[
Lists Home |
Date Index |
Thread Index
]
Michael Brennan wrote:
>>From: Eric van der Vlist [mailto:vdv@dyomedea.com]
>
>>That's not true at all. If I open http://rddl.org or follow a link to
>>http://rddl.org in my browser, the URL bar displays
>>"http://www.openhealth.org/RDDL/" and I have absolutely no
>>way to know
>>which namespace is documented.
>
>
> That's just because it is doing a redirect.
Yes, and that's a common practice on the web.
> Surely the application (or user)
> knows what URL was initially used to obtain the resource.
Not that sure. What if the user has followed a link and hasn't seen the
location? What if it's a crawler which has found it at its redirected
location?
> <snip/>
>
>>But this would break the simplicity of RDDL and to do more complex
>>things, I would much prefer using RDF...
>>
>>My proposal is more than a dirty hack when you think about it. The
>>document (in whatever location if is) is a container which
>>documents a
>>target namespace by defining a bunch of resources. This translates
>>pretty well into RDF to associate resources defined into
>>different RDDL
>>documents describing the same namespace if you had to.
>>
>>And it makes RDDL documents movable which IMO is fundamental
>>as well: I
>>feel very uncomfortable with documents which meaning changes when you
>>move them!
>
>
> Yeah, that's why I liked the notion of extended links. But I also like the
> current simplicity of RDDL, so I've been inclined to think of an extended
> link variant as an optional alternative that can be used in place of RDDL in
> certain contexts (such as in an XML Archive file), but not intended as a
> wholesale replacement of RDDL.
I see RDF (or extended links) more as backend vocabularies which needs
to be hidden from the users and RDDL (and simple links) as frontend
vocabularies which can be exposed to the end user.
RDF (or extended links) are only a XSLT transformation away from RDDL,
assuming the necessary information is available and the target namespace
is just missing!
> RDF may fit here, as well, but I tend to lean in favor of linking rather
> than RDF assertions. XLink is much more approachable to most developers than
> RDF. Also, it's not clear to me how you would support both an arcrole and
> role for a resource locator using RDF -- unless you use an RDF/XLink
> combintation as XPackage does, and XPackage has a rather unweildy syntax
> IMO.
IMO, this is just a problem of syntax and both are rather equivalent. I
tend to prefer RDF as promoting the use of "soft" (or logical) IDs while
XLink using XPointer tends to promote more "physical" links or IDs but I
won't fight on this point: we have the choice and each of us can go his
own way :) .
> I see your point, though, and I seem to stand alone in my opinion that
> extended links could be useful for this purpose.
No. I think that in principle you are right but that extended links are
too complex to be used in RDDL :) !
Eric
--
See you in Barcelona.
http://www.xmleurope.com/2002/schedule.asp
------------------------------------------------------------------------
Eric van der Vlist http://xmlfr.org http://dyomedea.com
http://xsltunit.org http://4xt.org http://examplotron.org
------------------------------------------------------------------------
|