Recovered from the Wayback Machine.
The fluff about namespaces in RSS 2.0 seems to have boiled down to: the major version number should have warned everyone that this version of the specification isn’t compatible with previous versions. The solution: generate both sets of Userland RSS (0.9x and RSS 2.0) until aggregators can properly work with the namespaces.
Tim Bray wrote in comments at Ben’s:
The best suggestion I’ve seen so far in the thread above is to leave RSS 2.0 with the all the elements in the RSS2.0 namespace, but for publishers to provide 2 different RSS feeds until people get used to it. And then turn off the non-2.0 feeds after a few months. -Tim
First, I agree with Dare Obasanjo — the breakage most likely did occur within aggregators that do support namespaces rather than the reverse; the namespace with RSS 2.0 ‘changed’ and this caused the breakage. However, I disagree with Dare that the solution is to just continue as is and have the RSS generators now create two separate Userland RSS feeds: one for 0.9x and one for 2.0.
How many feeds will we end up with by the time this is done — one for 0.9x, 2.0, and then the RDF/RSS, RSS 1.0 one?
Remember that old chestnut: Poor planning on your part does not make an emergency on mine?
Several things missed with all of this:
- Documentation of the namespace support in RSS 2.0 is non-existent, leaving a great deal of confusion about its implementation
- Most weblogging tools don’t have the capability of just adding yet another RSS feed, and most webloggers (or others who use software that provides RSS) don’t know how to program enough to generate their own RSS feeds (and those that do, don’t care)
- If RSS 2.0 is a major tool release, two weeks to hack it out, implement it, and then shove it into production is a farce — there was no time to allow for third party developers to adapt to the new specification
- Focusing on pure technical solutions to what is the result of poor business practices will only postpone these same problems until the next release of something like RSS
However, what I’m saying is not sexy and isn’t full of code. And since I don’t support RSS, it doesn’t impact on me anyway, so why am I talking about it?
One thing I will say, though, is that if RSS 2.0 had been based on RDF/XML, many of the questions arising now about RSS 2.0 would have been answered by the RDF specification, and there wouldn’t be this chaotic scrambling to understand what all of this means (namespace, default or otherwise). RDF/XML is an implementation architecture, and as such, provides a good understanding of what is, or is not, valid XML within the specification. That’s one thing RDF/XML would have provided.