Precision issue with generated HLSL and float literal

[rickbrew]

This shader:

[D2DInputCount(0)]
[D2DGeneratedPixelShaderDescriptor]
internal readonly partial struct BadShader
    : ID2D1PixelShader
{
    public float4 Execute()
    {
        return new float4(256, 60000, 131072.65f, 1.0f);
    }
}

generates this HLSL:

        /// <inheritdoc/>
        [global::System.CodeDom.Compiler.GeneratedCode("ComputeSharp.D2D1.D2DPixelShaderDescriptorGenerator", "3.0.0.0")]
        [global::System.Diagnostics.DebuggerNonUserCode]
        [global::System.Diagnostics.CodeAnalysis.ExcludeFromCodeCoverage]
        static string global::ComputeSharp.D2D1.Descriptors.ID2D1PixelShaderDescriptor<BadShader>.HlslSource =>
            """
            #define D2D_INPUT_COUNT 0

            #include "d2d1effecthelpers.hlsli"

            D2D_PS_ENTRY(Execute)
            {
                return float4(256, 60000, 131072.66, 1.0);
            }
            """;

Notice how the float4 in the C# has 131072.65f but the HLSL says 131072.66.

If I change to 131072.64f then it correctly emits 131072.64.

[rickbrew]

Now, to be fair, these two numbers do end up rounding to the same value, as evidenced by running this in the immediate window in VS:

131072.65f.ToString("N16")
"131,072.6562500000000000"
131072.66f.ToString("N16")
"131,072.6562500000000000"

But, I would still expect the generated HLSL to faithfully represent the C# code for something like this.

OR, I would expect it to emit 131072.65625 for both values

[rickbrew]

This came about while I was investigating #780 .

I don't think this issue (#779) is a high priority bug. HLSL is emitted that does not match the C# source, but it does produce the correct float value at runtime.

Instead, I consider this to be a debugging hazard: the HLSL will not match the C#, and it will be confusing.