Re: Choose your namespace (Was : Personal view)

> Excuse me - NS already use URI syntax so I don't see what your point is.

Currently they use URI reference syntax and specify character for
character comparison. It is the latter feature that is the important
one for a namespace parser, not the former. So if having both was for
some reason unacceptable, then it would be more important to keep the
character for character comparison.

> These *cannot* be two different namespaces with different semantics - this
> is *exactly* a case A - what at one level (the naming authority) is
> identical, you want to be different at a higher level.

That isn't the `Case A' that James referred to.  The two URI are
different URI, the fact that they are equivalent in the http scheme
does not mean that they are the same URI or that they name the same
namespace. Using xpath one can distinguish elements in those two
namespaces and so could apply different styles to them using xsl or
any other namespace aware language.
An XSL processor must reject as a stylesheet a document in a namespace
that differs from the specified namespace in any way, including case
differences.

> What is OK to
> say is that a namespace parser doesn't have to KNOW that these are
> identical but you when writing the document cannot assign different
> semantics to the two names.

If the names are different then you can specify separate schema,
stylesheets and anything else. The fact that if you dereference the
namespace names you get the same thing is not relevant to namespace
processing, any more than the fact that it doesn't matter if you
dereference them and get a 404 error from an http server (which is the
most likly outcome, given currently specified namespace names).

> You pick the namespace
> you want with the semantics you want

The semantics I want are the semantics provided by the namepace spec:

For all namespace names, if the names supplied in the xmlns attribute
value are different as a string then it is a different namespace, if
the names are the same, then the namespaces are the same.

the simplest way to get that semantics is not to change the namespace
spec. But there are alternatives: "fixed base" doesn't quite give that
(./foo and foo become equivalent) but is a concession so that there is
always an absolute URI for use with rdf etc. "forbid" as interpreted
by most people, including I think Tim berners-Lee would give that.

But applying arbitrary uri scheme specific normalisations is just
completely out of the question.

David

Received on Monday, 19 June 2000 19:17:49 UTC