[
Lists Home |
Date Index |
Thread Index
]
Uche Ogbuji wrote:
> [Joe English]
> >
> > Hm. I didn't think HLink was intended to work that way.
> >
> > I thought the intent was something like: someone developing
> > a new XML vocabulary wants to include HLink semantics,
> > so includes an HLink mapping along with the rest of
> > the vocabulary specification (schema, documentation, etc.)
> > Application developers who want to use the new vocabulary
> > consult the HLink mapping when building their own application.
> > Something like how WSDL works -- web service clients don't
> > download WSDL at runtime, the _developer_ does when _building_
> > the client.
> [...]
> I really don't understand any of this. I'll just latch on to the one concret
> thing that struck me.
>
> If HLink is like WSDL, then Dare is right about the security issues. These
> same security issues obtain with WSDL. Tainting a WSDL can cause subtle
> application failures (for instance, messing with the data type definitions in
> the <types> section). This is a security issue.
Apparently I'm very mistaken about how WSDL works.
I was under the impression that developers download
WSDL documents from service providers, feed them to
some kind of WSDL toolkit, point, click, drag, drop,
and out pops a working application that can access
the service; at that point the original Web Service
Description document is no longer needed.
Is that not the case? Does the end user need the WSDL
file at runtime as well?
> If HLink is not like WSDL, i.e. apps do not use it to affect processing durin
> instance processing, then it seems entirely useless to me. Why not just spel
> out the meaning of attributes right in the XHTML spec?
That would also work. But if the linking parts are factored
out into a separate specification, then it's easier to reuse
them in other vocabularies. As far as I can tell, HLink is
designed to aid XHTML modularization, not to enable blind
recognition of links.
> In either case, I don't remotely see how HLink is a potential replacement for
> namespaces.
I don't see that either, except in the sense that HLink
is a replacement for XLink, and XLink relies exclusively
on namespaces for identification while HLink adds other
mechanisms.
--Joe English
jenglish@flightlab.com
|
