Re: More namespaces perversion

David Brownell (db@Eng.Sun.COM)
Wed, 07 Oct 1998 10:46:56 -0700


Eric Hellman wrote:
>
> David Brownell wrote:
> >In summary: all namespaces (ever) do is provide a layer of indirection.
> >Their meanings change over at least time. When you use namespaces you
> >need to consider what that change will do to your applications.
>
> It seems to me that the current NameSpaces draft adds TWO things to XML.
>
> 1. A layer of indirection is added.

Come to think of it, one of the ways that "XML Namespaces" seems
most odd to me is that it doesn't even do this. That is, there are
no semantics whatever associated with the "namespace": no formal
way in which they identify a set of bindings between names and
whatever it is that they denote. (That's wanting a "namespace"
to be a "context", yes.)

> 2. A mechanism to associate XML elements with URI's is added.
>
> It seems to me that not much attention has been paid to the things that can
> usefully be done with that second function. After all, there is no role
> specified for the Namespace URI.

Exactly. This means that there will be no standard way to make use
of the bindings (between element/attribute names and "meaning") that
are necessary. Everything is allowed ... there's not even a useful
way to compare the namespaces, given mechanisms like HTTP redirects
which can turn one URI into another.

- Dave