To namespace or not to Namespace ....

[David <dlee@calldei.com>]

I'm embarking on a project to create a schema (or more likely a 
collection of schema(i?) ) to model a dataset.
This data is currently represented in a relational database as 
semi-normalized in about 20 tables.
The underlying data is prety complex, but not in terms of 
attributes/elements but rather some hidden 'business knowledge' which is 
implied by
fields with coded values and things like shared "string pools".

The end result of this schema will be to be able to re-represent this 
data as XML for mainly internal use, although may be published in some 
form in the future
(only to partners, not the general public).

This is one of a handful of content sets that are already in XML or are 
already translated to XML, its the 'last dog' of enterprise business 
data to be translated to XML.

Up until now we've never used namespaces. We've been blissfully 
unaffected by not using namespaces.   XQuery and XSLT and pure Java 
programs have had no problem with having no namespaces.   The data is 
typically in "silos" and while sometimes referenced together, is not of 
the "module" sort intended to be embedded in other XML documents, but 
rather fully standalone data (which may reference each other with loose 
couplings).

I think I know the main reasons to use namespaces, and tons of reasons 
to not to ...
Equally  confused by Balisage's opening talk last year that (pardon the 
paraphrase)
"Best Practices for XML"
1) "Always use Namespaces:"
N) "Never use Namespaces" ...

So whats a poor XML geek to do ?

Any *practical* suggestions ?


-- 
-------------------------
David A. Lee
dlee@calldei.com
http://www.calldei.com
http://www.xmlsh.org