[
Lists Home |
Date Index |
Thread Index
]
From: "Eric van der Vlist" <vdv@dyomedea.com>
> > I am not sure this technical distinction helps the average schema
writer.
> > For example, rewriting an ambiguous schema so it is merely
non-deterministic
> > won't help XML Schema users. ;-}
>
> This chapter is in a book about Relax NG and its number one goal is to
> help Relax NG users :-) ...
Good point. Then you could probably omit that section on rewriting patterns
so they are deterministic (or is it unambiguous?). RELAX NG users seldom
need to do that. ;-}
> > The only case I can think of where the distinction makes any difference
is
> > in XML Schema union types, which are allowed to be non-deterministic but
are
> > not unambiguous because of the first-match rule.
>
> Or for using Relax NG tools (such as MSV) which can derive type
> information out of unambiguous schemas whether deterministic or not
> (more precisely, it will provide type information even if the schema is
> ambiguous but I wouldn't always advise to rely on it).
That's interesting. I assume you're talking about MSV when used as a SAX2
filter. What does MSV do when it has multiple valid types, pick one?
Bob Foster
|
