[rdfweb-dev] RE: Is it OK to produce valid-but-strange-looking RDF?

[Danny Ayers]

>> <item rdf:about="http://blah/blah.html">
>> <title>One Page</title>
>> ...
>> <rdfs:seeAlso>
>> <item rdf:about="http://blah/blah2.html">
>> <title>Another Page</title>
>> ...
>> </item>
>> </rdfs:seeAlso>
>> </item>
>
>You're putting an item inside an item? That's not RSS. No reader
>out there is
>going to pick up on it. I'd go further and say you could do such
>a thing in
>RSS-1.0 but you'd have to use the <items> element in the <channel>
>to assert the
><item> being part of the channel. Otherwise RSS has no support for an item
>having an item of it's own.

I'm not saying the item is inside the other item, only something like:

item1
title = one
...
item2
title =two
...

item1 seeAlso item2

The syntax I quoted (apart from the shortcuttery) is a perfectly valid RDF
way of saying this. (I left out the channel only for brevity - the channel
would contain all the items listed). The resources aren't in any way nested,
only the XML syntax. The nearest the spec seems to come to forbidding this
is with the DTD-like models given with each of the elements. Also it talks
of RSS 0.9 compatibility "libraries ignore what they weren't designed to
understand" - but are they meant to understand RDF or not? If it is the case
(which I assume) that RSS syntax is a particular subset of RDF/XML, then I
reckon the spec should be a lot more explicit - e.g. say that the DTD-like
models override RDF syntax.

Try this with your favourite RDF tool - open an RSS file containing two or
more items, add a seeAlso (or whatever) between the two items, then
serialize it out. IsaViz for one gives the 'item inside item' syntax.

Cheers,
Danny.