> Of course that is more in agreement than in disagreement with your
> points. One of the implications of your breakdown, in fact, is the way
> it raises the question of scope. Historically, validation has been
> looked on as something that could be done on each instance
> independently and without reference to a broader set of instances (or
> indeed other 'dynamic' aspects of system state or context). But that
> isn't really the way the world works, at least not all the time.
>

That reminds me of a challenge in a past project where we wanted to validate a
document against different rules at different stages of its life-cycle.
(Whatbs acceptable as a submission from an author isnbt necessarily
acceptable as a finished publication).

Managing multiple schemas for the same instances is very cumbersome - how do
you keep them in sync?

One thing Ibve played with in Saxonbs XSD implementation is the idea that
a schema should be parameterized. Currently that only extends to variables in
assertions, but Ibve always thought it could and should be generalized,
e.g.

_minOccurs=b{if ($status=bdraftb) then 0 else 1}b

Meanwhile, if you need to do this kind of thing, there is always XSLT (either
to transform the instances, or to transform the schemasb&)

Michael Kay
Saxonica