Please, won't somebody think of the children!? (Upgrade Carefully)
Ok, here's a little tip -- turn off your RSS feed before upgrading your blog software. Really. Please. For the sake of all things good and holy. It just happens to be Anders' feed today, but it seems like about every 2-3 days somebody decideds to upgrade their blog software, thus flooding the planet with everything that they've ever written.
I fail to see the solution
Disclaimer: I'm not part of the planet, but I am curious.
When and how do you turn it back on without merely having a delayed flood?
Planet uses timestamps, so if the posts are a couple of days old, ideally they aren't shown. The problem is that usually updating blog software also resets the RSS timestamp.
Call me stupid, but how difficult would it be to detect someone´s RSS contains 10 headers with the same date?? ;-) That shouldn´t be too hard to implement and keeps the planet clean.
Problem is not upgrading
The bad news is that upgrading is not the only trigger for planet puking out all earlier stories. It appears that there's something fundamentally wrong with the planet software used (planetplanet, IIRC), merely removing spam comments from a blog can trigger the behaviour you're talking aobut in planet.
Stephan Hermann (\sh) told me some time ago that Serendipity is also able to act as web aggregator, and that it's less buggy than planet. Maybe this is a suitable alternative option for planet.kde.org?
Yep we use s9y at amarok.kde.org and have never had a problem. Granted, we have far fewer users then planetkde.org. Someone could setup a test s9y planetkde pretty easily I suppose.
The problem is ultimately RSS for not having a clear way to uniquely identify entries as part of its core standard.
Planet Is Greedy
I read that planet generates a signature based on various properties for each post and if this signature stops matching it reposts.
For instance I changed my blog settings so that the links it creates started with www.methylblue.com rather than methylblue.com (no www), that was all that was required to make planet regurgitate my whole posting history. Much to my embarassment.
Basically, planet needs some intelligence, why does it ever post stuff that was created weeks before the bottom-most post it is currently displaying? Crazy.
It ignores most dates
Planet doesn't even check all the dates, it seems to only use the date of the first blog entry, in case it fails (like Scott outlines) and uses it for all older ones. That might explain why the sorting order is completely screwed.
Clee is the one in charge of planetkde.org, we might ask him if he has some time to try out s9y ...