[p5.js 2.0+ Bug Report]:WebGL silently ignores mat2 shader uniforms (no error, no warning)

[harshiltewari2004]

Most appropriate sub-area of p5.js?

• Accessibility
• Color
• Core/Environment/Rendering
• Data
• DOM
• Events
• Image
• IO
• Math
• Typography
• Utilities
• WebGL
• WebGPU
• p5.strands
• Build process
• Unit testing
• Internationalization
• Friendly errors
• Other (specify if possible)

p5.js version

2.3.0

Web browser and version

Not browser-specific — the missing code path is in p5's source, so this reproduces in any WebGL-capable browser.

Operating system

Not OS-specific (same reason).

Steps to reproduce this

Steps to reproduce

1. Create a WebGL shader that declares a uniform mat2.
2. Set that uniform from sketch code with setUniform(...).
3. Observe: no error and no warning, but the value never reaches the shader — the uniform stays at the GL default (zero matrix). The identical pattern with a mat3 or mat4 uniform works.

Minimal illustration

The definitive evidence is the source analysis below — this snippet just shows the symptom.

let s;

function setup() {
  createCanvas(100, 100, WEBGL);
  s = createFilterShader(`
    precision highp float;
    uniform mat2 uM;
    void main() {
      // diagonal of uM drives the red & green channels
      gl_FragColor = vec4(uM[0][0], uM[1][1], 0.0, 1.0);
    }
  `);
}

function draw() {
  s.setUniform('uM', [1.0, 0.0, 0.0, 1.0]); // identity
  filter(s);
  // Expected: yellow (uM uploaded as identity -> 1,1,0)
  // Actual:   black  (uM never uploaded -> stays at the 0 matrix -> 0,0,0)
}

Root cause

The root cause is in src/webgl/utils.js. The setUniform switch handles FLOAT_MAT3 (line 226, via gl.uniformMatrix3fv) and FLOAT_MAT4 (line 229, via gl.uniformMatrix4fv), but there is no case gl.FLOAT_MAT2 and no default: branch — the switch ends with //@todo complete all types at line 327. A mat2 uniform falls straight through and is never uploaded.

A repo-wide check confirms this is structural, not a one-off miss: gl.uniformMatrix2fv — the WebGL call required to upload a 2×2 matrix — appears zero times anywhere in src/, while uniformMatrix3fv and uniformMatrix4fv each appear once. So there is no code path on which a mat2 uniform could reach the GPU.

getWebGLUniformMetadata likewise omits FLOAT_MAT2 from its isArray classification (around lines 360–369), consistent with mat2 never having been wired up.

WebGL vs WebGPU

For contrast (as motivation, not the core claim): the WebGPU backend handles square mat2. In src/webgpu/p5.RendererWebGPU.js it parses the matrix dimension generically (~line 2174) and maps mat2x2<f32> (line 3103), and throws a clear Non-square matrices not implemented yet for non-square (line 2172).

So mat2 works in WebGPU but silently does nothing in WebGL — and where WebGPU at least fails loudly on what it doesn't support, WebGL fails silently. (Non-square matrices are out of scope here; WebGPU's explicit throw shows that's separately tracked.)

Possible approaches (would like maintainer direction before implementing)

1. Add the missing FLOAT_MAT2 case (gl.uniformMatrix2fv) to the switch and to the isArray classification — completing the //@todo.
2. Or, if full mat2 support isn't wanted yet, emit a friendly warning for currently-unsupported uniform types instead of silently no-op'ing, in keeping with p5's friendly-error approach.

Happy to fix this myself once an approach is agreed on @ksen0 @davepagurek !

[harshiltewari2004]

Hi @ksen0 ! — about a week since I filed and I wanted to make sure it didn't slip past triage.

The one open thing is a direction call between the two approaches above: completing the //@todo with a proper FLOAT_MAT2 case, or emitting a friendly warning for unsupported uniform types for now. Happy to implement it myself either way once you've picked a direction. No rush — whenever's convenient. Thanks!