On Mon, Oct 4, 2010 at 7:21 PM, hermann pitton <hermann-pitton@xxxxxxxx> wrote: > thanks for the report and pointing to the details again. > > We can see, that my testings on four different machines and Dmitri's > tests have not been enough. Mauro had the Dual card=78 version from me > too at least for analog TV testing. > > And, that was on hg with most backward compat as possible. > > How good are our chances, to run in such and similar troubles in the > future, in fact staying only on latest -rc, rc-git and in best case on > -next stuff previously? > > It will all come down to the distros and such a bug fix might take just > a year in the future regularly ... > > So, if the quality control was not even sufficient on hg, what will > happen on latest -rc git stuff for that? > > Obviously zillions of people do much more prefer to crash around there > than on hg ... ;) I think it's been made pretty clear: we don't give a crap about whether users' PCs crash. Getting the code into the bleeding edge kernel is the most important thing. Reducing maintainership overhead is clearly more important than whether the code actually works. Forget about the hg backport system. We would rather get crap code into the bleeding edge kernel where almost zero users will test it than to put it into HG where there is actually a chance for users to see the problems before it goes into the mainline kernel (except for the 0.1% of users who are willing to install the latest bleeding edge kernel and make it work with all their other hardware). Yes, we should all be prepared for lots of regressions being introduced, and nobody notices them until the code is already in the distros and has reached the masses. And then maybe if the users are lucky the distro maintainers will backport fixes. It's been made pretty clear that reducing merge overhead is more important than delivering a quality product. I'll stop hijacking the thread now. Devin -- Devin J. Heitmueller - Kernel Labs http://www.kernellabs.com -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
