It makes sense.  Thanks again.

> From: mike@s...
> To: andrew.j.welch@g...=3B dragon-fly999@h...
> CC: xmlschema-dev@w...
> Date: Wed=2C 9 Jul 2008 13:49:38 +0100
> Subject: RE: Include multiple complex types into one element ...
> 
> 
> > - the element can be reused (so LastName can be used (ref'd) 
> > in multiple places)
> > 
> > - the element is defined in one place (so changing the type 
> > of LastName only needs to be done once=2C not everywhere that it's used=
)
> > 
> > - the types are exposed to applications=2C so you can match on 
> > type define variables as that type in schema-aware XSLT or XQuery
> > 
> > - it follows a consistent style=2C and turns the upper part of 
> > your schema into a data dictionary :)
> > 
> 
> And one more=2C if you consistently avoid using local element declaration=
s=2C
> then it's easier to avoid the trap of writing
> 
> <xs:element name="person"/>
> 
> when you meant
> 
> <xs:element ref="person"/>
> 
> It's a really nasty mistake to make=2C because your schema document is va=
lid
> and it validates all your valid instances=2C but it doesn't catch the val=
idity
> errors that you intended to catch. That's because <xs:element
> name="person"/> defaults to <xs:element name="person" type="xs:anyT=
ype"/>
> 
> I don't condemn local element declarations entirely=2C however. But I ten=
d to
> apply the rule that if you use a local element declaration then it should
> refer to a global type definition.
> 
> Michael Kay
> http://www.saxonica.com/
> 
> 

_________________________________________________________________
Making the world a better place one message at a time.
http://www.imtalkathon.com/?source=EML_WLH_Talkathon_BetterPlace=