Lists Home |
Date Index |
Mike Champion wrote:
> On Sun, 16 Feb 2003 17:35:52 +0000, Bill de hÓra <email@example.com> wrote:
>> We have two prescribed layers already:
>> 1: Infoset + W3C Schema + QNames
>> 2: RDF + Model Theory + URIs
> The second isn't even prescribed in any meangful sense by the W3CL.
If you trawl through the semweb side of W3C's house, you'll see the
second broadly defined, for some definition of meaningful. By the
end of the year the W3C will have signed off on the specs. It is an
alternative approach. Though perhaps the most interesting (and
underappreciated) option out there is FIPA's, which seems to sit
somewhere in the middle of the two.
> at www-tag ... the RESTifarians and the SemWeb people can't even agree
> on the details of URIs.
That's straightforward enough. Before we move past that permathread,
either some existing architectural constraints need loosening or the
20,000ft view needs to explicitly acknowledge a resolution layer. In
any case details of URIs are not stopping the higher level semweb
stuff being worked on (at that level everyone agrees that a URI
names a something), it's an engineering issue mainly.
> Uhhh ... There are PLENTY of difficulties with XML, as the most casual
> glance at the archives of this mailing list for the past 6 or whatever
> years will confirm.
Other than entities, I have very few problems when I stay inside XML
1.0 and a vanilla parser. It's the extraneous stuff (starting with
Namespaces) and the odd dissonant API call that tends to bite.
> I must be missing something -- a lot of people whose opinion I respect
> seem to be drawing the wagons around XML 1.x, warts and all. Serious
> question: what's driving this? Are people who practice XML 1.x now
> being forced by their customers, partners, tools, etc. to deal with the
> cruft that we've just complained about in the abstract for the last few
It was never abstract, not in a world where you have consume other
people's data as well as produce it. Too many people think base
interoperation starts with object models and type systems instead of
protocols and syntax.
Bill de hÓra