RE: What is the advantage of RELAX in comparison to Schemas?

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

Yes and no.  If done sensibly, context sensitive 
constraints in the schema are useful.  The problem comes 
when the information gets repurposed along a chain 
of users.  These rules can vary instance to instance, 
site to site.   Contracting for a schema that works 
in all of those sites is much harder if they have 
to share narrower and narrower context rules.  
It may be that the most practical approach is 
to separate out the context rules either into 
the application or into a separately cited 
document such as TREX, Schematron, etc.  

We know we need both but not both all the 
time and in all cases.  Again, the monolith 
definitions can strangle by attempting too 
much scope of control along multiple axes.  
Centralization gives the illusion of efficiency 
but quite the opposite is the case.  It simply 
leads to the emergence of cross-system checks; 
in organizations, committees that consume 
resources without creating product.  In systems, 
interfaces that consume system resources 
making sure every local system is behaving 
in accordance with global requirements, often 
when those requirements add nothing to local 
production.

The schema acts as a control. Scope accordingly.  
Look at use case for the schema and make very sure 
the constraints are shared.  Otherwise, let the 
locals add what they need as they need them.

Len Bullard
Intergraph Public Safety
clbullar@ingr.com
http://www.mp3.com/LenBullard

Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h


-----Original Message-----
From: Bob Kline [mailto:bkline@rksystems.com]

The most significant difference (from our point of view, at any rate)
between RELAX and W3C's Schema spec is that the former supports
context-sensitive content models.  Schematron also supports this
essential feature, which allows you to specify that a certain type of
content is allowed for a given element only if the element's parent is
<X>..</X>, or only if the element contains a certain attribute with a
specific value.  Very disappointing that W3C decided to leave this out.