> But still, it would be nice for validating implementations to not have to
> introduce hacks to work around this issue. I guess it's a pretty low
> priority erratum, though.
Obviously it's errata to record. However, I don't believe OASIS has a procedure to handle normative errata, only non-normative (or stuff you can finesse to be non-normative). This doesn't really qualify.
Going beyond that, there's the question of how to revise a schema, which is borderline impossible. Short of revving the version attribute inside the schema element, I would be at a loss as to what to do.
Lest this turn into a schema versioning thread, let's not and say we did. Nobody's going to change anybody's mind about versions, hints, namespaces vs. filenames, schema lookup, etc.
-- Scott
