[Intel-gfx] [PATCH i-g-t] intel-ci: Add fast-feedback-simulation.testlist

[Daniel Vetter]

On Thu, Aug 17, 2017 at 07:53:45AM -0700, Kelvin Gardiner wrote:
> On 17/08/17 00:50, Daniel Vetter wrote:
> > On Wed, Aug 16, 2017 at 05:30:08PM -0700, Kelvin Gardiner wrote:
> > > 
> > > 
> > > On 16/08/17 07:04, Daniel Vetter wrote:
> > > > On Wed, Aug 16, 2017 at 11:33 AM, Petri Latvala <petri.latvala at intel.com> wrote:
> > > > > On Tue, Jun 27, 2017 at 02:04:51PM -0700, Kelvin Gardiner wrote:
> > > > > > Added an initial list of fast feedback tests for simulation
> > > > > > environments.
> > > > > 
> > > > > Merged, thanks.
> > > > 
> > > > Yes I'm a bit late, just noticed this fly by: How does this interact
> > > > wit igt_skip_on_simulation? What's the significance of this list, are
> > > > we going to see CI run these? Have platform owners acked this as the
> > > > PO list?
> > > 
> > > This is a list of tests seen to be good in a simulation environment. It is
> > > meant as a starting point to have a reference list, to which we can add.
> > 
> > seen by whom? That was pretty much my question/concern here. I chatted
> > with Petri, and apparently this list is also used by the helsinki CI, and
> > that should have been noted.
> > 
> > But just today I've seen a mail fly by that e.g. Rodrigo has a power-on
> > testlist. Which I guess is again something else, but doesn't help making
> > things less confusing.
> > 
> > I mean you can add whatever you want to igt and let it rot there, but if
> > you expect platform owners, test engineers and developers to support it,
> > we need a consensus. Otherwise it won't really happen, and this patch here
> > looked like that consensus engineering work wasn't done (and that's really
> > the hard work, not the patch itself).
> > 
> > > With regards to igt_skip_on_simulation, some times this is erroneously
> > > included in a test, other times it is missing, some tests need to reduce
> > > iterations etc, (where this still give a valid test) when this is set (as
> > > some already do). In short some work is needed to clean up the use of this
> > > flag.
> > 
> > So ... who's doing that work? Or are we just going to let 2 half-solutions
> > rot side-by-side?
> > -Daniel
> 
> We have this on our to do list. I was thinking if we create a list of tests
> that need attention the 2 val teams can work through the list to make the
> fixes.

Ok, that was also not clear, and should have been explained in the commit.

A proper commit message doesn't just explain what you're doing (that
should be obvious from the patch itself), but _why_. Plus who's all on
board and what the longer-term plans are.

Rule of thumb for distributed teams like this: If you're not feeling
terrible about how you're overcommunicating, you're not communicating
enough. Please make sure your team knows this BKM when they start
submitting patches.

For this case here, the entire above quoted thread (maybe reworded to have
more structure) should have been captured in the commit message. Or even
better, in some docs somewhere in the code/list.

Thanks, Daniel
> 
> > > 
> > > > 
> > > > VPG validation is just one team here, imo adding something like this
> > > > needs a lot more buy-in. Or we're just once again adding a list no one
> > > > is actually using, which is pointless. Imo if we can't get acks from
> > > > platform owners that they are actively using this list, and CI that
> > > > they are also actively using this list, then it should be removed
> > > > again.
> > > > 
> > > > Thanks, Daniel
> > > > 
> > > > > 
> > > > > 
> > > > > --
> > > > > Petri Latvala
> > > > > _______________________________________________
> > > > > Intel-gfx mailing list
> > > > > Intel-gfx at lists.freedesktop.org
> > > > > https://lists.freedesktop.org/mailman/listinfo/intel-gfx
> > > > 
> > > > 
> > > > 
> > 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch