Re: Linux kernel version for wheezy

[Moritz Mühlenhoff <jmm@inutil.org>]

Ben Hutchings <ben@decadent.org.uk> schrieb:
>
> --=-H5TOWby5lstdx8paRFAT
> Content-Type: text/plain; charset="UTF-8"
> Content-Transfer-Encoding: quoted-printable
>
> Given a freeze in June 2012, we will have a choice between these Linux
> releases (with estimated dates):
>
> 3.2 (December 2011)
> 3.3 (March 2012)
> 3.4 (May/June 2012)
>
> Some other distributions with long-term support will be using:
>
> Oracle Linux 6:   2.6.32+ (RHEL-compatible), 2.6.39 (UEK)
> RHEL 6:           2.6.32+
> SLE11 SP2:        3.0
> Ubuntu 12.04 LTS: 3.2
>
> Mandriva and Slackware are supposed to have long-term support; should we
> try to find out what they're doing?
>
> Greg K-H will be supporting 3.0 and I believe there will also be a
> 3.0-rt longterm series.  However, this was far too early a release to
> use in wheezy.
>
> If we go with 3.2 then we can work with Ubuntu and someone from either
> kernel team can run a longterm update series (Greg is happy for people
> do this using much of the same kernel.org infrastructure).  But it's
> still going to be a year or so old by the time we release (no worse than
> for squeeze, but no better).
>
> If we go with 3.3 or 3.4 then we can release with more features and have
> less backporting work to do.  But we are also likely to get far less
> help with bug fixing (including regression testing).  It is possible
> that SUSE could use 3.4 for the next SLE service pack, but I doubt we
> would know that until some time after our freeze date.

I'd prefer going with 3.4. For Wheezy we have don't have to care about
external patch sets since Xen is merged and openvz/vserver have been 
dropped. 

I think the massive amount of bugfixes/features from 3.3 and 3.4 outweigh
the benefits of a common stable tree. If there were a major player basing
on 3.3 things would be different, though.

Maybe we should contact Greg or someone else from SuSE for a potential
3.4 in SLE11 SP3?

Cheers,
        Moritz