On 04/24/2011 03:07 AM, Emmanuel Benisty wrote:

On Sun, Apr 24, 2011 at 1:38 PM, Tobias Powalowski wrote:

...

Am Sonntag 24 April 2011 schrieb Emmanuel Benisty:

...

Hi,

I was having some sound issue with ARCH 2.6.38.3 stock kernel so I started to bisect it from Greg KH's 2.6.38.y stable tree. .38.2 was good and .38.3 was bad (so I thought) but I hadn't any single bad commit during bisecting. However, .38.4 /was/ bad. I could finally find the guilty commit (which is in .38.4) but couldn't understand why I was hit by this issue even with 2.6.38.3-ARCH. Then I diff'd both .38.3 patches and found out that Arch's one includes patches that are not in Greg's release. It seems we include patches that are still in -stable patch queue. Finally, I just have one question: is that normal? All I can say is that it made my bisecting session a real PITA. Please give me back my CPU cycles :P

Cheers. -- Emmanuel The .3 contained some prepatches from the stable queue. That is the explanation for it.

Thanks Tobias but I already learned that the hard way :P

Should we really do that? Or in that case, shouldn't the package be given another version? That is really confusing.

Furthermore, those patches are still being tested in a way. Here's what Greg KH says in the announcement: "If anyone has any issues with these being applied, please let us know."

That is before the .y release...

I don't think including additional patches is a big deal. But I've seen discussion lately on how to include them. I think separating the ARCH patches from the .y patch would be nice. And then additional patches should be added individually.