use premultiplied alpha for cairo argb surfaces

[jkrumbiegel]

Description

Cairo needs premultiplied alpha for image ARGB32 image surfaces, while everywhere else they don't. So we never noticed, which has been leading to all kinds of artifacts with transparent bitmaps. This should fix all of those hopefully.

Before:

CairoMakie.activate!()
x = collect(0:2.0)
y = collect(0:2.0)
z = x .* y'
f = Figure()
ax = Axis(f[1, 1], title="CairoMakie", backgroundcolor = :white)
heatmap!(ax, x, y, z, colormap=(:viridis, 0.5))
f

After:

Type of change

• Bug fix (non-breaking change which fixes an issue)

[jkrumbiegel]

Should fix #2162 as well but I can't select that in the link menu for some reason

[MakieBot]

Compile Times benchmark

Note, that these numbers may fluctuate on the CI servers, so take them with a grain of salt. All benchmark results are based on the mean time and negative percent mean faster than the base branch. Note, that GLMakie + WGLMakie run on an emulated GPU, so the runtime benchmark is much slower. Results are from running:

using_time = @ctime using Backend
# Compile time
create_time = @ctime fig = scatter(1:4; color=1:4, colormap=:turbo, markersize=20, visible=true)
display_time = @ctime Makie.colorbuffer(display(fig))
# Runtime
create_time = @benchmark fig = scatter(1:4; color=1:4, colormap=:turbo, markersize=20, visible=true)
display_time = @benchmark Makie.colorbuffer(display(fig))

	using	create	display	create	display
GLMakie	21.40s (20.68, 22.08) 0.47+-	22.78s (21.83, 23.42) 0.59+-	21.20s (20.45, 22.26) 0.61+-	21.17ms (20.76, 21.92) 0.39+-	117.68ms (114.57, 121.11) 2.13+-
master	20.74s (19.58, 21.81) 0.71+-	22.78s (21.75, 23.81) 0.73+-	21.18s (20.61, 22.18) 0.53+-	21.14ms (20.47, 21.79) 0.56+-	116.21ms (114.63, 118.11) 1.57+-
evaluation	+3.09%, 0.66s invariant (1.11d, 0.06p, 0.59std)	-0.03%, -0.01s invariant (-0.01d, 0.98p, 0.66std)	+0.06%, 0.01s invariant (0.02d, 0.97p, 0.57std)	+0.13%, 0.03ms invariant (0.06d, 0.92p, 0.47std)	+1.25%, 1.47ms invariant (0.79d, 0.17p, 1.85std)
CairoMakie	14.75s (13.32, 16.95) 1.25+-	19.96s (17.60, 23.18) 1.95+-	3.30s (2.90, 3.80) 0.32+-	17.91ms (16.25, 20.04) 1.36+-	26.90ms (25.29, 28.97) 1.61+-
master	14.58s (13.28, 15.68) 0.92+-	20.54s (18.42, 22.98) 1.39+-	3.32s (3.07, 3.73) 0.22+-	18.57ms (16.21, 20.49) 1.80+-	28.15ms (25.39, 30.33) 1.83+-
evaluation	+1.10%, 0.16s invariant (0.15d, 0.79p, 1.08std)	-2.89%, -0.58s invariant (-0.34d, 0.54p, 1.67std)	-0.53%, -0.02s invariant (-0.06d, 0.91p, 0.27std)	-3.68%, -0.66ms invariant (-0.41d, 0.46p, 1.58std)	-4.63%, -1.25ms invariant (-0.72d, 0.20p, 1.72std)
WGLMakie	25.15s (24.42, 26.48) 0.78+-	36.52s (34.89, 37.85) 1.00+-	60.32s (58.52, 61.77) 1.18+-	34.63ms (33.30, 35.64) 0.79+-	2.14s (2.07, 2.24) 0.07+-
master	25.13s (24.08, 26.72) 1.01+-	36.21s (34.22, 39.37) 2.02+-	58.50s (56.08, 62.57) 2.39+-	34.75ms (33.63, 36.07) 0.84+-	2.15s (2.04, 2.21) 0.06+-
evaluation	+0.08%, 0.02s invariant (0.02d, 0.97p, 0.90std)	+0.85%, 0.31s invariant (0.19d, 0.72p, 1.51std)	+3.02%, 1.82s invariant (0.97d, 0.10p, 1.78std)	-0.34%, -0.12ms invariant (-0.15d, 0.79p, 0.81std)	-0.33%, -0.01s invariant (-0.11d, 0.84p, 0.06std)

[SimonDanisch]

Yay finally :)

Before:

With this PR: