p5.Vector.fromAngle returns a 3d Vector while the Docstring says it creates a 2d Vector

[jan-23]

Most appropriate sub-area of p5.js?

• Math

p5.js version

1.11.10

Steps to reproduce this

p5.Vector.fromAngle returns a 3d Vector while the Docstring says it creates a 2d Vector.
The Z coordinate is set to 0.

Snippet:

const v = p5.Vector.fromAngle(1, 1);
console.log(v.values[2] === 0);
console.log(v.values.length === 3);

This is confusing and might cause errors.

[Iron-56]

I would like to work on this issue, I'm still quite new to contribution. I'll need to check if any of the examples use this function and change either the documentation or the function accordingly. @davepagurek

[davepagurek]

I'm not sure that this actually needs fixing, as in general we treat vectors with z = 0 as 2D vectors in 1.x. v.values.length is not really the way to tell if a vector is 2D or 3D. cc @holomorfo: how do you feel this should be handled in 1.x?

[holomorfo]

Hi @davepagurek and team, yes on version 1.x p5.js treats all vectors as 3D vectors internally, with 2D vectors simply being 3D vectors where the z-component is always 0. This has been the behavior from way back before I joined the project, @jan-23 In the sample code you mention you call v.values but the values array was introduced on version 2.x from my understanding calling values on 1.x would not work.

On that note, on v 2.x we added the flexibility for the vector to be a n-dimentional array, but to keep backwards compatibility we added the special case that 2D vectors would be 3D with z as zero, so this was intentional.

Having the 2D being a special case of the 3D gives an unified approach that simplifies the mathematics, since all vector operations (dot product, cross product, etc.) can be implemented once for 3D vectors and work seamlessly with 2D vectors where z=0. It's a common and efficient design pattern in graphics and mathematical libraries.

While I agree that it could be confusing I would like to kindly asked @jan-23 if you can provide some examples of instances where you think this might break or cause some negative behavior. I imagine if the intention is to use the values.lenght value as source of truth for the vector dimension.

At the moment I don't think we should change this to avoid breaking any downstream dependencies on the render methods, but we could explore if we also need a more general Vector class that behaves more "mathematically" similar to what @GregStanton has been suggesting for the matrix implementation.

Thank you @Iron-56 for your interest in this issue. Since you are looking to ways to contribute, we can find a small entry issue on the Matrix module if thats ok, what do you think?

[ksen0]

Thanks for the input everyone! I think this issue can be closed in a few days, because the behavior is intentional - unless someone can proposed an improvement to the 1.x documentation?

If there are ideas on how behavior could be more intuitive, then I think it can inform the design of Vector/Matrix future 2.x releases, but in 1.x, behavior would stay the same.