RE: [xml-dev] Pure syntax vs the Infoset permathread (was Re: [xml-dev] The subsetting has begun)

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

And that puts the question I asked back squarely in your lap 
as a subsetMustHappenInStandardOrItWillHappenInWild supporter:

o  The Disconnect:  Why should SOAP request a subset 
   of the syntax if SOAP is described in terms of an infoset?

o  The Ubiquity Challenge:  Can a standard subset that includes both be 
   defined that will meet SOAP's needs and still be useful for 
   other applications?

and the kicker....

o  The Opportunity:  Should that be essentially the XML-SW as 
   defined by Tim Bray?

If the parties could agree to that last one, you might have 
a chance getting something done at the W3C before the next 
winter season.  But don't argue for standardization unless 
you are sure about

1.  A subset of the syntax
2.  A matching infoset
3.  A crystal clear realization that this would be a new 
    XML core in implementation if not normatively.

If it can be done at the application level in the application 
specification, it should be.  Don't argue standards for standards 
sake.  IME, that quickly makes them irrelevant.  Until the 
infoset and syntax questions are resolved, we should not 
touch the core.  We can't paper over the disconnect.  However, 
if, and I think it is so, that disconnect is really a point 
of view about XML processor ubiquity, then it is possible 
to take the SW as a starting position, and get a maximal 
victory with a minimum of work instead of a minimum victory 
with enormous amounts of rabbit threaded discussion.

len

From: Mike Champion [mailto:mc@xegesis.org]


Can the XML world agree on one and only one conception of the Infoset?  
Interesting question.  Probably "no" if everybody gets a vote and a veto. 
Probabably "yes" in the sense that once one takes root and the dust 
settles, the benefits of standardization will outweigh the costs of losing 
things like one's preferred view of where the namespace information is 
attached to the Infoset.