Missing references should not return true, IMO.

[sprugman]

I just noticed that this was by design, based on the tv4.missing demo. Is that in the spec? It seems backwards to me -- I want missing reference definitions to fail. Is it possible to make that an option? Do you know of another (good), browser compatible validator library that works that way?

TIA.

[ramiel]

This is a problem for me too. And if the reference do not contains a full uri (which is permitted in the spec) the external reference doesn't appear using getMissingUris so I cannot neither know an external reference is missing.

[geraintluff]

Nice suggestion making it an option - I'll work on that.

[sanear]

+1 for this - any kind of silent failure in a validator is a bad idea... the default assumption has to be that the schema is invalid unless proven otherwise; an invalid reference certainly doesn't suggest that everything is fine.