Hi Kevin (and others),

It's a good thing that you don't let up on this, I have been way to slow
getting to commenting on this module...

First: You start out your "Design Decisions for the RSS 1.0 Link Module" post
with a comment regarding links in RDF/RSS being a misconception. I'm not sure
how you got that idea, but in any case it's not the best note to start a
discussion on.

Simply put, I think you are wrong - if you want hypertext, use the popular
HyperText Markup Language, RSS is (partly) for describing resources, hence
the use of an existing framework for this, Resource Description Framework.
That was probably a little strong-worded, but I think the distinction is
important.

With that out of the way, RDF *does* provide "linking", just as the use of
rdf:resource in the module specification suggests.
The link element itself, however, introduces another layer to the RDF/RSS
equation, a layer that is not necessary.

You state about mod_link, that "it provides a mechanism to provide discovery
of external resources by user agents without having to explicitly enable new
link types within an implementing application."
You suggest that the magic sequence here is an element named link, but how is
the presence of an rdf:resource attribute not enough in this context? A user
agent could just as easily recognize these attributes, and do something just
as useful as the with a link element.

I think it's important that you take a step back and realize what you want -
a way to express some relationships of the channel (or item) in question.
What you should do instead of adding a too-flexible layer to RSS, is to use
the existing framework, by defining specific elements with fixed semantics.
Why couldn't a link element with link:relation="#alternate" be replaced by an
element named link:alternate (or something along those lines)?

I don't see any added value in the link element contruct, only added
confusion.

I do however see the value in the relationships you define, and the user need
for these should not be forgotten - they should just be implemented
"properly" in the RSS/RDF context.

Also, despite the pro/con section, I think it is clear that all the
mod_link-defined attributes should be replaced with their Dublin Core
equivalents - I don't see any that doesn't have one, and I see no value in
adding properties with the same semantics.


Regards,
Morten Frederiksen