W3C home > Mailing lists > Public > public-xml-core-wg@w3.org > March 2012

More for XInclude 1.1 [was: Minutes for XML Core WG telcon of 2012 March 7]

From: Paul Grosso <paul@paulgrosso.name>
Date: Wed, 07 Mar 2012 11:14:03 -0600
Message-ID: <4F57975B.8090508@paulgrosso.name>
To: public-xml-core-wg@w3.org


On 2012-03-07 10:52, Paul Grosso wrote:
>
>> 5.  XInclude 1.1--see http://www.w3.org/XML/Group/Core#xinclude
>>
>> On 2012 February 14, we published
>> XInclude 1.1 Requirement and Use Cases
>> http://www.w3.org/TR/xinclude-11-requirements/
>> summarizing the requirements and use cases for possible
>> enhancements to XInclude addressing the issues:
>>
>
> Norm did post something to the DocBook TC:
> http://lists.oasis-open.org/archives/docbook/201202/msg00001.html
>

And at
http://lists.oasis-open.org/archives/docbook/201202/msg00002.html
was a follow up posting that said:

> would it be possible to extend the
> element scheme so you could do:
>
> <xi:include href="foo.xml" xpointer="element(someId/*)"/>
>
> Short of the full xpath support that was in the xpointer()
> scheme, that would provide a lot of benefit for a small change.


I'm not positive I understand the request, but I made
a guess.

Of course, our working on XInclude doesn't mean we can
change the XPointer element() scheme.

The element() scheme already allows starting at an id and
then (optionally) adding a child sequence from there.  So
both element(someId) and element(someId/1) are already
allowed.

I'm guessing what he wants is to be able to use the
element scheme to xinclude all the children elements
of an element with a given id without including the
id'ed element itself.

While I can see how this might be useful, the element()
scheme only addresses a single element.  That's part of
its design.

But this line of thought leads me to something that could
perhaps be added to xinclude:  an attribute that says
"include the *contents* of all the elements addressed by
the xpointer attribute, but not the elements themselves".

I could see that as being a bit tricky to define (since
not all xpointer schemes return just elements), but if
it would be useful enough, we could consider it.

Thoughts?

Norm, can you check to see if doing this would, in fact,
address David's need?

paul
Received on Wednesday, 7 March 2012 17:14:38 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:16:44 UTC