I know everybody says this, but man, there are a metric boatload of pretty badly formed RSS feeds out there. I have most of an application written that trawls a blogrolling.com blogroll for RSS feeds and generates a “friends” page from whatever it finds. I even respect “Last-Modified” headers. Unfortunately, everyone kicks out bad RSS. Some people also have pretty flaky RSS autodiscovery bumpf — Atrios, for example, winds up pointing my l’il robot to an RSS feed containing the last few comments on his blog. Weird. I could work around that, though, although it would kind of kill the glorious automation purity I’ve got now.
What really gives me hives is the malformed feeds, though. Yeesh. Maybe I can steal some code from Amphetadesk, or maybe Mark Pilgrim has some useful stuff. Oooh, or wait, mt-rssfeed has a liberal parser module which does not require me to run Python. Hm.