idea: json-schema based validation

Hi,

I've used this library in a small project and it server really well for my purposes. I'm now considering its use in larger projects, but the idea of having two sets of rules (front-end and back-end) in order to validate the same model seems to violate the DRY principle.

json-schema are useful because they provide a formal description of the data a given service is supposed to produce and/our consume. Depending on your development methodology, you can use it to automatically generate Java (or other language of choice) objects that can be de/serialized to the a compliant JSON representation. You also have the option to generate the schema from existing classes, though this path has some limitations.

Given this scenario, I was wondering that it would be very nice if I could derive validation rules straight from a json schema.

So, instead of this:

mymodel.observable1.extend( rule );
mymodel.observable2.extend( rule );
...
mymodel.a.nested.observable2.extend( rule );
mymodel.a.very.very.very.very.deepply.nested.observable2.extend( rule );
...

One could just do like that:

mymodel.extend(schema:someJsonSchema)

where 'someJsonSchema' is a json-schema object retrieved somehow.

There are a number of json-schema javssscript implementations, but none of them (as far as I know) can deal with observables(). One way would be to simply unwrap the structure and use the existing libraries, but this seems inefficient, specially for complex data structures.

WDYT ?

The issue names no files, tests, or entry points. Start by reviewing the existing observable-level extend(rule) validation API and how models and nested observables are handled; the work would need a defined schema-based API, supported JSON Schema behavior, and tests showing validation of nested data.

Written by the indexing model from the issue text.