Re: Level 3 XPath doesn't feel right

On Thursday, 10/31/2002 at 05:43 CET, Christian Parpart 
<cparpart@surakware.net> wrote:
> can't imagine _ANY_ usecase for a standalone XPath Module implementaion.

It's a separate module precisely because there are, and will be, DOMs that 
don't implement it, for various reasons. 

> It's usually for a W3 reference implementation to cover most parts
> recommented. 

Not all implementations are reference implementations. Not all 
implementations are intended to be fully general. The DOM very 
deliberately does not want to take an all-or-nothing approach; we'd rather 
folks be as compatable as they can be -- and at well-named, documented 
levels of compatabilty -- even if there are portions their customers don't 
need and that they don't want to support.

You're right, there will be market forces demanding more features. There 
will also be market forces demanding more efficiency in memory, including 
in code size. We can let the marketplace decide which trade-offs are most 
interesting.


Having said that: If someone has a DOM which does not want to implement 
the XPath module, do we really want to prevent their customers from easily 
working around that? To take your own examples, if Xerces hasn't yet 
implemented the DOM XPath API, do you really want to force them all the 
way to a completely different set of XPath APIs?

______________________________________
Joe Kesselman  / IBM Research

Received on Thursday, 31 October 2002 11:52:25 UTC