Rendering into RGIntegerFormat textures doesn't work

[Aloalo]

Description

When creating a WebGLRenderTarget with RGIntegerFormat a WebGL error occurs: WebGL: INVALID_VALUE: texImage2D: invalid internalformat or (GL_INVALID_OPERATION: Fragment shader output type does not match the bound framebuffer attachment type.). Managed to fix the issue by adding the follwing code to this function:

if ( glFormat === _gl.RG_INTEGER ) {

	if ( glType === _gl.UNSIGNED_BYTE ) internalFormat = _gl.RG8UI;
	if ( glType === _gl.UNSIGNED_SHORT ) internalFormat = _gl.RG16UI;
	if ( glType === _gl.UNSIGNED_INT ) internalFormat = _gl.RG32UI;
	if ( glType === _gl.BYTE ) internalFormat = _gl.RG8I;
	if ( glType === _gl.SHORT ) internalFormat = _gl.RG16I;
	if ( glType === _gl.INT ) internalFormat = _gl.RG32I;

}

Not sure if this is the only place this is needed?

Reproduction steps

1. Create a WebGLRenderTarget with RGIntegerFormat and RG8UI internal format.
2. Render into it.

Code

jsfiddle

Live example

jsfiddle

Screenshots

No response

Version

r160

Device

No response

Browser

No response

OS

No response

[Mugen87]

Would you mind filing a PR with your fix? The addition looks correct. getInternalFormat() is the only place where something is missing for supporting RGIntegerFormat.

BTW: In your fiddle, you are not creating the instance of WebGLCubeRenderTarget correctly. The ctor has only two parameters, not three. Besides, position shader attributes are not of type vec4 but vec3.

In any event, adding your code makes the fiddle work without errors.

[Aloalo]

Ok, thanks, I'll create a PR.

Fixed the WebGLCubeRenderTarget construction in the fiddle for posterity: https://jsfiddle.net/h3e4zncb/1/
The position attribute is vec4, I'm using a RawShaderMaterial. Even if the underlying buffer data is vec3, WebGL will automatically set the 4th component to 1.0.

From https://registry.khronos.org/OpenGL/specs/es/3.0/es_spec_3.0.pdf:

Unspecified y and z components are implicitly set to 0.0 for floating-point array types and 0 for integer array types.
Unspecified w components are implicitly set to 1.0 for floating-point array types
and 1 for integer array types.