On Mon, Apr 14, 2008 at 10:29:19PM +0100, Michael C wrote: > Haines Brown wrote: >> Michael, >> >> I wanted to put Debian on a new Thinkpad X61s, and to achieve that with >> minimal pain, I went with sidux. I created a USB-stick to install it, >> and it went as smooth as can be. I'm using the machine with wifi. >> >> All hitches were simply the result of my ignorance. The applications >> I've installed (not many so far because I started with only a base >> system), work fine. I don't bother with any desktop manager, but use >> fluxbox instead. > > Thanks, > > With Fedora as my current desktop I'm used to ongoing minor breakages, > though from the explanation in Martin Krafft's book Sid, and by > implication sidux, is probably too chaotic a proposition for my > (non-hobbyist) needs.
I have very strong opinions on the use of testing versus unstable for non-hobbyist needs. First the purpose of testing is just that, to test the distribution. A testing user should be willing and able to poke and prod the system looking for and reporting bugs. And that user should be willing to live with those bugs for an extended period of time. And be willing to live _without_ particular packages for a while. We saw a lot of this late in the game when etch was in testing. Users would use it for a while, see that it was in pretty good shape but suffer when a particular bug hung around for a while. The crucial bit that many miss is that new packages don't move into testing unless they've sat in unstable with no new bug reports for 10 days (I think). That means that if something breaks in testing, there's a minimum of 10 days of waiting after the initial bug report before the bug fix could even think of migrating into testing. If you throw a couple of other wrenches in the works, you could be looking at an extended period of breakage. So yeah, testing is more stable than sid, but could also bite you pretty badly. Contrast that with sid, bug fixes happen fast. It seems, in my limited experience, that serious bugs that get caught in sid rapidly disappear, sometimes within hours. Sure there's more churn and potentially more opportunities for breakage, but it seems to be pretty short-lived. I've run sid on my desktops for about 4 years now (wow! when did that happen) and I can count on one hand the number of times I've had a serious enough breakage to cause a real problem for my work. And I can count on one finger the number of breakages that required real work to get out of (unbootable system...). I personally wouldn't run a testing system for regular use. I would run sid or stable (with backports as needed). Of course, YMMV. .02 A
signature.asc
Description: Digital signature
