Loss of constant precision

[therontarigo]

My application requires the full IEEE754 precision as supported by the GPUs I target (with precision highp float). I have a constant in the shader source:
float pi = 3.1415927;
Shader Minifier translates this to
float M=3.14159;
resulting in measurably altered results.

Furthermore there is another bug when trying to use a workaround:

float pi =
//[
3.1415927;
//]

results in a parse error. I must wrap the entirety of the constant definition in //[ //] which has the unwanted effect of preventing Shader Minifier from shortening these constant names.

For the example case here, I can use acos(-1.) as my pi constant, but for other high precision constants the limitation stands.

[laurentlb]

Thanks for the report. That's indeed a bug that we should look into.

As a workaround, you can define a macro:

#define pi 3.1415927