El Dimecres, 11 de desembre de 2013, a les 21:06:20, Thomas Lübking va escriure: > On Mittwoch, 11. Dezember 2013 20:41:28 CEST, Albert Astals Cid wrote: > > I am hoping that Andras would let us just revert his changes > > since reading the > > commit logs seems like they are only optimizations that ended > > up wrong (i.e. i hope he wasn't really doing any bugfix). > > Depends on how fast Andras can fix this anyway. > > The introduced 50ms "initializeTimer" was originally not set singleShot() > but fixed so with commit 5fedc5ee921f87f320b5465edd80f74cabcfefcf only. > (likely bug #327190) > > Therefore I guess the initial constantly fired timer shadowed the now > present issues what basically means the actual changes have (unintendedly) > never been really tested. > > Patches strongly look like optimization (i guess: "this prevents an ugly > recursion" would have been commented somewhere....) to me as well, so i'd > relax, remove them for 4.12.0, this way get Andras enough time to inspect > the patches and the *actual* behavior and re-add the correct patch for > 4.12.1. > > (Unless Andras already *has* a fixing patch, of course ;-)
Hmmmmmmm, ok, my suggestion, let's do both what Martin and Thomas suggests, let's revert Andras patches and tag/release a RC2 today/tomorrow and delay the 4.12.0 release a bit more than a week and release on the 30th. This way Andras has a week and a bit to re-fix their commits if it is really to have them. Comments? Albert > > Cheers, > Thomas