/ blog. chaff. occasional witterings.


Feeding The Daemons

For over a year, the home page of has been a collection of content elsewhere. The reason's pretty simple; I don't write long form posts often enough for a standard blog to work as a front page, but I didn't like the static nature of a plain index either. Keeping things updated with links and photos as well seemed like a plan. (Nowadays, you can do this sort of thing within Movable Type itself, but a year ago, you couldn't. Also, I'm rubbish at getting around to upgrades anyway.)

The code that generates the front page is pretty simple, really; it fetches four RSS feeds, does a bit of data munging to group them by day then source, with a bit of caching around the edges. However, despite relying on RSS for the content, there wasn't a syndication feed of the aggregated material. Now there is, and, perhaps sensibly, it uses the same Perl library (XML::Feed) to generate as it uses to consume the feeds elsewhere. (There's actually another post in the technical problems there, if I find time.)

So, if you're following this blog via a feed, why not switch to one of the aggregated ones instead? There'll be more content (as they pull in my more-regularly-updated Vox blog, as well as other stuff), and you might even find it interesting.

There's actually a choice of four feeds; one that shows everything, two that show text (one for long form writing, like this post, only; the other, posts and links combined) and finally one for images, which, right now, is just my Flickr photos, but that might change. If I do add feeds, there'll be a post about how to exclude things. (It's already there in the code.)

Speaking of excluding things, I've started using a feed reader again, a little. This time, I'm not using it for news (I know to check that), or sites that have aggregation (like Twitter,, Flickr, or LiveJournal), but for those one-off blogs you don't check otherwise. Unfortunately, there are a few people I'd subscribe to, but they insist on posting either their links, twitters, or both, to their feed. I can understand why - hell, I bring in links - but I also offer a feed without, and that's what I'd really want from them, too.

It might just be me (after all, I already see things via the sites mentioned above), but I've also started seeing grumblings about this from other people. With the launch of friendfeed and other lifestreaming/aggregation services, for example, people are flagging "duplication, or the infinite echo problem" as an emerging problem. Meanwhile, others are finding that infrequent blogging suits them fine. The comments note that an RSS feed lets you subscribe to someone who posts interesting stuff, but rarely; I'd argue that bringing in links to pad your content might actually drive people to leave again.

Just to make this a constructive criticism, here's how I'd work around this if I was using a blogging engine alone. All my long-form posts would get a category or tag (say "longform") and, as well as an overall RSS feed, I'd advertise a feed for original posts only. The bottom line? Aggregation's good, but let your consumers choose what they want to subscribe to.


links for 2008-03-07 from davblog

trackback 11:26:09

chaff:: feeding the daemons Paul talks about an issue that I've been thinking about recently. When you are generating multiple RSS feeds, how do you deal with the overlap between feeds? Or users who only want to see particular...


Dave Cross

comment 12:28:45

Is the code that generates your front page available anywhere? Or is it a trivial hack that I could recreate in a couple of hours?

I'm beginning to think that Plagger is over-engineered for my purposes, so I'm looking for something a bit more light-weight.


Paul Mison

comment 12:52:33

It's not public, no, but it is fairly trivial to generate an Atom feed by combining things. It's a bit harder to make it valid (I think I referred to this in passing) and I really should write a follow-up to this with some code in; hopefully that'd be useful.


Phil Wilson

comment 20:06:40

Well, this is clearly now a year out-of-date, but in theory, isn't atom:source/id supposed to stop the duplication nonsense?