On Thu, Sep 21, 2006 at 09:55:13AM +0200, Francois Visconte wrote:
> Hello,
> I think the real question is : is there allways a backward compatibility
> of system calls accross patching ? ...
> I thinks this is mostly de case....
>
> Jason Dixon wrote:
> >On Sep 20, 2006, at 8:10 PM, Patsy wrote:
> >
> >>On Wed, 20 Sep 2006, John Costello wrote:
> >>
> >>>This is in regards to a 3.9 system that I installed and am patching.
> >>>
> >>>After rebuilding the kernel (patches 007 and 009), is it ,
> >>>unnecessary,
> >>>necessary, advised, or imperative to rebuild userland (FAQ 5.3.5)?
> >>>Thanks,
> >>>
> >>Imperative.
> >>
> >>Your programs might work, but they might not, or they might work
> >>unpredictably. The kernel, userland (and ports for that matter) are all
> >>intended to be kept in sync, not half -stable and half -release, so
> >>if you
> >>have a -stable kernel, you should have a -stable userland as well. i.e.
> >>yes, rebuild your userland.
> >
> >
> >The OP is referring to the patch branch, not -stable. The only time
^^^^^^^^^^^^^
> >rebuilding userland is necessary after a kernel errata is when the
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> >errata claims it is necessary.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
In other words, almost always yes and if no it will be noted.
Joachim
