Re: Moving on (was Re: URIs quack like a duck)

Michael Mealling wrote:

> >The namespace name is a character string, nothing more.  It
> >just happens to have the form of a URI, but in principle (though not
> >according to the namespace spec, of course) it could be any string that
> >we'd expect to be unique (i.e., no one >else would choose it by accident)
> >within the set of all possible character strings.   If we have another
> >attribute for namespace identification in the URI sense, then there's
> >no reason to demand anything more than literal, uninterpreted,
> >textual identification from the xmlns attribute.
>
> So relative URIs, being non-unique, should be deprecated?

I'd say they should be discouraged, but deprecation might be too strong.  If we
agree they're uninterpreted -- they just happen to have certain characters in them
like periods and slashes -- then the danger of using them lies in the likelihood of
accidental collision, just as if I name my widely-distributed namespace "foo".

> >So I suppose the direct answer to your question is that the scope is the
> >universe of character strings, but that's probably not what you're looking for.
>
> No, that's fine. I was hoping someone would say its the document but
> I kind of figured that wasn't the case...

Why were you hoping that?  Or have you changed your mind, so the question is
irrelevant?

Paul Abrahams

Received on Thursday, 1 June 2000 11:25:10 UTC