Re: [xml-dev] RE: Namespaces Best Practice

["Simon St.Laurent" <simonstl@simonstl.com>]

At 10:57 AM 9/20/2001 -0700, David Brownell wrote:
>But aren't there issues like "how far do you bubble down"?
>Normally I'll think that certain nodes in a given document will
>be natural spots to bubble down to, even if some decls could
>get bubbled down even further.

This is where the problems start to get interesting.  From the point of 
view of least-number-of-declarations, you need to check a big chunk of the 
tree to put namespace declarations in the element which is one above where 
the namespace actually gets used.

Putting namespace declarations in where they are used (tight scoping) works 
well when a namespace only appears once, but produces piles of declarations 
that could have been avoided when a namespace appears lots of time over the 
course of a document but is scattered.  (XLink tends to do that, but there 
are lots of other cases.)

It's the kind of problem humans are generally good at solving but which 
take some algorithmic tree-trickery to solve efficiently on a computer.

As a guideline, I'm pretty happy with "declare all namespaces on the root 
element whenever possible", while still recognizing that isn't always possible.

Finally, I don't think variable-scoping is a useful comparison for 
namespace-scoping in XML documents.  There aren't a lot of 
XLink-attribute-like things that get used in a scattered way across a Java 
program but are still in the same namespace, for instance.  Variable 
scoping is nicely contained by the logic of program flow, while namespace 
scoping may be much more arbitrary.

Simon St.Laurent
Associate Editor
O'Reilly & Associates, Inc.