Add show_data for band

[ffreyer]

Description

Adds a custom show_data for band which displays the x position of the mouse and the relevant (interpolated) min and max value of the band at that x value. This also affects density and whatever else relies on band. Currently this shows a bounding box for the underlying mesh.

I'm making the assumptions that

• band is used in 2D
• x values are sorted in plot.converted
• plots.converted is always 2 Point2f arrays, with the first being low values and the second being high values
• x values between low and high points match

Example with density(rand(100)): (3rd commit)

Type of change

• New feature (non-breaking change which adds functionality)

Well technically it's changing the current behavior but a bounding box is not a useful thing to get here...

Checklist

• Added an entry in NEWS.md (for new features and breaking changes)
• Added or changed relevant sections in the documentation
• Added unit tests for new algorithms, conversion methods, etc.
• Added reference image tests for new plotting functions, recipes, visual options, etc.

[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	29.06s (28.97, 29.14) 0.07+-	17.62s (17.43, 17.75) 0.11+-	16.63s (16.46, 16.81) 0.11+-	10.27ms (10.16, 10.33) 0.06+-	114.17ms (112.64, 115.88) 1.24+-
master	29.30s (29.16, 29.44) 0.10+-	17.61s (17.38, 17.85) 0.17+-	16.57s (16.42, 16.75) 0.13+-	10.47ms (10.31, 10.90) 0.20+-	115.20ms (114.09, 116.29) 0.76+-
evaluation	-0.82%, -0.24s faster ✓ (-2.73d, 0.00p, 0.09std)	+0.08%, 0.01s invariant (0.10d, 0.85p, 0.14std)	+0.35%, 0.06s invariant (0.48d, 0.39p, 0.12std)	-1.98%, -0.2ms faster ✓ (-1.40d, 0.03p, 0.13std)	-0.90%, -1.03ms invariant (-1.01d, 0.09p, 1.00std)
CairoMakie	26.90s (26.77, 26.96) 0.06+-	15.96s (15.83, 16.06) 0.10+-	2.42s (2.41, 2.46) 0.02+-	11.26ms (11.19, 11.37) 0.06+-	4.47ms (4.45, 4.51) 0.02+-
master	26.99s (26.80, 27.22) 0.14+-	16.27s (16.06, 16.67) 0.20+-	2.44s (2.40, 2.48) 0.02+-	11.16ms (10.99, 11.31) 0.11+-	4.44ms (4.35, 4.49) 0.05+-
evaluation	-0.31%, -0.08s invariant (-0.77d, 0.18p, 0.10std)	-2.00%, -0.32s faster ✓ (-2.03d, 0.00p, 0.15std)	-0.42%, -0.01s invariant (-0.48d, 0.39p, 0.02std)	+0.94%, 0.11ms invariant (1.15d, 0.06p, 0.09std)	+0.74%, 0.03ms invariant (0.92d, 0.12p, 0.03std)
WGLMakie	31.56s (31.32, 31.91) 0.19+-	25.62s (25.25, 26.02) 0.25+-	41.46s (40.94, 42.28) 0.52+-	15.28ms (12.78, 28.58) 5.87+-	1.83s (1.79, 1.87) 0.03+-
master	31.55s (31.42, 31.69) 0.10+-	25.60s (25.48, 25.73) 0.09+-	42.60s (40.54, 54.17) 5.10+-	12.37ms (11.71, 13.89) 0.70+-	1.63s (1.60, 1.66) 0.03+-
evaluation	+0.02%, 0.01s invariant (0.04d, 0.94p, 0.15std)	+0.10%, 0.02s invariant (0.13d, 0.81p, 0.17std)	-2.75%, -1.14s invariant (-0.31d, 0.58p, 2.81std)	+19.08%, 2.92ms noisy🤷‍♀️ (0.70d, 0.24p, 3.29std)	+10.91%, 0.2s slower❌ (6.56d, 0.00p, 0.03std)

[jkrumbiegel]

Looks good, the only thing is that I think (0.345, 0.000..0.932) like in the example is a bit hard to read. Maybe spaces around the .. like (0.345, 0.000 .. 0.932)? Or any other suggestions?

Oh and what happens with vertical bands or flipped densities?

[ffreyer]

Do we have (x, x, y) bands? I don't see a keyword or plot method for it.

[ffreyer]

Oh it's a raw input of Points. I'll have to tweak this then.

Looks good, the only thing is that I think (0.345, 0.000..0.932) like in the example is a bit hard to read. Maybe spaces around the .. like (0.345, 0.000 .. 0.932)? Or any other suggestions?

Some thoughts:

• ($x, $ymin, $ymax) (or $xmin, $xmax, $y) was my first thought, but that's even worse for readability imo
• Two lines like "($x, $ymin)\n($x, $ymax)" would be another option
• The $ymin..$ymax matches the Interval notation which I thought was rather fitting for band. I guess we could also have ($ymin, $ymax) or "$ymin <= y <= $ymax"
• Displaying the current position rather than band min/max would be an option, but I think less useful? I think that's more something for a separate function/tool which just shows mouseposition() regardless.
• Having two or even three (top, bottom, current pos) tooltips would require reworking tooltip! and is too much noise I think.

[ffreyer]

I got stuck for a while trying to figure out a direct way to find the line I need to draw and ultimately just used an iterative method. Now it can handle densities in :x and :y direction and also bands like this:

inner = [(1 + rand()) * Point2f(cos(phi), sin(phi)) for phi in range(0, 2pi, length=51)]
outer = [(4 + rand()) * Point2f(cos(phi), sin(phi)) for phi in range(0, 2pi, length=51)]
fig, ax, p = band(inner, outer)
di = DataInspector(fig)
fig

You can also swap outer and inner without issues. (It will break if the quads look like |><|, e.g. if you swap lows and highs throughout the band. At this point band is malformed anyway though)