RSS is the flavor of the week, clearly
Rewinding the discussion on summaries for a moment, I’d like to point to tha EJB for a moment with regards to his RSS feeds. He doesn’t provide full posts, he doesn’t provide summaries, and he doesn’t provide what is typically referred to as an ‘excerpt’. The amount of text excerpted in his feeds, defined specifically by him, is long enough to attract the reader’s attention… before presenting them with a link and a word count for the full version. In the case of shorter posts, the entire entry is presented. Quite well done, I must say.
I dare say this is why we have ‘excerpt’ fields in MovableType, so we don’t need to rely on machine–generated excerpts. Obviously not a widely–used feature.
Elsewhere, I’m pleased to see some people disagreeing with me (I’m fond of argument, if you hadn’t noticed) over at O’Reilly. An anonymous commenter on Marc Hedlunds’s weblog points out that bandwidth is perhaps as much a concern with full–post RSS as a regular web page, particularly with some news aggregators polling the server every five minutes. There are, of course, solutions to this problem; each useful on their own, or in combination.
- Cut down the number of entries your RSS feed houses. I believe MovableType’s default to be the last 15 posts… which might be a little excessive in the case of full–text feeds. Looking at my RSS 2.0 XML files: the full–text feed is 45 kilobytes and the excerpted–text feed is 14. I think it’s fair to say that I could just house the last five entries (hell, why not just the last three? Why not the last one?) without a single subscriber complaining about it… and I’d be saving the subscribers (and myself) a lot of bandwidth.
- Gzip your feed (already supported by a goodly number of aggregators).
- Ensure your server churns out the 304’s where appropriate (especially relevant to those of us serving dynamically–updated RSS).
- Serve a polite request to those on noncompliant aggregators to bug the developers to support conditional GET.
- Serve a polite, but firm request to IPs whose polling is a little over the top to slow down.
- Stop being such a wuss.
Personally, I like the first three combined with a spoonful of number six. Four and five require a little work, which I might bother to look into later.
Moving through the blogosphere, the big news in RSS this week is XML error handling. Mark, Brent, Dave, and too many others to credit here have weighed in on the situation… and I have to say I’m liking Nick Bradbury’s work on the solution. There are other ways to handle it, of course, but I’ll be interested to see what comes of all this; what other solutions lurk in the hearts of developers everywhere.