[meta-ti] [PATCH 1/3] SOC_FAMILY: add to MACHINE_OVERRIDES locally

[Denys Dmytriyenko]

On Sat, Mar 03, 2012 at 08:10:20PM -0800, Richard Purdie wrote:
> On Sat, 2012-03-03 at 20:01 +0100, Koen Kooi wrote:
> > Op 3 mrt. 2012, om 14:55 heeft Denys Dmytriyenko het volgende geschreven:
> > 
> > > On Sat, Mar 03, 2012 at 11:53:06AM +0100, Koen Kooi wrote:
> > >> This breaks the hard dependency on Angstrom for getting a sane set of 
> > >> recipes.
> > >> 
> > >> The variable name is kept the same and the actual code moved from 
> > >> angstrom.inc to soc-family.inc to make it clear what it is doing and how.
> > > 
> > > Ah, nice workaround until/if it gets accepted upstream to OE-Core! Thanks, 
> > > Koen.
> > 
> > The consensus at ELC was that these kind of tweaks should be done at
> > the machine level now that we have MACHINE_OVERRIDES seperated out. If
> > a lot of BSPs are going to use SOC_FAMILY we can try to push it into
> > OE-core again. But currently we wouldn't stand a chance.
> 
> MACHINE_OVERRIDES was added to give people a generic way of extending
> overrides from a machine context. This works for various situations such
> as the "x86" override that made more sense than the previous 386|486|
> 586|xxx madness as well as providing a mechanism for SOC_FAMILY and
> other uses.
> 
> I guess the .inc file could make it into OE-Core but I don't see much
> value in adding SOC_FAMILY directly as a standard variable at this
> point.

Thanks, Richard! That's a suitable compromise. I'll send a patch to OE-Core.

-- 
Denys