PUBLIC/SYSTEM distinction (was Re: typo/bug in the namespace spec)

At 10:38 AM 6/11/00 -0400, Michael Mealling wrote:
>On Sun, Jun 11, 2000 at 10:29:03AM -0400, Simon St.Laurent wrote:
>> So it sounds like a lot of our problems come from what's the equivalent of
>> dereferencing a PUBLIC identifier without a catalog file, just because it
>> looks like a SYSTEM identifier...
>
>So what happens if I write a URN namespace* for FPIs? I.e.:
>urn:fpi:-//IETF//DTD BLA//EN 
>(which would look kinda funky once encoded but you get the idea)
>
>URNs as defined do not require resolution. They just name. 
>But there is a resolution system available. I can take that
>URN and find a server (if one exists) that is authoritative for that FPI
>and ask it all sorts of questions. I can ask it for an XML Schema,
>I can ask it for a DTD. I can ask it for a picture of the author.
>
>I can also type that URN into a hypothetical search engine and find
>out other places where software might be available for processing
>that namespace. One could even setup that search engine such that
>it only does this service and returns all sorts of interesting
>third-party tidbits of information about that FPI. 
>
>Does this describe kind of what your after?

It does, to some extent, serve as a replacement for FPIs.  However, the
problem isn't simply that we're treating PUBLIC identifiers as SYSTEM
identifiers - it's that we only have one space for both kinds of
information.  While the spec makes clear - to me, anyway - that the NS name
like a PUBLIC identifier, there are those who would prefer that it be
treated like a SYSTEM identifier.

Using URNs can make clear the PUBLIC id-like nature of namespaces, but as
long as the rest of the URI possibilities are available, we'll have
interoperability problems caused by those who treat namespace identifiers
as SYSTEM ids, both using dereferenceable URIs and (worse) expecting URIs
to be dereferenceable.

That's the bigger problem, the one that availability of URNs doesn't solve.

Simon St.Laurent
XML Elements of Style / XML: A Primer, 2nd Ed.
Building XML Applications
Inside XML DTDs: Scientific and Technical
Cookies / Sharing Bandwidth
http://www.simonstl.com

Received on Sunday, 11 June 2000 10:53:59 UTC