allow float32 literals

[azur-wolve]

Sometimes want to use them for whatever reason.

value * 3.14f

[Calinou]

for whatever reason.

Unfortunately, this isn't sufficient on its own to add a new feature to GDScript. Did you run into performance issues that would likely be resolved by using 32-bit primitive floats? Note that vector types already use 32-bit floats unless you're using a double-precision build.

All floats in GDScript are 64-bit, but you can use PackedFloat32Array when performance or memory usage matters.

[azur-wolve]

there are some cases when want to reproduce with fidelity existing code like DOOM, Quake or Half-Life that the outcomes are different because they use float32 literals in some places but GDScript uses float64

[AThousandShips]

Why not just use the actual value of the literal? Every 32 bit float value is exactly representable in a 64 bit one so there's an exact representation in 64 bits for that value that's distinct from the more accurate value of the 64 bit value

This is so niche that it doesn't really make sense to add a feature for it, but you can easily just convert the values as needed

[Calinou]

If you want to simulate a lower level of precision, you could always work with a Vector2 types and only read its x component.

That said, I doubt you'd ever see the difference between float and double when replicating Doom/Quake/Half-Life physics. (Doom physics are purely fixed point anyway, so you'd want to use integers to replicate them.)