Re: Minutes for XML Core WG telcon of 2011 September 7

On Wed, Sep 07, 2011 at 11:47:52AM -0400, Norman Walsh wrote:
[...]
> > Extending Xinclude
> > ------------------
> > Norm sent email on behalf of DocBook at
> > http://lists.w3.org/Archives/Public/public-xml-core-wg/2011Jul/0013
> > asking whether we want to consider extending Xinclude.
> >
> > The key issue is that xincluding content can often cause
> > duplicate ids, and the question is whether we can define
> > some kind of fix up.
> >
> > See http://docbook.org/docs/transclusion-requirements/#uc-5
> > for a use case.
> >
> > See http://www.docbook.org/docs/transclusion/#d6e180 for
> > some DocBook proposals for id fixup.
> >
> > A processor might be able to find ids and/or idrefs if xml:id
> > is used and/or there is a DTD/schema available, and even if 
> > that is the case, fixup would only work within the included
> > module, but is that partial case worth augmenting XInclude.
> >
> > Liam suggests the only real solution is to do a transformation,
> > so he doesn't feel there is an XInclude solution.
> >
> > Norm will try to implement Jirka's proposals using xproc.
> >
> > ACTION to Norm:  Try to implement Jirka's proposals using xproc.
> 
> Norm concludes it's probably not.
> 
> Liam: I had in mind reimplementing XInclude in XSLT.
> 
> Norm: With doc-available or try/catch, I guess you just might manage
> that.
> 
> Liam: Another way to look at it is, is it a use case worth taking back
> to the XSLT WG.
> 
> ACTION to Norm: Try to implement Jirka's proposals using XSLT+XProc.
> 
> Norm: It did occur to me that if XInclude would just leave some sort
> of a marker on the top-level elements that it includes, then maybe
> that would be a hook for subsequent processing.
> 
> Jirka: That's what my implementation does, but it doesn't use XInclude
> elements because they don't have that semantic.

  Libxml2 does this too, internally it converts the old XML element
into an XInclude start node (which disapears when serializing), I
also added an XInclude end node to mark the end of the xincluded
fragment.
  The only problem is that this is incompatible with XPath as its
document model requires adjacent text node to be coalesced, so if we
keep the markers we can't merge the text node (if present).

Daniel

-- 
Daniel Veillard      | libxml Gnome XML XSLT toolkit  http://xmlsoft.org/
daniel@veillard.com  | Rpmfind RPM search engine http://rpmfind.net/
http://veillard.com/ | virtualization library  http://libvirt.org/

Received on Thursday, 8 September 2011 01:24:07 UTC