Sync defs

[tomfull123]

Code is from the godot repo, had to remove duplicate definitions from math.hpp for some types and moved out the math defs into their own file to mirror godot's file structure more closely.

I was looking at syncing some other classes, but there were some types missing, hence this PR.

Briefly tested it on one of my gdextension projects and there doesn't seem to be any issues when running it with Godot 4.4.

[tomfull123]

This PR also gives us the is_zero_constructible type, which is used by a lot of classes and would allow us to sync those classes too

[Ivorforce]

Thanks! Overall, this looks good to me (although it's always hard to know for sure with these complex updating changes). Assuming there's not many changes to how Godot's header looks.

I'm not sure how to handle compatibility breakages for refactors. E.g. ABS was moved to the math namespace. We could still keep it temporarily but deprecate it. But this seems like a lot of work long term.

[tomfull123]

Thanks! Overall, this looks good to me (although it's always hard to know for sure with these complex updating changes). Assuming there's not many changes to how Godot's header looks.

I'm not sure how to handle compatibility breakages for refactors. E.g. ABS was moved to the math namespace. We could still keep it temporarily but deprecate it. But this seems like a lot of work long term.

Yeah I saw that ABS was already deprecated and that it was only used in 2 places, so I changed those, slightly out of scope for the PR, but I didn't think it would impact PR readability too badly

And ABS has already been removed from Godot

[dsnopek]

I'm not sure how to handle compatibility breakages for refactors. E.g. ABS was moved to the math namespace. We could still keep it temporarily but deprecate it. But this seems like a lot of work long term.

Hm. Normally, it would be my preference to have a deprecated alias for ABS, but there's a high risk of us accidentally removing it when we sync with Godot next - I'm not sure how to avoid that...

We have the using real_t = godot::real_t one too that was correctly maintained in this PR, but I could easily see us forgetting it.

Maybe we could put the deprecated aliases in special headers that are included into the main one?

Like, math.compat.inc and we could start using DISABLED_DEPRECATED like in Godot, putting something like this at the end of math.hpp:

#ifndef DISABLE_DEPRECATED
#include "math.compat.inc"
#endif

[Ivorforce]

Maybe we could put the deprecated aliases in special headers that are included into the main one?

This sounds like a great idea to me!

[tomfull123]

I'm not sure how to handle compatibility breakages for refactors. E.g. ABS was moved to the math namespace. We could still keep it temporarily but deprecate it. But this seems like a lot of work long term.

Hm. Normally, it would be my preference to have a deprecated alias for ABS, but there's a high risk of us accidentally removing it when we sync with Godot next - I'm not sure how to avoid that...

We have the using real_t = godot::real_t one too that was correctly maintained in this PR, but I could easily see us forgetting it.

Maybe we could put the deprecated aliases in special headers that are included into the main one?

Like, math.compat.inc and we could start using DISABLED_DEPRECATED like in Godot, putting something like this at the end of math.hpp:

#ifndef DISABLE_DEPRECATED
#include "math.compat.inc"
#endif

Added the math.compat.inc file and moved the using real_t... def and the ABS macro into it and I've added the include at the bottom of math.hpp, how does it look now? 😊

[Ivorforce]

Assuming the bindings match Godot's closely, this works for me.

[dsnopek]

Thanks!

This is looking good to me. I've skimmed over the updated files while looking at the Godot files next to them, and I can't spot any issues