RE: How to Version XML Applications

Seems obvious but I may misunderstand you.  

For a schema to force a version 
number to exist or not exist, it must exist.  On 
the other hand, the document doesn't really care 
about that; the author has to.  So I don't 
know exactly what it solves architecturally.

What becomes a problem is is that meaningful given 
a document that includes multiple namespaces?
 
I don't really have a good idea for documents 
like that because the versioning of the primary 
document may no relationship to the 
versions of the ones included by namespace. 
That seems to be a thicket and a nightmare. 
Well... I guess one could include schematron 
assertions for coordinating version values as 
that is a sort of co-occurrence constraint.

We had this discussion on XML-Dev at least 
twice and after some time, there was no 
consensus, but my own conclusion was near 
to what you suggest: use a version attribute. 
It had the disadvantage of having to open 
and inspect the document, but oh well.  It 
still doesn't do anything for the multiply 
namespaced version unless those carry their 
own version attributes.  Ugly.  Again, it's 
like versioning a view instead of the query 
that creates it.  Essentially, what does 
"version" mean and when should it be a 
property?  What does version mean to identity? 
Is it the same as "representation"?

If the worms are still in the can, you 
can dream of going fishing.  Once you open 
them, you have to fish or cut bait. :-)

len

-----Original Message-----
From: Norman Walsh [mailto:Norman.Walsh@sun.com]

/ "Bullard, Claude L (Len)" <clbullar@ingr.com> was heard to say:
| Isn't that going to lead back to the potential of 
| having a document with multiple namespaces but no 
| normative schema? 

How so?

| That's a can of worms.

Versioning is a can of worms. But what good is a can of worms if you
never open it? :-)

Received on Wednesday, 11 September 2002 16:27:30 UTC