I’ve just finished a massive overhaul of my news aggregator page. It now orders items according to date rather than by source, so the top items are always the newest ones.

I’ve added colour-coding (by country for news, and by genre for the rest). I’m pleased with the newspaper graphic at the top, the result of half an hour in Photoshop. The hardest part was finding a suitably generic page in the newspaper—one that wasn’t covered with gaudy advertisements or too many pictures.

It’s heartening to see that in the interval since I developed my original RSS-reader, everyone is now doing character encoding properly. That’s enabled me to use—I shy away from the buzzword “leverage”—the iconv library instead of my hand-rolled heuristic character encoding translator. Proud as I was of that bit of hackery, I’m far happier to use a standard library where possible.

The new aggregator has two parts. The “fetcher” is written in Perl and based on much of the previous parser code. Instead of producing the static page straight away, it adds new items it discovers into a database table. The database is set to disallow duplicate items, which makes the processing fairly simple.

The second part is a simple piece of PHP code that just dumps the latest items from the database in descending date order.