Nothing now, by definition. We've defined all the required semantics now and compliant processors have to understand everything there is now.

But I could easily see an example where we added a feature that required special processing by a resolver or proxy/etc for which a failure should be generated if such processing could not be accomodated. For example, maybe there are policy statements (as extension elements in the XRID) about how often an authority should be contacted (not more than once a minute, for example) - and if you as a resolver don't understand that policy statement, then you shouldn't even attempt to use the XRID and should just throw up your hands and fail.

For me, this is a judgement call - nothing will break today if we don't have it - I'm just trying to anticipate extensions, etc that might come back to bite us if we don't have a way of making them explicit.

I'd be curious to hear other opinions. If this is something others don't see a need for, then we can probably pass. If people are designing extensions with mandatory behaviors, they'll just have to find other mechanisms to induce failure by resolvers if they resolvers don't understand the extensions that require the mandatory new behavior.

And of course, you may rightly believe that we'll never have such extensions, or that we should strongly discourage them. I'd be open to those opinions as well.

-Gabe