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] XPointer crisis

[ Lists Home | Date Index | Thread Index ]

At 12:48 PM -0500 2/1/02, John Cowan wrote:
>David G. Durand wrote:
>>Xpointer seems to be under imminent threat of being gutted. There 
>>appears to be a significant chance that there will be no XPointer 
>>at all, or only a drastically cut down one. Particularly endangered 
>>are the ability to have ranges of any kind, or any forms of 
>>addressing other than "bare names" or maybe child sequence 
>This is overdramatized.  The question at hand is whether a more minimal
>form of conformance should be permitted, so that people who want to
>implement streaming XPointers don't have to handle full XPath syntax.
>I know we have a lot of "streamers" around: the question is, do we
>need to do XPath in XML fragment identifiers and XLinks in order
>to declare victory?

With respect, not everyone is phrasing the issue in that way, and I 
am not opposed to conformance levels as long as they are a sorting of 
what the committee has recommended, and not an elimination of 
significant parts. What worries me is that I have seen significant 
pressure to eliminate everything but the minimal level, and on top of 
that to remove the "scheme" mechanism, which provides the only sane 
way to ever extend the standard to cover other features.

>I think the minimal set for victory is: bare names, child sequences
>(like /1/2/3 or tag/4/5/6, walking the element tree), the latter
>two with character-data counts, and ranges consisting of two of
>any of these.  This allows you to do practical
>pointing and transclusion (nobody points to an attribute, or a
>namespace node, or a comment).  Check out the following:

I find John's proposed minimal level just fine (except that scheme is 
a must). The original fixptr proposal removed ranges, and on that 
account is unacceptable. Their proposal for range support is a vary 
bad way to modularize the problem (as well as being incompatible with 

I'd rather have two conformance levels than nothing.

>XPointer is in CR state, meaning that feedback from implementers is
>desired.  This has been lacking.

It's been less than hoped, but not completely lacking. And there is 
no set standard for how much feedback is enough in any case.

   -- David
David Durand                    |  12 Bassett St.
david.durand@ingenta.com        |  Providence RI, 02903-4628 USA
VP, Software Architecture       |  401-331-2014 x111 Cell: 401-935-5317
ingenta plc                     |  FAX: 401-331-2015
http://www.dynamicDiagrams.com/ http://www.ingenta.com/


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

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