> You've got no prefix, but its in a
> namespace, which is inconsistent and requires special casing.
> Why couldn't they have just said that if you see an unprefixed attribute
named
> xmlns, that's all the signal that you need.
That would save special-casing during DOM construction, but add special
casing while actually operating on the DOM's contents. Remember, the
assumption we had beaten into us by the NS folks was "prefixes are semantic
sugar; you should NEVER look at a prefix while actually manipulating a
document contents."
Given how namespaces behave, and the fact that they were a significant
shift in XML semantics introduced after XML and the DOM was already in use,
I really don't think the DOM group did a bad job of handling them.
If I Ran the World, I would have wanted to fully develop the XML infoset
first -- including namespaces, links, includes, processing models and
schemas -- then devise the APIs and syntax, in parallel, to fit that
abstract model. Unfortunately we haven't had that luxury; instead we've had
the luxury of extremely rapid evolution and release. Tradeoffs...
As far as whether this should be being dealt with by the main parser body
or the DOMBuilder... The DOM doesn't yet adress parser APIs (though that's
on the list for Level 3) and I don't know what SAX says.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
