01,
    "hop_varieties": [
      {
        "name": "Cascade",
        "origin": "US",
        "alpha_acid_units": 5.5
      }
    ]
  }
}

couldn’t it read

  "beer_xml": {
    "version": 2.01,
    "hop": {
        "name": "Cascade",
        "origin": "US",
        "alpha_acid_units": 5.5
    }
  }
}

Is this allowed?

If not, although it would require more work, wouldn’t it be better to have a json-schema for single records and make the json-schema for record sets to have a list referencing the schema for a single record?

[krutilin]

Well, it is possible to make experiments in tests and check it. By default constructions without array will fail validation. We could rewrite root schema in this way:

            "recipes": {
              "description":"Records containing bla bla bla",
              "oneOf": [
                {
                  "type": "array",
                  "items": {
                    "$ref": "recipes.json#/definitions/RecipeType"
                  }
                },
                {
                  "type": "object",
                  "$ref": "recipes.json#/definitions/RecipeType"
                }
              ]
            },

to archive validation for arrays and single objects. But I can’t understand key reason for it. Brewing software can handle arrays, single objects and arrays with one item without any troubles.

[rmsrosa]

I managed to install the latest Node.js, which already comes with npm, and was able to locally run npm test. Pretty cool. I played a little bit with it, purposely making some erros to see how it goes. But I still need to understand the syntax of json schema to be able to modify it without making a mess. For instance, I tried to adapt in hops.json the modification above that @krutilin suggested on recipes.json to work with either an object or an array but I failed.

Anyway, if I succeed in understanding json-schema, then I might try addressing some of the issues, instead of only making remarks.

I will be following https://spacetelescope.github.io/understanding-json-schema/index.html to learn more about json schemas. It looks nice.