On 28/10/2007, Claude Jones <cjones@xxxxxxxxxxxxxx> wrote: > On Sun October 28 2007, Michael Schwendt wrote: > > > > Hyperbole. > > > > > > that's a high horse you ride > > > > Why? In the context of this thread, the recommendation not to use "yum > > -y" is beyond my comprehension. > > No it's not. You're on your mount. Your "my way or the highway" approach is > prototypical of what's been going on on this list for several months, now. > > Congratulations back to you for your unpleasant pedantry. Cast your fireball spells as much as you like. It is way too easy to point a finger at a Fedora user after an accident and blame him for installing a bad update or for not verifying whether an update set was safe to install. "You could have avoided the disastrous results if you had entered NO instead of YES". "You could have tried out the Test Update, which nobody has had interest in for two weeks." Great! That is the same stance as of those people who preach of backups, because the next distribution upgrade might do the completely unexpected and erase the /home partition's contents, although the installer was never told to do that. Fedora 7 is supposed to be a stable branch of the distribution. Fedora's precious users ought to be protected from disastrous results and should be given reason to trust *all* updates, especially the security related ones. Unfortunate regression left aside, I don't like it when some packagers play with fire and pipe out upgrades for Fedora 6 and 7 that put it on par with the Fedora 8 development branch just to offer new package builds. We have a series of test releases, even freezes, but shortly after a gold release of the distribution, we literally throw away all that and upgrade hundreds of packages to give the users something fresh and less tested to play with? Stuff that puts the user into a loop, in which to evaluate lots of updates every week, trying to gain experience with cherry-picking instead of simply applying or ignoring all updates? Version upgrades, changes in .rpmsave/.rpmnew files, regression - for some packages we've had almost daily updates from upstream source code repositories. I know testing is not easy. When upstream developers miss a problem, package maintainers and co-maintainers don't spot it either, the built update is made a Test Update without anyone reporting a problem [or without anyone testing it], the single (y/n) confirmation prompt in Yum is no silver bullet either. For instance, it won't help at all if an RPM package post-install scriptlet runs wild on your file-system. So, suggesting users to avoid yum's -y as the way to handle Fedora Updates, is only going to scare users.
