Different anchors for `x` and `y` labels.

[patrick-kidger]

I have the following:

#import "@preview/cetz:0.4.2"
#import "@preview/cetz-plot:0.1.3"
#cetz.canvas({
  cetz.draw.set-style(axes: (label: (offset: 0.2, anchor: "north")))
  cetz-plot.plot.plot(
    size: (2, 2),
    axis-style: "school-book",
    x-label: [Foo],
    y-label: [Bar],
    x-tick-step: none,
    y-tick-step: none,
    cetz-plot.plot.add(((0, 0), (1, 1)))
  )
})

however this only produces a reasonable-looking label for the x label. The y label overlaps with the y-axis, and for this label I would like to be able to set anchor: "south".

I've already consulted the manual and couldn't find any documented way to adjust both labels independently. (I've also tried various suggestively-named arguments, e.g. axes: (y-label: (anchor: "south")), but these don't seem to have any effect and are presumably just silently ignored.)

Is this possible?

(And whilst I'm here: setting anchor: "south" mostly gets me what I want for the y label, but a little extra vertical space is still needed to get it far enough away. I can hack around this by adding #v(0.3em) to the y-label, but it also struck me that there might be a better way to do this.)

[johannes-wolf]

You can specify styles per axis:

cetz.draw.set-style(axes: (label: (offset: 0.2, anchor: "north"), y: (label: (anchor: "south", offset: 0))))
                                                                  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~

For the #v(0.3em) problemen, there is no better solution.
I'll add support for passing vectors to offset:.

[patrick-kidger]

Awesome, thank you! I didn't manage to figure out the correct dictionary nesting there. Happy to confirm that that works :)

In passing, it would be nice to a little more discoverability in the API – AFAICT the manuals don't give a comprehensive list of dictionary keys, and there's no easy way to discover them at runtime. Conversely, extra keys are silently ignored, rather than loudly rejected. It would be a huge usability boon to have some means of resolving these points.

(One way of doing this would be to use classes instead of dictionaries, though obviously I'm advertising my own package there – I'm sure you could construct something analogous too.)

[johannes-wolf]

Awesome, thank you! I didn't manage to figure out the correct dictionary nesting there. Happy to confirm that that works :)

In passing, it would be nice to a little more discoverability in the API – AFAICT the manuals don't give a comprehensive list of dictionary keys, and there's no easy way to discover them at runtime. Conversely, extra keys are silently ignored, rather than loudly rejected. It would be a huge usability boon to have some means of resolving these points.

(One way of doing this would be to use classes instead of dictionaries, though obviously I'm advertising my own package there – I'm sure you could construct something analogous too.)

Yeah, cetz-plot needs an overhaul… I made a lot of bad decisions when implementing it. Maybe lilaq (https://typst.app/universe/package/lilaq/) suits your needs? I am not sure if working on the plotting part of cetz-plot is really worth it. Ideally it would be a thin layer on top of cetz supporting axes and viewports etc. with less automagic.

[patrick-kidger]

Oh, very nice! I wasn't aware of Lilaq. I'll take a look, and thank you :)