Delaunay recipes

[SimonDanisch]

Continuation of #3102

[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	14.18s (13.05, 15.14) 0.69+-	1.34s (1.28, 1.45) 0.07+-	942.30ms (908.66, 971.39) 22.98+-	16.42ms (15.84, 16.95) 0.36+-	191.63ms (188.08, 196.77) 2.98+-
master	13.58s (13.07, 14.15) 0.38+-	1.32s (1.24, 1.43) 0.08+-	954.34ms (894.85, 988.36) 32.20+-	16.57ms (16.03, 17.01) 0.41+-	194.30ms (187.98, 197.15) 3.07+-
evaluation	+4.22%, 0.6s invariant (1.07d, 0.07p, 0.53std)	+1.53%, 0.02s invariant (0.28d, 0.61p, 0.07std)	-1.28%, -12.04ms invariant (-0.43d, 0.44p, 27.59std)	-0.91%, -0.15ms invariant (-0.38d, 0.49p, 0.39std)	-1.39%, -2.67ms invariant (-0.88d, 0.12p, 3.03std)
CairoMakie	11.70s (11.50, 12.12) 0.20+-	1.28s (1.25, 1.31) 0.02+-	258.59ms (242.18, 271.21) 9.08+-	15.38ms (15.24, 15.58) 0.13+-	8.49ms (8.43, 8.63) 0.07+-
master	11.62s (11.47, 11.76) 0.10+-	1.24s (1.20, 1.28) 0.03+-	304.70ms (292.17, 317.87) 9.70+-	15.34ms (15.11, 15.55) 0.13+-	8.50ms (8.36, 8.63) 0.09+-
evaluation	+0.69%, 0.08s invariant (0.51d, 0.37p, 0.15std)	+2.44%, 0.03s slower X (1.34d, 0.03p, 0.02std)	-17.83%, -46.11ms faster✅ (-4.91d, 0.00p, 9.39std)	+0.21%, 0.03ms invariant (0.24d, 0.66p, 0.13std)	-0.07%, -0.01ms invariant (-0.07d, 0.89p, 0.08std)
WGLMakie	16.44s (16.12, 16.88) 0.29+-	1.58s (1.53, 1.64) 0.03+-	15.02s (14.76, 15.42) 0.27+-	19.17ms (18.26, 20.74) 0.84+-	1.47s (1.36, 1.56) 0.07+-
master	16.56s (16.34, 16.77) 0.17+-	1.59s (1.48, 1.69) 0.07+-	15.05s (14.91, 15.25) 0.14+-	18.87ms (17.26, 19.75) 0.89+-	1.48s (1.44, 1.52) 0.03+-
evaluation	-0.75%, -0.12s invariant (-0.52d, 0.35p, 0.23std)	-0.58%, -0.01s invariant (-0.17d, 0.76p, 0.05std)	-0.24%, -0.04s invariant (-0.17d, 0.76p, 0.20std)	+1.57%, 0.3ms invariant (0.35d, 0.53p, 0.86std)	-1.03%, -0.02s invariant (-0.27d, 0.62p, 0.05std)

[SimonDanisch]

Can someone check:

https://docs.makie.org/previews/PR3159/examples/plotting_functions/voronoiplot/
https://docs.makie.org/previews/PR3159/examples/plotting_functions/triplot/
And best check:

# from the project where you developped Makie
using ReferenceUpdater; ReferenceUpdater.serve_update_page(pr=3102)

I'm not 100% sure what the right output is supposed to look!
If this all looks good, I think we can merge, right?

[DanielVandH]

Can't edit but I think those changes I've just reviewed will make the docs look better.

[ffreyer]

I simplified a few docs examples and updated the bounding_box docstring. Regarding bounding_box I'm wondering if

• we should add Rect2 -> tuple conversion for triplot to make it more Makie-like
• we should rename bounding_box to clip for voronoiplot since we have the Circle clipping in there too

Also, is this correct with the pink line cutting through? (From docs)

[DanielVandH]

we should add Rect2 -> tuple conversion for triplot to make it more Makie-like

This would be a good idea.

we should rename bounding_box to clip for voronoiplot since we have the Circle clipping in there too

I think that's okay, go for it.

[github-actions]

Missing reference images

Found 8 new images without existing references.
Upload new reference images before merging this PR.

[SimonDanisch]

So... Is this ready now? ^^ I've been updating the refimages quite often, so if this isn't ready yet, I won't update them right away ;)

[DanielVandH]

As far as I can tell, everything is good. Those failed tests worked fine for me locally - did something go wrong with them in the test, or were they just not updated properly? If it's the latter, then I'm happy with it all

[ffreyer]

I just realized that a lot of tests pass the RNG with triangulate(..., rng = RNG.STABLE_RNG) but a lot also don't. Are the ones that don't at risk of failing in the future?

[DanielVandH]

For all those that don't, the triangulation is unique and so there is no risk of failure. For Voronoi diagrams, even if the triangulation is non-unique then the diagram is still unique. So, there should be no risk of failure (passing rng = RNG.STABLE_RNG into those is still fine if you would like to, but I just didn't need to).

[ffreyer]

Ok, I just got worried for a second. From my side this is good to go as well

[DanielVandH]

Hurrah! Thank you @SimonDanisch and @ffreyer :)

Do you know when a new Makie version might be registered so that these functions are available again @SimonDanisch? No rush, just want to check.

[SimonDanisch]

Trying to tag in the next day or two!