(No, I’m not contemplating suicide. :p)
Work sucks. 
I’m being totally wasted like I predicted. I’m not interested in the subject of the document I’m OCRing, the writing is dreary and not particularly easy to read, the expectations are unfairly rising, the workplace isn’t particularly pleasant (not least due to an annoying noisily flickering fluorescent tube lamp), and the work is tiresome enough that I’ve caught myself about to nod off.
Anyway, so the syslog writeup is now in CVS. Take a look and leave comments. Also, comments on topics to write about are very welcome. (Actually, any comment is very welcome, since I seem to draw so few of them. :p)
I’ve been fighting with 2.6.1 patches, lately, especially trying to edge Nick Piggin’s scheduler fixes into stock 2.6.1. (Stock because Software Suspend 2 is valuable to me.) It isn’t easy, since it seems there have been lots of scheduler changes in the -mm tree.
Oh well, it doesn’t matter that much, anyway. The ALSA in 2.6.1 is a major CPU hog in comparison with OSS in 2.4.24. Whereas playing an mp3 in 2.4.24 takes up about 5-10% CPU time, in 2.6.1 it takes a whopping 25-50%. And I can’t use OSS in 2.6.1, either, since I get oopsen galore. So I probably won’t be using 2.6 on a regular basis until I get beefier hardware capable of running ALSA decently. That and the major problems with the 2.6.1 scheduler (namely lots of jerkiness and unpredictability, and especially the fact that static priorities aren’t respected very much in Con Kolivas’ interactivity work) are what are keeping me from going to 2.6.1 — I’ve dealt with much of the remaining 2.6 braindamage relating to things like the revamped plug-and-play support (due to which ISA PnP devices that aren’t explicitly known to the kernel — with their ISA IDs hard-coded — won’t be recognized, and 2.6 devfs won’t automatically load drivers, and modprobe seems not to respect options specified in /etc/modules.conf, and so on).
As far as I’m concerned, 2.6 is one big regression from 2.4. Sure, you’ve got nifty baubles like ATAPI CD burning (I don’t have a CD writer in my box) and an O(1) kernel-preemptive scheduler with interactivity boosts (I don’t have enough processes to bother about constant-time scheduling, got the preempt patches backported to 2.4 so I have the lion’s share of the responsiveness of 2.6, and I hate the interactivity boosts and silly things like the anticipatory scheduler whose assumptions simply don’t work on my box, etc.). I can do without what’s new in 2.6 — it needs a lot of time to mature.
I think my early guess that early 2.6 wouldn’t be much better than early 2.4 isn’t too far off the mark, judging from LKML traffic. Oh well. I’m sure Andrew Morton will do a decent job of getting 2.6 actually workable after Linus gets a few alpha-quality releases in.
I’m off to bed.
