W3C home > Mailing lists > Public > public-grddl-wg@w3.org > October 2006

Re: section 2. Adding GRDDL to well-formed XML for review

From: Chimezie Ogbuji <ogbujic@bio.ri.ccf.org>
Date: Tue, 17 Oct 2006 13:17:52 -0400 (EDT)
To: GRDDL Working Group <public-grddl-wg@w3.org>
Message-ID: <Pine.GSO.4.60.0610171306080.8806@joplin.bio.ri.ccf.org>

On Tue, 17 Oct 2006, Murray Maloney wrote:
>> Also, regarding...
>>  TODO: explain why not <?xml-stylesheet ?>
>> Murray says: No explanation is needed. This is a design decision
>> that is in keeping with Web Architecture. PIs are deprecated.
>> That works for me, so I took the TODO out.

I have a *big* problem with that as there is a *substantial* precedent for 
this particular mechanism and browsers that support it.  Even though the 
specification for ?xml-stylsheet does not require user agents to support it, the reality 
is that most *major* browsers do interpret these processing instructions 
and they are very useful for *generally* associating XSLT transforms with 
source documents.  Ignoring this overlap would IMHO be irresponsible 
(and misleading - more on this later) for developers of GRDDL implementations who are aware of this precedent and 
*will* eventually ask:

"So why not <?xml-stylesheet ?> and how are GRDDL transforms different from transforms identified this way".

I don't think a paragraph explaining the following points is too much to 
ask to explain the need for a significant departure from the ?xml-stylesheet precedent:

1) ?xml-stylesheet is for the most part used to associate 'presentation' transforms,
whereas GRDDL associates 'semantic' transforms.

2) Browsers were never required to support ?xml-stylesheets (although most do).

3) The possibility of using both in tandem

The latter point is important IMHO, because seperation from presentation 
and content is orthogonal from seperation from interpretation (or 
semantics)  and content and so *not* deprecating PI gracefully would be 
irresponsible with regards to the senarios where seperation from content and presentation is the 
intended goal.  The reality of XSLT and how it is used with XHTML is that 
the this is the more common scenario and GRDDL will be the exception.

Chimezie Ogbuji
Lead Systems Analyst
Thoracic and Cardiovascular Surgery
Cleveland Clinic Foundation
9500 Euclid Avenue/ W26
Cleveland, Ohio 44195
Office: (216)444-8593
Received on Tuesday, 17 October 2006 17:18:06 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:39:09 UTC