On Sun, Sep 14, 2003 at 04:22:23PM +0100, Adam Williamson wrote: > On Sun, 2003-09-14 at 12:16, Simon Oosthoek wrote: > > > Maybe something should be done (put on the Wiki) to have a more measurable > > quantity to verify for a release? > > Why? Why should some entirely arbitrarily defined quantities of "bugs" > be a more sensible arbiter of release quality than the judgement of the > developers?
because these developers are personally involved with their particular package(s). And there are no developers who can say "go ahead" of the distro, they can just say that of their own packages. The Q&A department of mandrake should decide whether a release is ready, but I presume they have some quantifiable measure to decide this (even management push is quantifiable ;-). My proposal was just to get these measurements out in the open and have them referred to when this kind of discussion pops up. I also had as a first measure, ZERO critical bugs on a set of known hardware. I believe Mandrake should at least be guaranteed to work on a few well known machines. (Like the ones the Q&A department of mdk has access too). Perfect Quality is an ideal situation, it should be a goal, but not a requirement. On the way towards the goal, it would be good to attempt avoiding regressions and even set stricter requirements as well. Cheers Simon