[arch-general] ARCH .y kernel releases do not match kernel.org's

[Matthew Monaco]

On 04/24/2011 03:07 AM, Emmanuel Benisty wrote:
> On Sun, Apr 24, 2011 at 1:38 PM, Tobias Powalowski<t.powa at gmx.de>  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.