Taken from Bug 103082: ------- Additional Comments From Brendan Eich 2001-10-04 12:50 -------
hyatt, perhaps we should take this to a newsgroup, but I still think you're on dangerous ground by acting as high back-out sheriff. If everyone did that for a 5% performance regression, I'd be happier. If only you do it, you risk becoming bad-cop, and (what's more important) others get to be victims. Even if you can manage to avoid being capricious (what, you didn't back out the NSS changes that have slowed down startup horrendously?! Why not?!!), you'll do a disservice to others who need to back-out responsibly, too. I still don't see why this had to happen so fast that Gerv or sballard couldn't do it. What am I missing? I like zero tolerance of performance regressions (as opposed to "zero tolerance policies" misapplied in other domains, e.g., weapons bans leading to frisks for nail clippers and expulsions for pointing one's finger). I think the policy needs to be clearly articulated by [EMAIL PROTECTED], and I strongly believe that individuals who (wittingly or not) regress performance need to have a chance to make good, if the regression is not so horrible that we can't live with it for a day. /be
