On Mon, 2010-03-01 at 18:33 +0100, Ralf Corsepius wrote: > > Right now, the only proposal for doing so is to restrict what can be released > > without spending some time in testing. > The issues that at least I have been trying to point out: > > * Is "testing" an adequate safety net? > * Is "karma" an adequate means to "assure quality" > * Is banning a direct pushes an adequate means to improve quality ? > > My answer to all: Neither of them are. They are components in the net. > The solution to actually improve quality are along the lines of > * maintainers to acting more carefully and think twice about what they > are pushing. As Peter said, people *always* make mistakes. No matter how hard they try, no matter how hard they think about it. It is in the nature of people that this will happen. You can't code it out of the wetware by telling people to be really, really careful. To make another analogy, if you're in the lab handling some incredibly noxious chemical substance, you are (if you're a good scientist) very, very, *very* careful. You also are wearing big gloves and safety goggles...because you know no-one is ever perfectly careful. > * rel-eng to implement automated procedures to catch at least the worst > mistakes (e.g. dep breakages, SONAME-breakages). We're working on this, as I've said several times. But why does it preclude other components in the safety net? Automated tests can never catch everything. The major problems that have occurred in updates in the past have not always been things that are amenable to automated testing. Just as people will *always* make mistakes, machines will *never* be able to catch all problems. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Fedora Talk: adamwill AT fedoraproject DOT org http://www.happyassassin.net -- devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/devel