Lists Home |
Date Index |
> From: Evan Lenz [mailto:email@example.com]
> Let me rephrase the question: Is it too late to require
> QNames in values to
> be resolved with an application-level namespace declaration?
Ah. I see what you are getting at. I misread your initial post.
I still think it doesn't make sense to have one mechanism for attribute
values and another for element and attribute names. I think consistency for
this is going to be less complicated and less confusing than having one
mechanism for element and attribute names, and another for values.
I think the issue of whether both should not rely upon some
application-level declaration is a different issue from whether there should
be separate requirements for each. I think the former is highly problematic
as a general model, but is far more acceptable than having different models
for names and values.
The issue of relying upon some top-level declarations for namespaces vs. the
current model has been rehashed many times, though, and there have been use
cases brought up to defend the current model. If we are stuck with this
model, then I think it should be used consistently rather than mixing and
matching mechanisms for namespace declarations (although specific
applications may have a need to require some top-level declarations).