RE: [xml-dev] Interoperability [long]

["Michael Kay" <michael.h.kay@ntlworld.com>]

> Right, just having clear explanations of the "early" and
> "late" binding
> models more widely used would help.  Could you summarize how
> that's different from the XPath model?  I keep thinking of XPath as
> being a read-only addressing scheme, and the problem you described
> seems like it wouldn't show up in a read-only world.

XPath, of course, has two great advantages over DOM: it doesn't do in-situ
update, and it doesn't preserve compatibility with a pre-namespace model.
Nevertheless, the exact behavior of namespaces in XSLT/XPath, and in
particular the question of whether or not prefixes are significant, is an
extremely arcane subject probably understood by only a handful of people in
the world (those lucky enough to have had a private tutorial on the subject
from James Clark).

I have to concur with my colleague Mike C that namespaces in the DOM are a
pain. Saxon allows the source tree for a transformation to be supplied as a
DOM, and catering for all the different things you might find in terms of
namespace representation is a nightmare. (It's not the only nightmare,
someone just tripped me up by feeding me a DOM containing text nodes that
contained no text; but namespaces is definitely one of the worst areas).

Mike Kay
Software AG