RE: lang implementation report

At 3:22 PM -0700 7/12/04, Jonathan Marsh wrote:

>We recognize that this is a cost-benefit tradeoff with room for
>legitimate viewpoints on each side.  We are willing to escalate your
>dissent through our Proposed Recommendation process if you'd like.
>Please let us know whether you are satisfied that we've considered your
>point of view adequately, or whether you'd like the Director to consider
>this point of contention during his review.
>

The practical side I'm OK with. It's a pain in the ass, but it's 
doable which I know because I've done it.

I am still concerned with the broader theoretical implications of 
this scheme though. In particular,

1. Using the XInclude spec to create a new generic Infoset property 
strikes me as very questionable. This should be done with a revision 
to the Infoset spec, not through the back door like this. The 
inclusions property is a little less objectionable because 1) It's 
clearly in scope for the XInclude spec. 2) It's unlikely to be of any 
interest to anyone beyond XInclude. 3) It has no impact on the 
serialized result document.

2. The new Infoset property is redundant with the values of the 
attribute properties for xml:lang attributes. This is a bad thing. It 
means they can get out of sync.

3. The idea that was expressed recently by Glenn Marcy that attribute 
inheritance somehow went beyond xml:lang, or that it could in certain 
circumstances, I think is actively harmful and not supported by the 
spec. 
<http://lists.w3.org/Archives/Public/www-xml-xinclude-comments/2004Jul/0003.html> 
I' don't know if this was just a personal opinion or the opinion of 
the working group. If it's just a personal comment, no big deal; but 
if this is the opinion of the working group then I think there's a 
much bigger issue hiding underneath the discussion of this one 
attribute that should be brought to the surface and discussed 
explicitly.

What I suggest is that the XInclude spec be rewritten so that the 
current behavior remains but is defined purely in terms of the 
appropriate attribute information items. I see no need to introduce a 
new property for the element information item to have the desired 
effect. It would be much simpler and more consistent with existing 
specs and APIs to define this purely in terms of attributes.

-- 

   Elliotte Rusty Harold
   elharo@metalab.unc.edu
   Effective XML (Addison-Wesley, 2003)
   http://www.cafeconleche.org/books/effectivexml            
   http://www.amazon.com/exec/obidos/ISBN%3D0321150406/ref%3Dnosim/cafeaulaitA

Received on Monday, 12 July 2004 19:42:28 UTC