OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: XInclude vs SAX vs validation

On Tue, Aug 21, 2001 at 12:41:02PM -0700, David Orchard wrote:
> But I do understand your point about the amount of machinery being higher 
> to support these use cases.  I would argue that if most xml parsers support 
> xpointer, then there's not too much overhead for xpointerized xinclude 
> versus nonxpointerized xinclude.  Kind of like XML Base.

  I think the *big difference* between an XInclude with XPointer
support is that it allows the importer to slelect precisely
what he wants to embed. It's like a collage, the artist is free to
pick precisely the part of the initial picture to compose its own.
It is very different from a system limited to global inclusion
or even including an ID, where there is a need of apriori contract
between the producer of the included parts and the designer of the
XInclude based document.

  The implementation time and processing cost has been pointed out
quite a few time, I think the former is a question of getting the
right toolkit, if you have XPath implementing XPointer is not very
hard (took me too weeks part time). The latter is a question of 
improving implementations, there is a streamable subset of XPath
(pointer ?) and this implies there is a streamable subset of XPointer.
If toolkits were good enough they would optimize this out in an
XInclude implementation, it's like any powerful language it takes times
for the users and implementors to understand the costs of each
features of the language and optimize to get the minimum.

  Should recursivity have been banned from languages because it's
harder to implement, or using it may fill up the stack, no, compilers
have learnt to use the right amount of resources for the job.

  The real question is: do people want to have the power to pick
       just what they need from their sources

In one hand if you have a 200MBytes document source, you might be
interested in selecting only a reasonable part of it, so the feature
will be useful, on the other hand if you have a small document source
using XPointer on it with current implementations is not a resource
problem so having that feature should not be a problem.

  My view is that XInclude is far more useful with XPointer than
without, because it doesn't require an explicit contract between
the producer and the consumer of the included resources.


Daniel Veillard      | Red Hat Network http://redhat.com/products/network/
veillard@redhat.com  | libxml Gnome XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/