3 things you need to know about this change:
1. We use diff(1) in various tests to check generated text against
reference output
2. Windows locates executables not just in $PATH, it also looks in
$cwd and the directory of the current process' binary
3. Multiple tests live together in the same directory
Windows is fun.
On Windows, git defaults to maintaining line endings, which means it
changed \n to \r\n on all files it identifies as text. And that includes
our test output.
Luckily diff(1) has an option to undo that. And since we do not care
about line endings in those tests, we can just use it.
When GL or Vulkan is not supported, the test should not fail.
It would be nicer if we could detect GL/Vulkan not being available
otherwise, but I'm not aware of a better solution, in particular because
rendeers might have stricter requirements than GTK itself.
So this is the next best fix.
No.
This fix is not that much better, but I'm too tired to fix stuff
like this properly.
And the Cairo renderer did at least work everywhere during 4.x
If that flag is set, we keep the bounds of the original node when
rendering the modified node.
Gets around the replay test having to draw a transparent color node to
ensure the same bounds.
The clip might be different from the scissor due to incompatible
intersections.
But the resulting intersection might be fully clipped, so we should
consider it.
Testsuite with longer explanation attached.
Fixes#7044
Due to a Mesa bug, RGBA16 images aren't properly handled sometimes and
can cause random failures.
In this case, generating the modified reference images for the tests
fails.
Fixes CI breakage.
Related: https://gitlab.freedesktop.org/mesa/mesa/-/issues/11750
Opaque textures don't clamp to transparent but instead to black.
We didn't consider this, so we were blurring their edges into blackness
not into transparency.
Fix this by adding the GSK_GPU_AS_IMAGE_SAMPLED_OUT_OF_BOUNDS flag
and respecting it in the implementation that uses it.
Test included.
Fixes#6980
If all four of the random colours have alpha channel exactly 0.0,
then the computed premultiplied average will also be zero.
Normalize the expected colour to (0,0,0,0) rather than (NaN,NaN,NaN,0).
Resolves: https://gitlab.gnome.org/GNOME/gtk/-/issues/6977
Signed-off-by: Simon McVittie <smcv@debian.org>
This requires adding infrastructure to generate per-test data, so that
the random clip rect can be computed and reused for both test and
reference generation.
So add this infrastructure.
... and port the colorflip test.
This is so we can factor out generic parts of the code. This allows
making changes easier to those parts, like if we want to introduce
rules for what colorstates and memory depths to do diffs in.
When comparing textures, always pick the colorstate from the reference
texture. This allows us to define what color state we expect.
For now, there's no check that the color states are equal, because they
don't really have to be as long as the pixels are.
We use the renderer to create the reference for the rotate test by
applying the same rotate transform to the reference image instead of the
tested node.
This is somewhat suboptimal because they run very similar codepaths, but
this method works with high bit depth content and different colorstates
This concludes the port away from gdk-pixbuf and means that all rendered
content and reference images can now use high bit depth and colorstates.
We use the renderer to create the reference for the colorflip test by
applying the same colorflip matrix to the reference image instead of the
tested node.
This is somewhat suboptimal because they run very similar codepaths, but
this method works with high bit depth content and different colorstates
We use the renderer to create the reference for the clip test by
applying the same clip node to the reference image instead of the
tested node.
This is somewhat suboptimal because they run very similar codepaths, but
this method works with high bit depth content and different colorstates
We use the renderer to create the reference for the flip test by
applying the same transform node to the reference image instead of the
tested node.
This is somewhat suboptimal because they run very similar codepaths, but
this method works with high bit depth content and different colorstates
and the gdk-pixbuf method does not.
We use the renderer to create the reference for the repeat test by
applying the same repeat node to the reference image instead of the
tested node.
This is somewhat suboptimal because they run very similar codepaths, but
this method works with high bit depth content and different colorstates
and the gdk-pixbuf method does not.
