Add logging for experimental features - #9109

buildNormalShader also has beta tag, that should also be here?

yep, good catch!

baseComputeShader() (L2379) and compute() (L2725) are @beta and live on Renderer3D, so they are reachable in WEBGL mode where the RendererWebGPU constructor warning never fires...

Compute shaders only actually work in WebGPU mode and will throw an error in WebGL mode so I figured that path was less critical to flag.

"WEBGPU mode is experimental, and API related to it (functions and constraints) may change in future versions. You can get involved by giving feedback to help direct its development!"

What do constraints refer to in this context?

ah sorry "constants"

can be rephrased I am just trying to be explicit that API may change (and that means functions or constants may change) in the future

Liekwise "p5.strands API (build shader functions, the hooks in them, and related constants) may change in future versions. You can get involved by giving feedback to help shape its future!"

There is a trick that you may or may not want to use (I only just thought of it so it may not work) which is that the pattern matcher (first argument) can be a function and if it returns true the decorator will apply and false it won't, which means that if say you add some flags onto the functions itself to mark it as experimental:

p5.prototype.buildStrokeShader.experimental = true

you can just call registerDecorator once and return true whenever the flag is present on a function. That way you don't have to call registerDecorator every time a new experimental API is added.

that's an interesting idea! I'm going to keep it as is for now because sometimes addons augment our functions by doing something like:

let oldMethod = p5.prototype.buildStrokeShader;
p5.prototype.buildStrokeShader = function(...args) {
  // ... do something
  oldMethod.apply(this, args)
}

...and that might accidentally make the property not be visible, or at least makes it a bit more complicated. But I'm thinking this might be a good way to implement FES on p5.strands functions that don't work quite like normal functions, and where the overriding is more dynamic. So I will definitely be revisiting this!