Portrait of Edd Dumbill, taken by Giles Turnbull

Subscribe to updates

Feed icon Atom or RSS

or get email updates

What I make

expectnation
a conference management web application


XTech Conference
a European web technology conference

Have your pie and eat it

My relationship with RSS is pretty long and, as befits the technology, tortuous. Behind the scenes, I was one of the prime advocates of a return to Netscape's intention of using RDF for RSS. It seemed a perfect match to me, still does, and I argued as much within O'Reilly and in public during developers day at WWW9.

I also wrote an article in June 2000 on why RDF was suited to RSS, but unfortunately it has since dropped off the web (one more reason to take blogging into my own hands.) Happily, I found a copy of it thanks to the Internet Wayback Machine. It's basically the arguments that I made at the time of WWW9.

I'm still going to carry on using RSS 1.0, and thanks to the way we built it, I can. I can mix it with vocabularies like FOAF and re-use terms from successful projects like the Dublin Core. Happily, URIs can (and should) be forever, and nobody can take RSS 1.0 away from me.

So what do I think about Pie then, the project to throw away all the RSS specs and start over? Tim Bray writes about why he thinks it's the right time for this, and it's hard to disagree with the rationale.

Initially I thought "oh no, here we go again." Then I thought harder. One of the real difficulties with RSS over the years has been that of divergent goals. My personal use for RSS has always been close to the original intent of Netscape, which said the "SS" stood for "Site Summary", that is to say for description (metadata).

However, the predominant use in the weblogging community has really been for syndication. Indeed, many reinterpreted the acryonym RSS in this light. Where syndication of content is concerned, I will admit that trying to cram everything into RDF really did make things difficult -- when the line between sharing metadata and the actual data itself was crossed.

From what I can tell, it looks like Pie is about content representation and syndication, and most specifically doing that for weblogs. So, really, this is about kicking free of an ill-fitting heritage and addressing the problem without legacy constraints. For that reason, the initiative seems sensible.

It also takes the pie-makers into areas that have already been researched, and I hope they take the time to look at these and see if they can reuse or steal anything useful. For starters, I'd take a look at the work from the news industry on NewsML. Unfortunately NewsML are doing their best to discourage adoption through having an excruciating web site. This overview page on the IPTC site provides a better starting point. Interested readers might also take a look at my July 2000 article about XML in news syndication -- although the technologies mentioned have matured since then.

After some years in the open source and open standards world, I'm a little sceptical about the suitability of the consensus model for building something from the ground up. I do, however, wish the pie-makers well.

blog comments powered by Disqus


You are reading the weblog of Edd Dumbill, writer, programmer, entrepreneur and free software advocate.
Copyright © 2000-2012 Edd Dumbill