On Friday 2011-03-04 12:26 +1300, Robert O'Callahan wrote: > On Fri, Mar 4, 2011 at 11:40 AM, L. David Baron <[email protected]> wrote: > > > What sort of plan for turning things off would you expect for > > somewhat architectural performance improvements like these: > > https://bugzilla.mozilla.org/show_bug.cgi?id=636029 > > https://bugzilla.mozilla.org/show_bug.cgi?id=636039 > > These are things for which we have pretty solid test coverage, but > > there is a risk of some small (but easily fixable) regressions, and > > I don't see a realistic way to plan for turning them off other than > > planning to back them out. > > > > Probably backing them out is the plan. > > Some changes can be split up into "huge but relatively safe" patches and > "small but less safe" patches, which is worth doing if it gives us the > ability to only back out the "small but less safe" patches. I don't know if > that's an option for you.
It's already been done that way. Although in the latter bug one of the bigger risks is probably a stray typo in the "huge but relatively safe" patch that somehow isn't picked up by tests. > I think the goal here should be to minimize the expected cost of disabling > stuff--- the cost of disabling multiplied by the probability of having to > disable. Agreed. -David -- L. David Baron http://dbaron.org/ Mozilla Corporation http://www.mozilla.com/ _______________________________________________ dev-tech-layout mailing list [email protected] https://lists.mozilla.org/listinfo/dev-tech-layout
