decoration: Allow dynamic amplitudes

[johannes-wolf]

Allows passing a function of the form ratio -> float or an array of floats as amplitude: to path decorations.

Adds a square square-wave decoration.

[Jollywatt]

One more nitpick:

  //   ╭ ma ╮        ▲
  //   │    │        │ Up
  // ..a....m....b.. '
  //        │    │
  //        ╰ mb ╯

Since the wave points ma and mb are at different positions along the curve, you could evaluate the amplitude at both points independently.
This leads to slightly nicer results such as:

compared to what you currently get:

Both of these are made with amplitude: t => calc.min(float(t), 1 - float(t)).
To make the first image, wherever you do

let fn(i, a, b, norm) = {
    let ab = vector.sub(b, a)
    let up = vector.scale(norm, resolve-amplitude(style.amplitude, i, num-segments) / 2)
    let down = vector.scale(
      up, -1)

I replaced it with

let fn(i, a, b, norm) = {
    let ab = vector.sub(b, a)
    let up = vector.scale(norm, resolve-amplitude(style.amplitude, i + 0.25, num-segments) / +2)
    let down = vector.scale(norm, resolve-amplitude(style.amplitude, i + 0.75, num-segments) / -2)

and reverted my earlier suggestion of changing (amplitude)(segment / num-segments * 100%).

This just means there's twice as many amplitude evaluations, but may lead to better results for large segment lengths… what do you think?

[Jollywatt]

This now needs to be reverted back to (amplitude)(segment / num-segments * 100%), since segment now ranges from 0.25 to num-segments + 0.75 😅

You can see what I mean with the test case

wave(line((0,3), (4,3)), amplitude: t => { calc.sin(t/100%*calc.pi) })

[johannes-wolf]

Yeah, I am playing around with it. With the .25 addition, you won't get 0%/100% in the callback, which is a bit unexpected, I think.

[Jollywatt]

I guess the segment variable is no longer an index, but rather the coordinate along the curve 0 <= t <= num-segments

[Jollywatt]

Have you considered amplitude.at(calc.rem(int(2*segment), amplitude.len()), default: 0)? That gives the user control over individual peaks and troughs. With int(segment), you can only control them in full-wavelength pairs.

[Jollywatt]

I just had one possible suggestion:

Have you considered amplitude.at(calc.rem(int(2*segment), amplitude.len()), default: 0)? That gives the user control over individual peaks and troughs. With int(segment), you can only control them in full-wavelength pairs.

[johannes-wolf]

I just had one possible suggestion:

Have you considered amplitude.at(calc.rem(int(2*segment), amplitude.len()), default: 0)? That gives the user control over individual peaks and troughs. With int(segment), you can only control them in full-wavelength pairs.

Will look into that.

[Jollywatt]

Perfect:)
I think length amplitudes should also be allowed, but I'm happy to make that a separate PR. It's just a matter of adding util.resolve-number(ctx, amplitude) and adding some tests.