Working Yocto (and other) goals more directly in oe-core, post Yocto 1.1

[Tom Rini]

On 07/15/2011 02:12 PM, Khem Raj wrote:
> On Fri, Jul 15, 2011 at 1:18 PM, Tom Rini <tom_rini at mentor.com> wrote:
>> Hey all,
>>
>> So in the last meeting I put out my beg for folks to try more oe-core
>> alone and less poky or oe-core+meta-oe+meta-distro.
>>
>> Here's what I'm driving at.  oe-core is what everyone uses as their base
>> and is what a lot of important things go into.  But there seems to often
>> be a "lemme start in ${what I use every day} and then I'll move it to
>> oe-core".  And I get that.  Heck, I'm guilty of it too.  But there's
>> problems with that:
>> - Not everyone uses what you do.  This can make testing and trying your
>> prototype out a bit more of a challenge.
> 
> it goes either way as I have seen. You fix per oe-core and it breaks distros
> and you fix using some distro and it might break the sample distro's in OE-core
> Its just the nature of beast. But testing both is what we should aim at.

I would say this is somewhat true.  meta-oe and any other external layer
needs to be keeping in sync with oe-core, but since both are pull models
with different maintainers, there's only so much that can be done there.

That said, there's a +1 for end users using some layering tool goes from
building rev to building rev :)  And I'm open to suggestions on what to
do to make this particular use case easier on everyone that's not poky
(which fixes these problems as they happen by updating their layers at
the same time).

-- 
Tom Rini
Mentor Graphics Corporation