[OE-core] [PATCH 1/2] piglit: move to meta-oe

[Richard Purdie]

On Wed, 2020-03-18 at 14:49 +0100, Alexander Kanavin wrote:
> On Sat, 29 Feb 2020 at 14:42, Alexander Kanavin <
> alex.kanavin at gmail.com> wrote:
> > > > I'm not sure about this. The intent was to use piglit as a way
> > of
> > > > > improving our graphics testing, particularly allowing it to
> > > be
> > > > > automated.
> > > > > 
> > > > > Whilst we've had to focus on getting the basics right, I'm
> > > not sure
> > > > > that objective isn't still a worthy goal over time?
> > > > 
> > > > Piglit is meant for testing and validating OpenGL drivers for
> > > real
> > > > hardware. While we can put it on top of software Mesa driver or
> > > > virgl, I am not sure there is much value in that? Software
> > > rendering
> > > > might even be too slow to run in practice.
> > > 
> > > OE-Core is meant to be usable for validating BSPs amongst other
> > > things
> > > though. Testing virgl does prove the graphics stack is working
> > > too from
> > > a software perspective so does have some value.
> > 
> > Right, I can put Piglit on top of virgl and see how well that
> > works. It would be good to enable virgl ‘out of the box’ at some
> > point, I’ve been proposing that for a while :)
> > 
> 
> I found an old ticket where piglit was proposed for the core image
> testing, and the idea was rejected due to its huge install size, so I
> won't be pursuing this further.
> https://bugzilla.yoctoproject.org/show_bug.cgi?id=10047
> 
> Would you still say it's worth to keep the recipe in core? Python-
> numpy (that piglit and only piglit needs) isn't trivial to maintain
> either (for instance, update to latest version needs cython to be
> added to core).

That bug is for piglit as a default in testapps. We decided that was a
bad idea, not that testing with piglit was a bad idea overall.

Cheers,

Richard