On 8 August 2012 00:51, Mathieu Trudel-Lapierre <[email protected]> wrote: > On Tue, Aug 7, 2012 at 4:59 PM, Brian van den Broek > <[email protected]> wrote: >> Hi all, >> >> My foray into Fedora has come to a close and I am returning to the >> debian and debian-derived fold. >> >> I'd run stable before and found it too stable. So, I was thinking of >> running testing. Then, to my surprise, I found that the debian FAQ >> recommends unstable over testing >> <http://www.debian.org/doc/manuals/debian-faq/ch-choosing.en.html>. >> The major reason seems to be that problems can persist in testing >> longer than they will in stable. > > There will usually be a waiting period of at least 10 days before > fixes done in unstable are copied over into testing. The waiting time
<snip> > After release, testing will just be the equivalent of unstable 10 days > prior. The side effect is that if you run into a major issue that > wasn't caught while the package was in unstable, you'll be stuck with > it for 10 days. The Debian FAQ bases its unstable over testing recommendation on the fear of a degenerate case: A package breaks in testing, a patch to fix it lands in unstable and, before the 10 days elapse, another patch lands on that same package in unstable, reseting the 10-day clock. Repeat enough times and testing can be broken for long stretches. That this is possible seems a consequence of the migration policy, but I do wonder how often that occurs. > You can work around that by installing packages from unstable directly > by including unstable in your sources.list; and using a technique > called pinning. See 'man apt_preferences' for the details... or ask on > the list, I'll be happy to share examples. I used pinning a bit on ubuntu and crunchbang. It mostly worked. Using losts of pinning on a test box convinced me not to use lots of pinning :-) >> So, I'm a bit confused about which to choose. Advice? > > Choose testing; if only because there's a slightly higher chance that > issues are caught while new packages are still in unstable and not yet > transitioned to testing. If you want something really safe though, > better stick with stable. It's not super old at this point, and there Super-old is in the eye of the beholder, I guess :-) One of the things that made me want to try Fedora was the "early to adopt" value. > will be a new Debian release eventually (like, reasonably soon, though > I can't say when). Yes you can: when it's ready :-) >> As I've already downloaded a stable live CD, my plan was to install >> that, then update the repos to point at testing, and possibly once >> again to point at unstable. So, unless I misunderstand the way things >> work, I can try testing for a while and then go to unstable if >> desired. (I do understand that testing -> unstable is largely a >> one-way journey.) Does that install plan sound sound to those with >> experiencing running other than stable debian? > > Installing from stable and upgrading to testing or unstable should > work. I sometimes used that to install my debian system. You could run > into issues though, since the upgrading process from the current > stable version to what will be the new one hasn't necessarily all been > tested. If it's a new install, it should be fine. If you run stable > for a while before, install a bunch of packages, and upgrade to > testing/unstable then, you could have surprises. OK, thanks. I'll be sure to point to the testing repos before I add anything else. (Would it be better to install stable and immediate point at testing or install stable, update what came in off the live CD I've got, and then point to testing?) This would I guess be better if it were being done in the n-weeks it is going to take wheezy to bake. But, I've got a box with a fresh install of Fedora that does not please, so I've little to lose. Thanks for the advice and the offers of help! Best, Brian vdB _______________________________________________ mlug mailing list [email protected] https://listes.koumbit.net/cgi-bin/mailman/listinfo/mlug-listserv.mlug.ca
