I think this is ready to be reviewed.

Note to the person who will be pressing the Merge button, can you merge and squash all commits into one so the commit history doesn't look so messy? I made a mistake with merging branches.

This is starting to look really clean! As Ankith mentioned, it needs tests. It could also probably use a note in the docs.

You got the explicit methods, but it still fails on math.

>>> class VEC(pygame.Vector2):...     pass...>>> a = VEC(5,6)>>> b = VEC(7,8)>>> a + b<Vector2(12, 14)>>>> type(a + b)<class 'pygame.math.Vector2'>>>> d = a * 2>>> type(d)<class 'pygame.math.Vector2'>

It passes-a though. You should also add a test for that.

Apologies if this has already been done somewhere, but I couldn't see it in the PR comment thread.

Yup, Starbuck5 has mentioned this issue. I'm currently working on a solution to that.

is this what we want to happen or should it also return an instance of the subtype?

Nope, this isn't what is supposed to happen.

There are a few more that needs to support subtype, but I'm pushing my latest changes to show some progress.

pygame math tests are failing right now, so please hold the review.

There's a little bit of an oopsie with the subclassing PR, so for swizzling, let's say you have this piece of code

classMyVector2(pygame.Vector2):passprint(MyVector2(3,5).xy)

This piece of code will work just fine, assuming it'll return aMyVector2 type instead of apygame.Vector2, but what if we add this at the end?

print(MyVector2(3,5).xyx

This will return apygame.Vector3 type instead of aMyVector2 since it has 3 dimensions instead of 2.

Any ideas on this problem? Should we disable swizzling if the Vector class has been subclassed, only returnpygame.VectorN for swizzling, or keep it how I described it, returnpygame.VectorN if it can't return itself.

Something strange happened on my end while trying to push my new commits, I thought I initially squashed some of these commits into one and they just somehow reappeared? I think it's best to squash and merge this.

For those OOTL, we discussed swizzling return types on discord, and the current decision is to keep old behaviour for swizzling, returningVectorN instances even when vector subclasses are swizzled.

This PR is bit of an API change, but it's a step in the right direction. User code that subclasses vectors should not break if those subclasses were designed with Liskov substitution principle in mind