The scheduler has been rewritten by Matthew Dillon - again! Except this time, it's very close to the original CSRG implementation. The eventual goal is to allow other schedulers to be used, on the fly.
OSNews has an interview about what's planned for FreeBSD. Not much bearing on DragonFly, other than it's interesting to see where the design goals match and diverge.
Matthew Dillon hints that using polling may be a way to get a finicky PCMCIA network card to work.
Hiten Pandya describes 'makewhatis -o local-manpages.txt' as a quick trick to make a reference list of available utilities.
Simon 'corecode' Schubert is looking for people willing to test his DRM/DRI changes; if you've got a 3D card (ATI Radeon, Matrox, etc.), contact him for instructions on testing.
Jeremy C. Reed, of fame, is the newest DragonFly committer, and has been tackling the dangerous task of documentation!
Chris Pressey announced a new version of the most excellent BSD Installer. Among other changes, it is rewritten in Lua, allows kernel module loading, gettext support, and ... go read the announcement yourself.
Simon 'corecode' Schubert posted these steps for trying out gcc 4.0, which is in DragonFly but not yet part of the build process.
Todd Willey posted on that pkgsrc is frozen for 2005Q2 release - try building packages from it, and report problems now to get fixes into the release.
The street where is hosted (my house) has been having electrical outages over the course of today; they're lasting longer than my UPS can handle, so access may be intermittent if the power goes out again.
Chris Coleman has, with free binary security upgrades for FreeBSD 5.3. There's a good description on BSDNews about the service, though it doesn't cover how it works in-depth. "BSDUpdates" suggests that this may be extended to the other BSDs at some point; it'd sure be handy for DragonFly once the development pace slows.
It's a little late to mention this, because it's either complete or close to complete, but Matthew Dillon and others have been removing SPL sections in code (there were more than 500 locations) and replacing them with critical sections. This is an important step on the road to Giant Lock-free SMP.