Hi Folks,
Ouch!
I've just been bitten and I'd like
to share the lesson I learned.
Namespaces are great. I highly recommend them.
But here's the thing ...
Namespaces are not just important for markup,
they are also important for data.
...
Here is a short article I wrote describing the lesson I learned about the importance of using namespaces for both markup and data:
http://www.xfront.com/namespaces-is-for-markup-and-data/
Your comments are welcome.
Yuck! Just Yuck! You'll never catch me doing that, and moreover, I hope your idea doesn't catch on ;)
In the run-up to Balisage I've started carving out a bit of time to keep up with XML-DEV again, for the first time in years, and I've noticed all the activity looking for ways to "fix" XML namespaces (still). The past few years have reinforced for me the feeling I was starting to have that what we need is less of the ghastly mess we call namespaces, not more. Amelia ran down the problems quite well. XMLNS is a hotch-potch of crazy, inconsistent, illogical rules that even the experts can't reconcile, not to mention the poor folks to whom we recommend XML. QNames in content have compounded the problems and complicated every potential fix. Spreading that even more into the field of data and data modeling is the last thing we need.
Scope is a fundamental aspect of data modeling, and I don't think it should be handled by a syntactic device such as namespaces. It should be incorporated into the heart of data models, which would provide superior remedy to the issues you raised. In other words, you need more than XML Schema, and you should acknowledge that properly, rather than further stretching hacks that are already at the breaking point.
