On Wed, Jul 11, 2012 at 12:11 PM, Martin Jansa <martin.jansa@gmail.com>wrote:

> On Tue, Jul 10, 2012 at 11:43:58PM +0100, Phil Blundell wrote:
> > On Wed, 2012-07-11 at 01:15 +0300, Andrei Gherzan wrote:
> > > And to move a little further, busybox should be configured with
> > > CONFIG_HWCLOCK=y only if it makes sense for that MACHINE. In my
> > > opinion this is machine specific.
> >
> > That's a DISTRO decision.  I suspect most that have binary feeds would
> > prefer to have a single busybox binary per architecture, and accept the
> > few wasted bytes on machines without RTC, than to make it MACHINE
> > specific and end up building it multiple times.
>
> Agreed.
>
> It would be better to package initscript + hwclock to separate packages
> and then pull it to image only for machines with RTC.. but not making
> whole busybox machine specific.
>
> Another advantage of this would be option to use busybox-hwclock as utility
> and hwclock-systemd instead of busybox-hwclock-init for images which are
> using systemd.
>
>
Thank you all. It makes sense to keep busybox machine independent. I agree.
Now if i package this init script separately how would it be pulled in by
images? Think this could brake some image which now relay on this being
pulled in by busybox.

ag