W3C home > Mailing lists > Public > www-tag@w3.org > February 2003

Re: [xmlProfiles-29] TAG recommendation for work on subset of XML 1.1

From: Mike Champion <mike.champion@softwareag-usa.com>
Date: Wed, 05 Feb 2003 09:46:00 -0500
To: www-tag@w3.org
Message-id: <oprj4lyynpt3hq37@smtp.comcast.com>

[hatless, speaking only for myself]

Jonathan Borden said:

> We really need "XML" to be a common point of
> reference -- there doesn't seem to be anything else the collective "we" 
> can
> agree on.

It seems to me that "we" (the actual consumers of XML) don't agree on that. 
 XML subsets (whether they are "bugs" or "features"), profiles, alternative 
syntaxes that parse into infosets, etc. abound out there.  The XML "brand" 
is in danger of dilution/confusion because inconsistency (e.g, a message 
that validates against the SOAP schema but is not legal SOAP) is a fact 
that people stumble across and complain about.

There would seem to be two basic ways to deal with this.  One would be to 
get strict about conformance testing/certification for tools that claim to 
be "XML", and to be heavy-handed with WG's that develop specs that forbid 
legal XML constructs.  That doesn't seem pragmatic at all, IMHO.  AFAIK, 
there really aren't  many (any?) *fully* conformant XML processors out 
there.  Also, the damage to the credibility of the W3C if it were to derail 
SOAP 1.2 at this very late date would be profound, I believe.  I'm not even 
sure that a "standard" for SOAP that mandated full XML 1.x support would be 
widely implemented, making the whole exercise pointless. [I could very well 
be wrong on this point, I'm not a SOAP implementer!]

The other approach is essentially what the TAG recommends here -- accept 
the reality that customers are creating subsets/profiles out there in the 
wild and try to tweak the basic specs to rationalize the situation.  
Perhaps there is some way to accomodate Henry Thompson's suggestion that 
these be "conformance levels" rather than "profiles."  Maybe a "minimal" 
processor such as SOAP would parse but MUST NOT process entity 
declarations.  (Hmm, what would it do with references to those entities in 
the instance? SOAP's "draconian" approach may be much cleaner!). Perhaps 
there are some viable ideas there that would lead us to a better situation, 
but this was *extensively* discussed on the XMLP mailing list over the 
course of the last 2 years, and I for one saw the current SOAP 1.2 language 
as the best approach on balance.

Anyway, this is out of the TAG's domain now ... I guess we should be making 
our comments to the xml-core WG on this issue.  Can someone point to the 
most appropriate "comments" list on which to register opinions on the best 
way forward for XML itself?
-- 
Mike Champion
Received on Wednesday, 5 February 2003 09:47:27 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:16 GMT