RE: xmlns created literally

Owen Rees wrote:
> It seems to me that this ends up meaning that an XML namespace is
> identified by a string.

This is precisely what made a lot of people cranky when namespaces were
first rushed to standardization.  They look and smell like URIs, but they
are not.  There is no guarantee that they are dereferenceable nor that
anything useful will be there if they are.

The choice to use URIs was as a way of namespace allocation; if I publish
a schema with a URI at sun.com, I can reasonably expect Sun Microsystems
to, at some point, get around to suing me for trademark infringement at
the least.

This was inspired in part by the Java class naming convention; I was never
clear on why pseudo-URIs won the day over a different syntax. 
Familiarity, I suppose.

All that said, it is done now.  Users should just be *very aware* that
namespace so-called URIs are merely magic strings, to be used with
caution.

~Chris