* Adrian Bunk <bunk@kernel.org> wrote:

quoted text
> But that was one year ago when we had only half as many regressions 
> per release as they do now, with 2.6.25 we had a peak of 66 pending 
> regressions...

we have twice as many commits, and we have better test coverage. I get 
the impression that user trust is coming back as well: regressions are 
being reported sooner and more persistently - because we are handling 
them in a more structured and more dependable way. We also seem to have 
more users of latest -git.

so it _appears_ to be an increase in bugginess but i believe it's an 
increase of activity and it's all good IMO, we close 90% of the 
regressions within a week or two, and most of the regressions are for 
obscure cases.

I had 2.6.25 running on most of my boxes from -rc1 on without any 
unprovoked crash. (provoked bugs were another matter) Bisection became 
more practical and more widespread as well. And i periodically find bugs 
that came from ancient kernels so we are fixing bugs faster than we put 
them in i think. I didnt have that feeling in the .18-.19 kernels.

also, now that the kerneloops.org client is in Fedora 9 by default, 
we'll start to have really objective long-term statistics about how our 
users react to the bugs we put into the kernel.

	Ingo
--
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/