On 2013-06-17 06:46, Paul Berry wrote:
> On 24 May 2013 19:49, Fabian Bieler <fabianbie...@fastmail.fm 
> <mailto:fabianbie...@fastmail.fm>> wrote:
[...]
>     +/**
>     + * \file ignore-adjacent-vertices-indexed.c
>     + *
>     + * Test that adjacent vertices are ignored when no geometry shader is 
> active.
>     + * Draw the adjacency primitive in red and blend the non adjacency 
> version in
>     + * green on top of it. Then test that the entire framebuffer is either 
> yellow
>     + * or black.
> 
> 
> I really like this yellow/black idea.  Previously we have done this sort of 
> test by drawing images on the left and right side of the window and then 
> using piglit_probe_rect_halves_equal_rgba() to check that they're the same, 
> but that has two problems: (1) sometimes rounding errors in the GPU cause 
> slight variations between the two halves, which leads to false failures.  (2) 
> when there are subtle differences between the left and right halves, it's 
> difficult to spot the differences visually, making debugging hard.  Your 
> yellow/black technique neatly addresses both of these problems.
Thanks, but I have to admit I took the idea from 
tests/general/triangle-rasterization.cpp which was written by James Benton.
> 
> You might want to consider giving check_framebuffer() a more descriptive name 
> and moving it into tests/util/piglit-util-gl-common.c so that it can be used 
> by other tests.
I'll do that in a subsequent, separate test from this series.

_______________________________________________
Piglit mailing list
Piglit@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/piglit

Reply via email to