KHR extensions

[pandaGaume]

Fix #872 default test behavior and prepare #892 support for clearcoat, sheen and transmission

[Drigax]

I reccomend using tools.isAlmostEqualTo() instead of doing floating point equivalency.

[pandaGaume]

I agree, Then we use the MathUtilities wich is already used in GLTFLight and the ArrayExtension. Both on Utilities shared project
ok ?

[Drigax]

Sounds good, thanks

[Drigax]

can we instead add a definition to our Tools.Math file with definitions of vector2, vector3 and the zero and one constants, similar to babylon.js?

[pandaGaume]

Yes we may have this , but if we create vector, i recommend to introduce this vector class directly into the GltfExport.Entities to avoid new dependencies. The gltf project is supposed to be standalone and self sufficient (wich its not the case today). So, similar to babylon.js, introduce one Maths.cs file with Vector2 and Vector3, and some logic to replace the MathUtilities used in only one place so far.

in the mean time, as you pointed out a poor code design option, i propose to clear this static and introduce

        public static bool IsAllAlmostEqualTo(this float[] current, float value, float epsilon)
        {
            if( current == null)
            {
                return false;
            }

            for (var i = 0; i < current.Length; ++i)
            {
                if (!MathUtilities.IsAlmostEqualTo(current[i], value, epsilon))
                {
                    return false;
                }
            }
            return true;
        }

into ArrayExtensions.cs

then we may discard the static and use the extension as

        public bool ShouldSerializescale()
        {
            return this.scale != null && !this.scale.IsAllAlmostEqualTo(1,float.Epsilon);
        }

then make the code more resilient without the need to introduce new classes.

[Drigax]

That seems like an alright decision, I just like code reuse, and not making architectural decisions for convenience sake, Currently we take dependencies on some shared utilities, helper functions and structures in SharedProjects, as long as we don't take any additional dependencies directly from Max or/Maya2Babylon.

Some standalone math utility, as opposed to an appendix to the gltf entities, but I don't think it matters that much to hold up the PR if you decide otherwise. Where we're using them, we also reused some entities defined in our babylon utilities as part of the glTF exporter, as they are used by the babylon entities.

Ultimately, it would be a good goal to architect such that we can easily make Babylon2GLTF its own standalone executable.

Contributing to our utilities should still allow the project to be relatively standalone, right? That would be akin to using a shared math library or scientific computing library in my opinion.

[Drigax]

I definitely agree with adding a helper function to compare arrays via IsAllAlmostEqualTo,

Alternatively you can override IsAlmostEqualTo to use float[] as well.

[pandaGaume]

Ok this is what i done. I guess PR is ready now.

[Drigax]

Thanks, checking in.