[yocto-ab] YP Advisory Board: VOTE meta-ti BSP layer YP Compatible

Denys Dmytriyenko denis at denix.org
Tue Mar 6 17:30:59 PST 2018


On Tue, Mar 06, 2018 at 03:34:36PM -0800, Richard Purdie wrote:
> On Mon, 2018-03-05 at 14:37 -0500, Denys Dmytriyenko wrote:
> > On Mon, Mar 05, 2018 at 11:12:42AM -0800, akuster wrote:
> > > 
> > > On 03/05/2018 10:48 AM, Otavio Salvador wrote:
> > > > 
> > > > On Mon, Mar 5, 2018 at 3:41 PM, Denys Dmytriyenko <denis at denix.or
> > > > 
> > > The current explanation in the dev-manual does not make that a
> > > requirement for the Yocto-check-layer to be run. If it is unclear,
> > > then
> > > we need to update doc's or clarify things in the docs.
> > > 
> > >  If the script is mis-behaving, then please open a bug. 
> > Yes, the script is not perfect. I'm being a guinea pig here to use it
> > first :)
> > 
> > For example, qemuarm includes tune-arm926ejs.inc, which sets
> > DEFAULTTUNE to 
> > "armv5te" by default. Other machines, including beaglebone-yocto
> > reference 
> > from meta-yocto-bsp, includes tune-cortexa8.inc, which sets
> > DEFAULTTUNE to 
> > "armv7athf-neon". Then it fails in go-cross-arm from oe-core, since
> > signatures 
> > change due to different DEFAULTTUNE. And that is even w/o meta-ti...
> 
> I'm travelling at the moment and am not able to spend time on this
> right now.
> 
> It is good to see people using the script. You're not the first person
> and Wind River and Intel have used it in a variety of contexts, there
> were posts from Mark Hatle and Patrick Ohly as they did so, they found
> and reported bugs which we did fix. You are still an early user of it
> though and we've by no means found all the bugs.
> 
> The spirit of YP Compat remains quite clear, a package feed generated
> for one MACHINE should not change when you change MACHINE from one in
> meta-ti so some other meta-* even if the tune is the same. That causes
> significant problems for end users and is something we need authors of
> layers to be mindful of. 
> 
> In that sense the changes should be machine specific or distro specific
> and there are changes being made here which cross boundaries. If the
> script doesn't detect those changes, it should and its a bug we need to
> fix.
> 
> I do think its worth working on and will make a big difference to end
> users ultimately, we'll have to lead by example. Thanks to everyone for
> working through this!
> 
> If a bug hasn't been filed, please do so we don't lose track of this.

https://bugzilla.yoctoproject.org/show_bug.cgi?id=12586

-- 
Denys


More information about the yocto-ab mailing list