Allow overriding how scripted objects are converted to strings

[LeonardMeagher2]

solves #26796

• ADD String to_string() method to Object which can be overriden by scripts
• ADD String to_string(r_valid) method to ScriptInstance to allow langauges to control how scripted objects are converted to strings
• IMPLEMENT to_string for GDScriptInstance, VisualScriptInstance, and NativeScriptInstance
• ADD Documentation about Object.to_string
• Changed Variant::operator String to use obj->to_string()

[akien-mga]

Please edit the commit log to be more explicit, the first line of a Git commit log is it's title and should be a short but clear indication of what the commit does.

[LeonardMeagher2]

oops forgot to add my code

[neikeq]

We should cache the StringName for "to_string" (check core_string_names.cpp). Otherwise, it looks good.

[LeonardMeagher2]

@neikeq, I'll get this done during my lunch break today

[Calinou]

Should the method be named to_string or _to_string, considering it's a function that can be overridden?

[neikeq]

@Calinou There are inconsistencies about this. Have a look at #19209.

[LeonardMeagher2]

@Calinou I felt like omitting the _ made sense regardless of it being overridden since it always returns something. You can do obj.to_string() and I wouldn't expect people to do obj._to_string() in their code.

My perspective about using the underscore is it appears most virtual functions that don't have the underscore are meant to act like an interface where you can request some information or tell it to do something, and virtual functions that have an underscore are usually responding to or expecting input from some process. It's all very confusing haha

[LeonardMeagher2]

@neikeq I've added to_string to CoreNameStrings hopefully I did that right.

Edit:
Looks like I'll need to change what line I add the core_string_names header according to the clang-format rules

[LeonardMeagher2]

Ran clang-format, we should be good to go now.

[akien-mga]

I think this should be bound with BIND_VMETHOD so that it is properly identified as virtual in the documentation (see _get_property_list in object.cpp, or all the virtual callbacks in node.cpp).

The documentation should also be improved to mention that the method can be overridden from script to customize the String representation of objects. And I would describe what the default representation means ([name:rid]).

Finally, I know that it's not done consistently but I think we do favor putting a leading underscore on virtual methods that can be overridden. See e.g. how get_minimum_size calls _get_minimum_size from scripts if overridden in control.cpp. So you would still call to_string() in your code (or it will be called behind the hood by print()), but you can customize it by overriding _to_string().

oops :P

[LeonardMeagher2]

@akien-mga I've implemented you requested changes.

[karroffel]

NativeScript parts looks fine

Changes done.

[akien-mga]

Seems mostly good to me, just a few nitpicks on documentation.

[LeonardMeagher2]

@akien-mga I've made changes to the documentation. Let me know if you'd prefer a better example, or that I use this kind of wording.

which will be used both when calling [method to_string] manually but also in [method @GDScript.print] and similar.

[akien-mga]

Thanks a lot for your contribution! :)

[LeonardMeagher2]

Amazing! thank you @akien-mga