> There are a whole lot of problems that make it almost impossible: > - Build time and buildd resources Granted - but Debian builds a lot of packages that require a lot of processing power. I can't imagine building xulrunner 2x will be that fatal. > - Upstream support (PGO is not supported yet on linux) https://bugzilla.mozilla.org/show_bug.cgi?id=418866 somewhat true... but there are cases of successful building with PGO. https://bugs.launchpad.net/ubuntu/+source/xulrunner-1.9/+bug/213708/comments/1 links to an arch package that does. > - Profiling requires a xulrunner application, and I think PGO upstream > scripts care about firefox. This means 2 things: By my understanding of https://developer.mozilla.org/en/Building_with_Profile-Guided_Optimization xulrunner itself can be used as the profiling app. That way, Firefox isn't needed at all. > - xulrunner would be profiled for one application, and not others > - xulrunner would need to depend on the application, introducing a > dependency loop, or the packages should be merged. > > Let's add to that that I don't know if PGO builds require network at > build time, but I sure hope it doesn't because that'd be a big NO. The build doesn't seem to use the network, so I don't think this is an issue. > > Not coming any time soon, I'm afraid. Considering the massive improvement in user experience, and the number of users who use firefox/iceweasel, it would be nice if this change could be done. For example, it doesn't look terribly good for Debian (justified or not) that Firefox for Windows under Wine on Debian performs better than native Iceweasel on the same Debian install. > > Mike > Thanks again, ~Craig
-- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
