Don't assume poly always has two children

[jkrumbiegel]

Otherwise the fallback logic doesn't work when poly gets an external override which plots a poly internally again

[asinghvi17]

It looks like this also forwards to the vector draw_poly when it is applicable for the child plot? Just to confirm.

[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(fig)

	using	create	display	create	display
GLMakie	4.23s (4.20, 4.29) 0.03+-	104.93ms (103.35, 107.98) 1.46+-	479.22ms (476.35, 484.07) 2.70+-	8.44ms (8.28, 8.55) 0.11+-	25.64ms (25.45, 26.17) 0.26+-
master	4.23s (4.19, 4.26) 0.03+-	104.57ms (102.71, 105.70) 1.12+-	471.96ms (430.69, 484.40) 18.72+-	8.50ms (8.37, 8.71) 0.12+-	25.68ms (25.47, 26.07) 0.23+-
evaluation	1.00x invariant, -0.0s (-0.14d, 0.80p, 0.03std)	1.00x invariant, 0.37ms (0.28d, 0.61p, 1.29std)	0.98x invariant, 7.26ms (0.54d, 0.35p, 10.71std)	1.01x invariant, -0.07ms (-0.58d, 0.30p, 0.11std)	1.00x invariant, -0.05ms (-0.19d, 0.73p, 0.25std)
CairoMakie	4.03s (3.88, 4.22) 0.11+-	110.39ms (104.33, 118.66) 4.71+-	136.91ms (128.52, 151.09) 9.31+-	8.64ms (8.38, 9.02) 0.26+-	984.68μs (976.10, 1003.06) 9.21+-
master	3.71s (3.63, 3.86) 0.07+-	108.02ms (104.25, 113.77) 3.11+-	135.60ms (127.61, 148.55) 6.61+-	8.36ms (8.23, 8.58) 0.11+-	984.26μs (975.06, 996.85) 8.44+-
evaluation	0.92x slower❌, 0.31s (3.39d, 0.00p, 0.09std)	0.98x invariant, 2.37ms (0.59d, 0.29p, 3.91std)	0.99x invariant, 1.31ms (0.16d, 0.77p, 7.96std)	0.97x slower X, 0.29ms (1.41d, 0.03p, 0.19std)	1.00x invariant, 0.42μs (0.05d, 0.93p, 8.82std)
WGLMakie	4.41s (4.38, 4.44) 0.02+-	104.26ms (102.48, 106.16) 1.17+-	8.88s (8.81, 8.91) 0.03+-	9.64ms (9.02, 10.47) 0.72+-	69.89ms (69.03, 71.39) 0.74+-
master	4.41s (4.39, 4.45) 0.02+-	105.18ms (103.84, 109.94) 2.12+-	8.90s (8.83, 8.98) 0.05+-	10.44ms (9.00, 13.40) 1.46+-	69.91ms (68.53, 70.65) 0.72+-
evaluation	1.00x invariant, -0.0s (-0.09d, 0.87p, 0.02std)	1.01x invariant, -0.92ms (-0.54d, 0.34p, 1.64std)	1.00x invariant, -0.02s (-0.50d, 0.37p, 0.04std)	1.08x noisy🤷‍♀️, -0.8ms (-0.69d, 0.23p, 1.09std)	1.00x invariant, -0.02ms (-0.02d, 0.96p, 0.73std)

[jkrumbiegel]

@asinghvi17 i didn't check but I think it just continues the normal way, so it should work