On Nov 17, 2025, at 20:59, Mahesh Jethanandani <[email protected]> wrote: > > What happens to attribute that a tool does not recognize?
That is a great general question. XML and JSON have developed very different default answers. RFCXML is an XML vocabulary, and one that aligns very closely to the traditional XML answer that is dominated by the concept of validation: a processor does not accept input that cannot be validated (we use Relax-NG grammars to describe what valid instances are). (The JSON philosophy is “ignore unknown”, which gives you great extensibility, but limited interoperability, because there is no way to express that some new feature really must be understood to perform some processing.) > Does it ignore it, or does it stop processing the document? In xml2rfc, you get a validation failure very early in the processing by xml2rfc, which doesn’t even get a change to look at information that doesn’t validate to the grammar in the first place. > Based on what Carsten says in this thread "xml2rfc will simply croak when it > sees that”, it appears we have to campaign to get xml2rfc to supporting the > new attribute of ‘role’. But please confirm. Yes. There are ways to combine full validation of some things with “ignore unknown” of other things even in XML, as shown by the way Julian uses an x: namespace to feed extension information to his tools. Grüße, Carsten _______________________________________________ netmod mailing list -- [email protected] To unsubscribe send an email to [email protected]
