WebGPURenderer: Align integer attribute check of WebGL backend.

[Mugen87]

Fixed #28898.

Description

This PR aligns the WebGL backend of WebGPURenderer to WebGLRenderer when deciding whether buffer data are integer or not.

Only gl.INT or gl.UNSIGNED_INT are integer data input or when attribute.gpuType is set to IntType.

three.js/src/renderers/webgl/WebGLBindingStates.js

Line 345 in 9834113

const integer = ( type === gl.INT || type === gl.UNSIGNED_INT || geometryAttribute.gpuType === IntType );

[github-actions]

📦 Bundle size

Full ESM build, minified and gzipped.

Filesize dev	Filesize PR	Diff
683.7 kB (169.3 kB)	683.7 kB (169.3 kB)	+0 B

🌳 Bundle size after tree-shaking

Minimal build including a renderer, camera, empty scene, and dependencies.

Filesize dev	Filesize PR	Diff
460.9 kB (111.2 kB)	460.9 kB (111.2 kB)	+0 B

[Mugen87]

three.js/src/renderers/webgl/WebGLBindingStates.js

Line 345 in 9834113

const integer = ( type === gl.INT || type === gl.UNSIGNED_INT || geometryAttribute.gpuType === IntType );

Um, I think this policy can be changed. With WebGL 2 singed and unsigned int32 are also a valid inputs for float attributes in the shader. So maybe it's best to just change the line to:

const integer = ( geometryAttribute.gpuType === IntType );

[RenaudRohlinger]

This just breaks Uint16Array support in the WebGL Backend. And now generate this kind of error:

[.WebGL-0x13000c4ea00] GL_INVALID_OPERATION: Vertex shader input type does not match the type of the bound vertex attribute.

Firefox:
WebGL warning: drawElementsInstanced: Vertex attrib 1 requires data of type INT, but is being supplied with type FLOAT.

Are you sure about this one? /cc @Mugen87

[Mugen87]

This change was required to make compressed models work with the WebGL backend so I would say the previous approach wasn't right.

Do you mind demonstrating with a fiddle how the error occurs?

Ideally, the GLSL builder should only generate iuvec* if the gpuType is IntType or when Uint32Array and Int32Array is used. In all other cases, the shader type should be float.

Related: #28920 (comment)