ReflectionConstant should have a `getType()` method

[Girgias]

Description

Especially as it can be typed as of PHP 8.3

[nielsdos]

ReflectionClassConstant::getType() exists though (but not documented yet it seems).
But ReflectionConstant::getType() indeed doesn't

[iluuu1994]

Does getType() make sense for ReflectionConstant at this point? Not saying it's likely, but what if we want to allow adding a type to consts? FWIU, getType() doesn't refer to the type of a concrete value anywhere else in ext-reflection.

[nielsdos]

Does getType() make sense for ReflectionConstant at this point?

Unsure, it's consistent though.

Not saying it's likely, but what if we want to allow adding a type to consts? FWIU, getType() doesn't refer to the type of a concrete value anywhere else in ext-reflection.

I don't see the problem. If we were to add type support for constants then we change the implementation and there should be no BC break.

[iluuu1994]

I don't see the problem. If we were to add type support for constants then we change the implementation and there should be no BC break.

Would no type become null? If not, we have an inconsistency with other methods. I don't care too much, I was just pointing it out.

[nielsdos]

Would no type become null? If not, we have an inconsistency with other methods. I don't care too much, I was just pointing it out.

I see what you're saying. But why would the lack of a type become null? I understand this is how types for properties work, but it's a constant, so it can only have a type NULL when the constant has the value NULL.

[iluuu1994]

I don't understand your point. null indicates that no type was specified, not whether the value itself can be null. A class constant with no type currently returns null, even if the actual type is never null. https://3v4l.org/UR3nG#v8.3.12

Confusingly, dumping the entire ReflectionClassConstant outputs type CONST_NAME. https://3v4l.org/RrZIG#v8.3.12

[nielsdos]

Okay, got it, sorry.
In that case implementing this is indeed risky as we'd get stuck with this behaviour...

[iluuu1994]

No worries. It's not a huge issue, but I think it's better to stay consistent and keep the get...Type() convention for user-specified types. We could pick a different name, or just leave it as is. If the user needs the type for some reason, they can just inspect the constant value.

[Girgias]

Right... it is only class constants.

Was working on something, and it would have been handy but turns out it needs to be derived from the value.