Re: xmlmd:foo

"Simon St.Laurent" wrote:

> At 11:01 AM 5/30/00 -0400, Paul W. Abrahams wrote:
> >There's another piece of the puzzle that needs to be filled in: having
> some more
> >or less standard way of attaching metadata to namespaces.  That's what I
> >originally understood (according to the duck principle) that the namespace
> name
> >did.   TimBL initiated this whole brouhaha because that was his
> expectation also
> >(Tim, correct me if I'm wrong on that).  I think we need another attribute
> for
> >that purpose, and I nominate xmlmd:foo for the job.

> Could you flesh out this proposal a bit more?

It's pretty straightforward, really.  Here's the idea, modulo debugging and
criticism.  The xmldm:foo attribute is used in conjunction with the xmlns:foo
attribute.  It is a URL that locates (yes, really locates) a resource that
contains metadata concerning the namespace.  Much like an external entity in
an ordinary XML document.  That metadata might be a DTD, a schema, or merely
documentation.  Or it could be the location of a node of some kind whose
subnodes contain several different kinds of metadata describing or providing
validation for the namespace.

The design challenge here, but it's probably not very difficult, is
distinguishing among different forms of metadata through the nature of the
resource.  Here it's important to be as consistent as possible with
conventions used elsewhere.  Tentatively I would suggest that the name itself,
or possible some suffix of it similar to `.htm', indicate the nature of the
resource, e.g., `.dtd' for a DTD.  I don't think it's a good idea to elaborate
on the attribute name itself, e.g., `xmldtd:foo', but I wouldn't rule that
possibility out entirely.  I'm not committed to any particular solution to
this problem.

> I like it very much, but I'm concerned that URI-related issues identical to
> the ones we've been discussing here will crop up in xmlmd:foo work.

I don't think so, because we agree at the beginning that the URI isn't just a
name; it locates something.

> I'd also like to know what, if anything, your expectations for the resource
> described by xmlmd:foo are.

Pretty flexible.  Examples, as I said above, are DTDs, schemas, and
documentation.

> Addressing these issues seems like an important part of any 'next step'
> toward closing this discussion - I'm really hoping that this plan doesn't
> lead us back into the same thicket.

I think that's less likely, because we now know the thicket is there and where
the worst brambles are, from our traumatic experience with the namespace name
thicket.

Paul Abrahams

Received on Tuesday, 30 May 2000 12:42:12 UTC