Add a testsuite called gsk-compare-vulkan to run
the gsk renderer tests with the Vulkan renderer.
The current stats:
Ok: 184
Expected Fail: 0
Fail: 204
Unexpected Pass: 0
Skipped: 2
Timeout: 0
For now, we mark all the tests as failing to
avoid ci breakage. To run the tests locally,
you can do:
meson test -C_build --suite gsk-compare-vulkan
When decomposing curves that are too straight, we may emit lines for
long parts of the curve. These lines do not properly map
t => distance
and it is better to treat them as a regular line than a curve.
This reason argument gives that information.
No users so far, that will happen in followup commits.
This way we can default to the siplest possible foreach() output - like
cairo_copy_path_flat() decomposing everything into lines - and add flags
to get more and more fancy.
This will be useful to have conics automatically decomposed for Cairo
drawing or if we want to add more line types in the future.
This test includes an implementation of a gsk_path_equal() func with
a tolerance that is necessary because parsing does not always work
100% exactly due to floating point rounding, so we can't just
compare the to_string() output.
Write out the commands for rects and circles in a special
way, and add code in the parser to recognize this, so we
can successfully round-trip these through the SVG path format.
The special way - for people who want to use it for debugging -
for now is that we use uppercase "Z" to close standard paths, but
lowercase "z" to close our special paths.
A test is included, but the random path serializations should take care
of it, too.
1. Allow specifying the max number of contours
2. Be smarter about creating the paths:
With 10% chance, create a "weird" path like the empty one or only
points or things like that.
Otherwise create a bunch of contours, with 2/3 a standard contour,
with 1/3 a predetermined one.
The relevant question here is about details, because we have to choose
if we declare alpha-only formats as having their (nonexistant) color
channels premultiplied or not, so that the code paths using them can do
the right thing.
Because we are premultiplied by default, it makes sense to treat alpha
like that, because then the alpha-only code doesn't need to do
workarounds for straight alpha.
Where this is relevant of course is when expanding the alpha channel
into color channels, where we want to end up with white.
So make sure we do color = alpha there instead of color = 1 like we did
before.
We need them for mask-only textures.
For tiffs, we convert the formats to RGBA (the idea that tiff can save
everything needs to be buried I guess) as tiffs can't do alpha-only.
Inverted alpha masks have an effect on the source, even if the mask
doesn't cover the source at all - or worse, is completely clipped out.
The GL renderer handles this fine, but Cairo and Vulkan had
optimizations that got this wrong.
In particular, fix the combination of luminance and alpha. We want to do
mask = luminance * alpha
and for inverted
mask = (1.0 - luminance) * alpha
so add a test that makes sure we do that and then fix the code and
existing tests to conform to it.
