These followups after the first are likely to be pretty short; not a lot happens in any given 2 week span. All changes this time were invisible.
fOSSiL refactored the audio engine to make adding new audio codecs easier.
He has also begun work on the design of a portable dynamic-loading backend, which is a major precondition towards the ability to have executable add-ons. He's mentioned having PC "video" decoding as a proof-of-concept, but he's also mentioned just going with pluggable audio codecs instead.
I've begun looking at the way the graphics locks are used; one of the crash bugs we found related to paths where code that demanded certain locks be held first were not held appropriately, producing a double-unlock or double-lock condition. These are generally bad news, and it's not particularly immediately obvious which routines have to do this; I've begun documenting the locking requirements of the locks in the code. With luck it will be easy to make the programming model much easier to check later.
The music remixing team has been cackling evilly to themselves lately, but no other information is available at this time. Perhaps more information will surface by the next update.