angle results are hard to predict and control

[knuesel]

The angle API is a bit tricky to use in programmatic drawing, where the angle value is not known statically. One must play with inner and mark to get the right result depending on the angle value. For example I often need figures for different trigonometric angles:

#let trig-angle(alpha) = cetz.canvas(length: 2cm, {
  import cetz.draw: *

  let O = (0, 0)
  let P = (1, 0)
  let Q = (alpha, 1)

  line(O, P, mark: (end: ">"))
  line(O, Q, mark: (end: ">"))
  cetz.angle.angle(O, P, Q, mark: (end: ">"), label: [#alpha], label-radius: 150%)
  rect((-1, 0), (1, 0), stroke: none) // for bounding box
})

#for alpha in (60deg, 240deg, -60deg, -240deg) {
  box(trig-angle(alpha))
}

The first drawing (60deg) is what I expected, but the others all require different fixes:

• For 240deg I need inner: false
• For -60deg I must replace end: ">" with start: ">"
• For -240deg I need both of the above

(Here I could implement the four cases with conditional logic, but in general the angle can be specified by all sorts of coordinates with no explicit angle value.)

What would you think of the following changes to the API:

• make angle(O, A, B) always draw from OA to OB (this makes the effect of mark start and end more predictable)

• allow inner: auto, in which case the angle is drawn counterclockwise for positive values, and clockwise for negative values (I think this should be the default, but having this opt-in would be fine)

[johannes-wolf]

I've created a PR #729. Could you please validate that it fulfils your expectations. I did not implement the auto behaviour, as it is not clear to me, when the angle would be negative. Do you mean if it is > 180deg?

[knuesel]

Thanks! I'm going to comment on that PR then.

[knuesel]

(By negative angle I meant clockwise, but then indeed the whole sentence doesn't make sense, sorry! But I'll elaborate in the PR comments)