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


Help: OASIS Mailing Lists Help | MarkMail Help



   RE: [xml-dev] ANN: xpath1() scheme for XPointer

[ Lists Home | Date Index | Thread Index ]

At 02:33 PM 10/26/2002 -0400, Elliotte Rusty Harold wrote:
>Another tricky bit about using XPaths on post-included documents is that 
>you can get very different documents depending not only on whether the 
>XIncludes are resolved but how the different fallbacks play out. The most 
>obvious case is something like //chapter[1] which could select different 
>chapters depending on which includes got resolved or not. Furthermore, 
>they're a lot of XPath functions like string() or number() that take a 
>node-set as an argument, but really only operate on the first node in that 
>set, so which node is first can be very important, even if you don't see 
>an explicit positional dependence.
>I also suspect the following, sibling, following-sibling, and 
>preceding-sibling axes could well give significantly different results 
>depending on which includes got resolved and which didn't. I don't think 
>the ancestor, child, descendant, self, attribute, namespace etc. axes 
>would be affected.

I think I'm just going to have to note that these are possibilities in the 
I-D, and let people sort them out in their own processing systems.

I've made it quite clear in the past that I don't approve of XInclude's 
approach or lack of a concrete position in the XML processing 
stack.  Unfortunately, I can't fix those at all, and the most I can do is 
apply chewing gum, duct tape, and warning labels in specifications I 
develop which might be affected by XInclude.

(Somewhat paradoxically, I'm hoping that xpath1() will encourage the 
creation of more XInclude implementations, in order that these issues are 
encountered more frequently and best practices can start to evolve.)

>We're not just dealing with the question of whether or when to perform 
>XIncludes. It's a question of the results of performing them. With 
>fallbacks, it's no longer a binary success or failure operation.

Fallbacks make this a lot more complicated.  I think it's time to write 
that xinclude() scheme document just to walk through the issues raised by 
that processing.

Simon St.Laurent
"Every day in every way I'm getting better and better." - Emile Coue


News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS