Every few months for the last 2 years I’ve gone on the hunt for an RSS aggregator. Every time I’ve been disappointed. Everything I’ve tried has been clumsy and kludgy and made my blog reading harder, rather than easier. And so I’ve stuck with the routine of actively visiting all the blogs I read (of course my fancy “last updated” blogroll scanner that I wrote myself in Perl helps a lot with not wasting time checking pages I know haven’t changed!)

Until now. Last week I came across SharpReader, loved it instantly, and one week on I’m still a fan.

The interface is the standard 3-pane Oulook-esque approach that most aggregators seem to take now, but it seems to do it better than most. Subscribing to feeds is also much easier than the clumsy approach many others take – you just drag a webpage from your web browser and it does the auto-discovery dance for you. (It also imports OPML so it was easy to get my initial sites set up).

You can categorise all the blogs into, erm, categories, and perform tasks on a universal, category-wide, or single blog basis (so, for example, I poll our internal work blogs at 15 minute intervals, blogs of poeple I know well every 30 minutes, keep most at the default hourly polling, and have some daily.

It also has feedster searching built in, and because feedster provides RSS search results, it’s trivial to subscribe to searches for topics that interest you. (This is also a great way to find new blogs to read regularly).

I have two major wishes for it, which I sent in email to the author (and received a response within an hour, saying that he liked both ideas and would try to get them in either the release that’s about to come out, or the next one!):

Firstly, I’d like the polling to be a bit more staggered. I have a lot of feeds, and every hour I get pop-up notifications telling me of all the changes – which can drag my computer significantly if there’s a lot of changes have happened in the last hour. I suggested that instead of waiting 60 minutes for the next poll it should wait for 60 minutes + rand(3). Over time the feeds will all start to drift apart, and be distributed all over the hour. I’ll get a constant stream of new articles all the time rather than a surge every hour.

Secondly, I’d really like to be able to mark articles to come back to later. At the minute I have to keep marking things as “unread” if it’s something I want to read in more depth.

But these are minor nits in a fantastic product, and now I can subscribe to lots of feeds that were just too much hassle to go check manually on a regular basis (such as the CPAN reviews of my various Perl modules!).