John Doe <jdmls <at> yahoo.com> writes: > From: Carl Eugen Hoyos <cehoyos <at> ag.or.at> > >> On the other hand, the dev branch keep alternating between > >> removing bugs and adding new ones, all the time... > > Could you elaborate? > > We release daily, and I can't remember regressions atm. > > Do you mean that when devs add new features to the code, these are > immediately 100% bug free all the time? Certainly not (actually I think the contrary is true), but bugs in new features are not the point (they actually are an important argument for the way we/FFmpeg develop because it means bugs are actually spotted). > I am sure they are good coders; but even good coders have bugs... > Maybe it will be fixed in a few hours... maybe a few days... maybe more... > Another aspect is the compilation/cmdline options that suddenly change. Again, could you elaborate? There are several command line options that are deprecated since years, and we still don't remove them because we don't want to change cmdline options. > Or one feature that disapear because another replaced it. I would be surprised if there was an example where this wasn't a clear advantage for the user, but please show me I'm wrong! > So, each time you run into a bug, you're being told to checkout the latest > release, and so have to go throught > the whole thing to try to find what still works and what has changed. > At least that's my experience... Could you ... (well you know) > As a comparison, as a sysadmin, I like the feeling of being able to > transparently/safely update my distro, > without having to go through all the packages/services/configurations to hunt > for what could havechanged or gonne wrong. Imagining for a moment it would be an option to have a "stable branch" (it isn't), please note that I am absolutely convinced this would not in the slightest way achieve the goals you imagine. Note that my questions are not to proove you wrong (I am absolutely sure we had many regressions, I just don't remember any atm). What is important, imo, is that we try to improve and that is only possible if you show us our faults (and no, not having something mistakenly called "stable branch" is not a fault). Please fix your line length, Carl Eugen
