Scripsit Yann Dirson <[EMAIL PROTECTED]> > But that last point raises another issue: does anyone really use > testing ? Would anyone use pre-testing after all ?
I think very many people use stable plus bits and pieces from testing. I have two machines set up that way. Getting the bits and pieces from testing instead of unstable gives some amount of protection from getting something that is horribly broken enough to hose the entire system. For a non-mission-critical desktop system behind a firewall, I prefer to run testing, upgrading infrequently. I care relatively little about the bleeding edge as such, but little bugfixes in otherwise mature packages propagate into testing much quicker than they reach stable. Again, not blindly tracking unstable gives me a reasonable potection against bad packages breaking my system completely. In fact my desktop system runs unstable, not testing. But I do not perceive any particular benefit in that. On the contrary it makes me dread seeing "Unpacking new libc6... oops, kernel panic" each time I update (not that it has actually happened since I converted, mind you). I run unstable solely by altruism; after all *somebody* has to run unstable in order to detect such snafus before they propagate to testing. But if somehow I completely lost my respect for Debian's social process (say, if my NM application gets rejected with rude words and insults and somebody adds a script to the BTS that automatically tags all my wishlist bugs wontfix and closes them, and the consensus on debian-devel is that both are completely reasonable actions), then I would waste no time in selfishly going back to testing instead of unstable. -- Henning Makholm "I didn't even know you *could* kill chocolate ice-cream!"
