Gaah. Trying to fix a bug in guidance didn’t work out (yet?), so I guess
I’d better call it a day, again.
The past days have been quite OK, last weekend Kim and I started building a new home for the
two chinchilla’s we share a flat with. That should increase the space that they can use
considerably. The biggest gain will be in the height, which more than doubles. They’re climbers, so
that’s a big plus.
Last night, I build a new kernel for my notebook. Took the 2.6.16 vanilla and patches it
with suspend2 to have faster and nicer suspend to disk.
Suspend to RAM is quite useless for this notebook since the battery is more or less dead. It loads
up to 18% which is just enough to change power sockets, but nothing you would trust offline for
longer than 5 minutes. The second patch I applied is Con Kolivas’ -ck patch which has a different
i/o scheduler as the default. This patch improves system responsiveness under (I/O) load
considerably, and given the slow harddisk, that’s more than noticeable. The fbsplash and vesafb-tng
patches add a nice splashscreen to the suspend-to-disk process, for which I’ve created my own
theme.
What buggers me about suspend2 though is that it’s not yet found its way into the
linux kernel. To explain that a bit, suspend2 has been developed out of tree, but is well tested,
documented, has an active maintainer (= one that doesn’t reply “send a patch” by default), supports
encryption and compression (doubling the speed of reading and writing the memory image) of the
suspend image, is much more stable in terms of freezing processes, supports writing the suspend
image to regular files and is much more configurable in general. Yet Pavel Machek refuses to even
consider working together with Nigel Cunningham to get suspend2 merged for the sake of happy users who just want a fast
and reliable suspend-to-disk implementation in Linux. Nigel on the other hand has been helpful for
users for years, has done everything the kernel people asked to make the code mergeable. Pavel has
gotten a lot of criticism lately, not that that changed anything in his attitude. In november he
began realizing that the in-kernel swsusp isn’t all that scalable and that users demand more. So he
began writing swsusp from scratch in userspace(!), adding another one or two years until the
featureset of suspend2 is available to users who don’t want to patch their kernels. That pisses a
lot of people of. It’s time someone kicks that guy swiftly in the butt. I’ve been helping testing
suspend2 releases for about 3 years now, and the contrast between two developers just cannot be
bigger. Lately, Pavel seems to have won the battle of PM maintainers, Nigel announced that he’s given up trying to merge his
patches, but will maintain them as an out-of-tree patch. Sad. Well done, Pavel. Sigh, I’ll stop
ranting at this point, just wanted to raise awareness for that topic though.
Today, I started writing a paper which I plan to submit for FrOSCon. It’s about Guerilla Marketing and the approach we’re
taking in KDE. I’m also writing on an article for the german c’t magazine, which deals with KDE in
general, stuff like dcop, kio, a couple of desktop-wide concepts. In order to get a better idea what
the ‘improved workflow’ in plasma means, I asked Aaron if he could clear that up for me. He swiftly
called me and we had a nice and interesting phone call for about an hour. Thanks for taking the time
to explain that stuff, dude!