On Wed, 2013-12-18 at 14:38 +0100, Hartmut Figge wrote: > Regarding regressions, i am compiling SeaMonkey, much more code than > wine, nearly daily. Unless there is a bustage. Currently there is one > since 5 days. > I'm not as bothered if a developer build breaks or fails regression tests, though IMHO changes that do either shouldn't be committed, as I am if a release candidate gets published while its still failing regression tests. > Regressions do occur there and it takes time to fix them. And there is > the issue with using much of the code from Firefox and Thunderbird, > which changes often and has the tendency to damage SeaMonkey. Sigh. > Indeed, but: (1) devs that commit code that breaks a compile or fails regression testing deserve a warning followed, if needed, by having their commit privileges revoked. Yes, I know they may be volunteers, but does any project need devs that make such gross mistakes? (2) publishing a version that fails regression or has incomplete changes testing is a really good way to destroy its reputation. Gnome 3 anybody? > In wine itself i had the luck to meet regressions very seldom. And > regressions *are* fixed. See the release notes of wine-1.6.1 from Alexandre. > I've regularly seen apps that initially ran OK gradually develop runtime faults and eventually stop running altogether. My apps weren't changing at all, but the current Wine version certainly was. That certainly looks like a failing in the regression test system to me: I don't tolerate that in my own code, so why should I put up with it from others? > Maybe there exists a lack of developers and likely they are all > volunteers which cannot be forced. > As I said above, volunteer dev or not, does any project need devs that don't follow the project standards and procedures? Martin
