mnot’s blog

Design depends largely on constraints.” — Charles Eames

Saturday, 24 August 2002



Interesting; I’m glad thiswas written, because RDF is good stuff, and this is a good walkthrough.

However, it still doesn’t approach what I see as the critical problems with using RDF in RSS.

First of all, considering that it’s trying to enable the Semantic Web, RDF/RSS is pretty light on semantics. RDF is about making assertions about resources, which are identified by URIs; in this case, the channel is making assertions about the items’ links. However, there’s currently no way to attribute the statements made in the items to a particular channel; they’re just out there, because of the way that RSS uses RDF. In other words, if you subscribe to two different channels that make a statement about the same Web page, your RDF-aware parser won’t be able to say who said what, or even keep straight that title 1 goes with description 1, and so forth.

Second, RSS is fundamentally about an ordered sequence, but there’s no machinery to do what most people use RSS for - to have an ever-growing list of items that is added to as you fetch the feed. This is possible, of course, but the current RDF/RSS view is that you fetch a representation of the entire list; if you want to add it to what you’ve already got, some magic is presumed to happen in the background. This should be exposed (perhaps as a module?)