Re: The xmlChunk-44 problem statement (resend)

Norman Walsh wrote:
> I think the issue xmlChunk-44 asks, essentially:
> 
>  1. Should there be a standard way to communicate context information
>     for a portion of an XML document?
>  2. If so, what should it be?
>  3. And to what extent should it provide a "canonical" form?

Thanks Norm, I think this makes the problem a lot clearer. There are a 
few related things that are covered in what you say but that I think 
should be mentionned explicitly as well:

  - What is the relationship of this issue to the dormant (but not 
dead!) XML Fragments Interchange[0] that's been in CR for just over 
three years now?

  - Should the context be extensible? There are zillions of ways to 
decorate a tree and one might wish to save more context than just that 
which the Infoset requires. CSS properties might have been inherited 
into the tree. There might be a need for PSVI context so as to type the 
fragment. One may wish to store some information about the rest of the 
document so that for instance XPath would still make sense of the 
fragment, and while you can't provide all the context XPath needs (since 
that would be the entire document) you could allow people to pick which 
subset of the rest of the document they need for certain things. It 
wouldn't be hard to produce a fairly long list of this, and it would be 
very hard to separate what one would consider to be legitimate context 
from what is going over border (and you're unlikely to get two people to 
agree on this). As such, room should probably be made for people with 
specific requirements to extend the contextual information in meaningful 
ways.


[0] http://www.w3.org/TR/xml-fragment

-- 
Robin Berjon

Received on Wednesday, 18 February 2004 11:37:29 UTC