[igt-dev] [PATCH i-g-t] lib: Use unsigned gen for forward compatible tests

[Zbigniew Kempczyński]

On Mon, Aug 10, 2020 at 09:56:34AM +0100, Chris Wilson wrote:
> Quoting Petri Latvala (2020-08-10 09:22:42)
> > On Mon, Aug 10, 2020 at 10:09:46AM +0200, Zbigniew Kempczyński wrote:
> > > On Thu, Aug 06, 2020 at 03:45:29PM +0100, Chris Wilson wrote:
> > > > Unknown, so future, gen are marked as -1 which we want to treat as -1u
> > > > so that always pass >= gen checks.
> > > 
> > > Do we really want to enable the tests when platform is not fully
> > > enabled in IGT?
> > 
> > What does "fully enabled" mean?
> > 
> > If the test is checking for just "gen > x", the test should work
> > already. If the test is also checking for "gen < y" then we get a
> > spurious failure, but either way CI is going to tell you that
> > something is not passing. Without this it will be a skip, along with
> > skipping in the case that should just work already without actual test
> > changes.
> 
> And for a very large fraction of tests, the behaviour of next gen
> follows current gen. Those that do not are very interesting, and will be
> a much smaller number than the volume of skips we have to process.
> -Chris

So patch only defers in time adding explicit platform definition.
For each tests which requires exact gen test will fail anyway, but
other will likely work. If they won't we'll know it from CI.

I see no big risk but let Petri decide, from my side:

Acked-by: Zbigniew Kempczyński <zbigniew.kempczynski at intel.com>