RE: Substitution Groups Reconsidered

["Eve L. Maler" <eve.maler@sun.com>]

I've talked by phone with Phill a little while ago, and he tells me that I 
missed a very crucial 45 minutes during F2F #4 when I had to leave to join 
a phone call.  In that time, the TC apparently agreed that substitution 
groups are important in extension schemas because if you're extending a 
"null assertion statement" (the head of the chain of assertion types, which 
has nothing particular specified in it), xsi:type gives you no more 
information than a substituted element would.

I'm sorry I wasn't present for that discussion.  I do understand the 
argument (though I would submit that it's a pretty darn weak argument for 
substitution groups!).  I'm not morally opposed to allowing extension 
schemas to use substitution groups; I just (erroneously) thought this was 
something we had definitively decided against.

I'm sorry I was being dense about this.  (I do wish there were more detail 
in the minutes so I could have seen this change in approach...)  Phill and 
I did agree that it's a good idea to have some non-normative text somewhere 
explaining the benefits and costs of different extension approaches 
(substitution group vs. xsi:type).  Perhaps this is something I can add to 
my proposed schema design rationale document.

	Eve
--
Eve Maler                                    +1 781 442 3190
Sun Microsystems XML Technology Center   eve.maler @ sun.com