Lists Home |
Date Index |
Yes. One can ask what we might call the lego question: is it
a logical fit and is there a component available? Are namespaces
only a means to disambiguate or a means to represent a different
component? I realize there is a working group looking at
components. We have to face this issue head on because
effective XML and effective application language design
are the same problem.
From: Leigh Dodds [mailto:firstname.lastname@example.org]
> From: Bullard, Claude L (Len) [mailto:email@example.com]
> Is it better to use a namespace or to copy an element type
> or attribute type definition into a new schema? IOW, by
> what criteria should one choose between a namespace reference
> and a clone?
> Maintenance is obvious but not a problem. I would suggest
> that it is cheaper to copy if the overlap among several
> schema is small (just a few common element and attribute
I'd agree with this.
I'd guess that the most common example at the moment of use
by reference is to include XHTML elements where one needs to
include markup text (e.g. product descriptions) within another
The overlaps might not be just be defined in terms of numbers of
elements/attributes, but also whether the 'concept' (for want of
a better term) that is being marked up is a natural subset of
the existing schema. For the XHTML example, this is paragraphs of
Another way to consider this and a potential advantage of this
kind of modularity is that these document fragments in another
namespace can be handed off to other components in a
pipeline. E.g. an HTML renderer. Reuse of existing markup definitions
allows reuse of these components, or where these don't exist,
provide a means of specifying a standard input for a new component.
Copying references increases the size of the interface to such
components or requires adapters.