Re: How are correct, unambiguous results possible with implementation-defined XML pre-processing? (#faithful-infoset)

On Thu, 2007-05-24 at 11:34 +0100, Jeremy Carroll wrote:
> 
> 
> My own view is that the GRDDL spec is not clear enough about faithful 
> infoset not being wholly adequately resolved, but the resolution being a 
> pragmatic decision based on what we felt we could realistically do.
> 
> I suggest we reopen it and postpone - perhaps with the same resolution, 
> but restricted to this version of GRDDL, noting that it is an area where 
> future work may be needed. In doing that we can also draw the postponed 
> issue to the attention of the XProc WG, and formally give them the 
> option of (partially) addressing it.

I think that's a good idea.

I found a chance to discuss this with Liam and TimBL.

Liam is sympathetic to the GRDDL WG decision of 2007-01-31:
there are use cases for XInclude to go before GRDDL and for
it to go after GRDDL, so W3C shouldn't specify which
way it goes.

Tim is sympathetic to David's position: you haven't defined
a language until you say whether the xinclude-elaborated
infoset/xpath-datamodel is part of the meaning of the document
or not.

So if the GRDDL WG tries to claim that this issue is really
resolved, over a formal objection, we're not likely to get
Tim's support.

However, Tim sees it as a problem not just for GRDDL but for
every XML-based language/project, and
W3C has chartered the XML Processing WG to answer this question:

"In the absence of any guidance, what default processing, if any, should
be done in what circumstances?"
 --
http://www.w3.org/2005/10/xml-processing-model-wg-charter.html#Deliverables

So let's please re-open and postpone this issue.

> 
> 
> On David's specific question:
> [[
> But if the XML parser is permitted to expand the
> xi:include directive, before my GRDDL transformation even sees it, then
> I do not see any way to write my transformation such that it always
> produces the correct results.  In other words, short of superceding the
> GRDDL spec with GRDDL 2.0, I do not see how XProc or any other spec can
> solve this problem.
> ]]
> 
> While I don't claim to understand XProc at all, it is at least possible 
> for a spec other than GRDDL to say apply GRDDL to this XPath NodeSet, 
> and if it does so, (at least some of) the ambiguity problems go away.
> 
> My view is the overall problem space is too difficult to address 
> adequately, and the faithful infoset resolution is an appropriate 
> response to the difficulties.
> 
> Jeremy
> 
> 
-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Tuesday, 5 June 2007 20:12:37 UTC