[OE-core] [oe] Piglit in Poky

Paul Eggleton paul.eggleton at linux.intel.com
Fri Jan 3 13:50:48 UTC 2014


On Friday 03 January 2014 14:37:27 Martin Jansa wrote:
> On Fri, Jan 03, 2014 at 01:26:05PM +0000, Paul Eggleton wrote:
> > On Friday 03 January 2014 13:25:13 Andrei Gherzan wrote:
> > > On Sun, Dec 29, 2013 at 04:44:52PM +0100, Koen Kooi wrote:
> > > > -----BEGIN PGP SIGNED MESSAGE-----
> > > > Hash: SHA1
> > > > 
> > > > Philip Balister schreef op 28-12-13 23:33:
> > > > > On 12/28/2013 10:28 AM, Koen Kooi wrote:
> > > > >> Paul Eggleton schreef op 28-12-13 12:48:
> > > > >>> Hi Koen,
> > > > >>> 
> > > > >>> On Tuesday 24 December 2013 15:22:32 Koen Kooi wrote:
> > > > >>>> Burton, Ross schreef op 23-12-13 19:01:
> > > > >>>>> We'd like to integrate Piglit (an OpenGL test suite) into Poky
> > > > >>>>> so that we can run automated QA on the GL stack.  Piglit is
> > > > >>>>> currently residing in meta-oe, but as Poky is a self-contained
> > > > >>>>> project we can't just add meta-oe to it:  apart from the size of
> > > > >>>>> meta-oe, we can't ensure stability if meta-oe makes incompatible
> > > > >>>>> changes that affect Poky.
> > > > >>>>> 
> > > > >>>>> Piglit isn't a stand-alone package, there are the dependencies
> > > > >>>>> of waffle, python-mako and python-numpy to consider too.  There
> > > > >>>>> are two possibilities I can see:
> > > > >>>>> 
> > > > >>>>> 1) Move piglit and deps to oe-core.  Piglit is for QA purposes
> > > > >>>>> only and pushes the boundaries of "core platform".  In a sense
> > > > >>>>> this is a repeat of the discussion we had with Midori...  does
> > > > >>>>> oe-core contain everything needed to sufficiently exercise the
> > > > >>>>> core components it ships or not?
> > > > >>>>> 
> > > > >>>>> 2) Add piglit and deps to meta-yocto.  Probably a new layer
> > > > >>>>> called meta-yocto-qa (or similar) because the Yocto Compatible
> > > > >>>>> guidelines forbid mixing distribution policy and recipes.
> > > > >>>> 
> > > > >>>> Speaking of layers, can you *please* rename meta-yocto to
> > > > >>>> meta-poky? It's what it's actually is and would remove a lot of
> > > > >>>> confusion when trying to explain that yocto is not a distro, even
> > > > >>>> if the distro layer is called 'meta-yocto'.
> > > > >>> 
> > > > >>> This is a tangent, but a couple of points:
> > > > >>> 
> > > > >>> 1) This rename would not come for free. We'd need to update
> > > > >>> people's
> > > > >>> existing bblayers.conf files on the fly, as we did when
> > > > >>> meta-yocto-bsp was split out of meta-yocto, and thus bump
> > > > >>> LCONF_VERSION; however, doing this only in poky has resulted in
> > > > >>> annoying problems when users remove poky from their configurations
> > > > >>> (since LCONF_VERSION is out-of-step between Poky and OE-Core,
> > > > >>> leading
> > > > >>> to confusing errors in this situation). Thus I think we'd want to
> > > > >>> solve this once and for all by bumping the value in OE-Core as
> > > > >>> well
> > > > >>> as Poky.
> > > > >>> 
> > > > >>> 2) If you propose this rename, perhaps you will also consider
> > > > >>> renaming meta-oe, since that name within a similarly named
> > > > >>> meta-openembedded repository leads to a similar level of
> > > > >>> confusion...?
> > > > >> 
> > > > >> I have no problems with renaming that layer since I get confused by
> > > > >> this a few times a week myself :)
> > > > > 
> > > > > What would we we rename it to?
> > > > 
> > > > I'm very tempted to suggest 'meta-yocto'
> > > 
> > > I definitely find meta-yocto a better option here. It would save me from
> > > some confusion when talking about yocto to other people.
> > 
> > I'm not following; meta-yocto is already called meta-yocto ... ? Maybe you
> > didn't realise Koen was joking...
> 
> My understanding was that Koen was talking about renaming
> meta-openembedded repository to meta-yocto, which would be kind of nice,
> but too late for that now, it would be very confusing with the other
> meta-yocto repository.

I don't think that would be possible anyway. The Yocto Project itself does not 
officially maintain support for the contents of that repository. It's true that 
I'm maintaining meta-webserver and folks at Wind River are maintaining meta-
networking and some other bits and pieces, but AIUI that work isn't officially 
under the Yocto Project umbrella - e.g. we do not run these through our QA / 
testing / release processes like we do Poky / OE-Core.

> > > Related to meta-oe, even if that would be a smaller problem, I think
> > > meta-openembedded is a better name for that layer too.
> > 
> > That doesn't solve the problem I was talking about, namely that there's
> > little distinction between meta-openembedded the repository (that
> > contains a number of layers) and meta-oe which is one of those layers.
> > These are two different things and the similar naming makes it hard to
> > always know which one people are talking about.

FWIW, I didn't yet mention my suggestion - I was thinking about meta-oe (the 
layer) being renamed to "meta-misc" in the absence of some better suggestion. 
I'm sure we could have a transition mechanism via a specialised layer.conf 
where the old name continues to work for a limited time.

> What's even worse is that github mirror names the repositories
> meta-oe/oe-core so even the small distinction "meta-openembedded" =
> repo, "meta-oe" = layer doesn't work there.
> 
> https://github.com/openembedded

Yes, that should definitely be changed as well. They're supposed to be mirrors 
so they should have the exact same names, IMHO.

Cheers,
Paul

-- 

Paul Eggleton
Intel Open Source Technology Centre



More information about the Openembedded-core mailing list