Update (7th August 2017): This and other articles are kept for reference, however these days cold pragmatism has replaced the hacker mentality: The overhead of maintaining a custom blog engine - especially given the many breaking changes in Scala over the years - has led me to migrate the blog to a static Blog made with HUGO. As HUGO also re-created the RSS feeds I am afraid that the IDs of the posts have changed again.
I have updated my Blog Engine today. Among some small internal improvements, the biggest change is a new way to generate feed ids. That might mean that you might have to delete the feed if you have subscribed and have to add it again. This depends on the feed reader you use. If your reader behaves funky, it is better if you do that. I am really sorry for any inconvenience. But at least my feed is now 100% Atom 1.0 compliant. For anybody building a software that publishes feeds I can only recommend the W3C feed validator. I wish I had discovered it sooner.