Re: Editions on p:make-absolute-uris

Hi Vasil,

> Still, there's no reason not to allow all sorts of nodes (text nodes in
> particular) be matched, is there? And I still don't see why should an error
> be raised if no node(s) are matched.

I think that the wording's a bit off: what it's really trying to say is 
that it's an error if the pattern *might* match something other than an 
element or attribute (or possibly if it *does* match something other 
than an element or attribute). I certainly agree that the current 
wording implies that it's an error if nothing gets matched, and that 
that shouldn't raise an error.

I'm torn on allowing other kinds of nodes.

I think the argument for the status quo is probably that only elements 
and attributes can/should hold data: that it would be bad practice to 
have a markup language in which you'd *need* to match text nodes, 
comments or PIs. For example, for matching text nodes to be useful, 
you'd need a design like:

   <links>index.html<sep />archive.html</links>

which isn't something we'd want to encourage. Also, we're never going to 
be able to do *every* kind of absolutisation with this step, so we 
should perhaps only focus on the really common ones.

On the other hand, I generally believe in giving people tools even if 
they might use them incorrectly. And I've certainly used PIs that hold 
URIs, so I have some sympathy with providing support for all kinds of nodes.

Hopefully Norm will add it to the agenda and we'll see what the rest of 
the WG thinks.

Cheers,

Jeni
-- 
Jeni Tennison
http://www.jenitennison.com

Received on Friday, 10 August 2007 21:09:04 UTC