<fdobridge>
<fooishbar> usually if a Piglit update causes 239874 new tests to fail, the automation captures that in new expectation files, but a human stops to pause and wonder if it's sensible
<fdobridge>
<gfxstrand> Well, it should only cause two new tests to fail (or maybe flake)
<fdobridge>
<gfxstrand> But enough that CI might get a little less stable if we don't catch it
<fdobridge>
<gfxstrand> Because, at least on Intel, it flakes since it's all race conditions.
<fdobridge>
<zmike.> dang I'm not sure the technology exists to handle that