The test_move() function may not be behaving properly, when rotating

[vesk4000]

Godot version:
Godot Engine v3.2.3.stable.official

OS/device including version:
Windows 10 Pro, Version 2004

Issue description:
In this demo, I've got this T shape and it can move around in an invisible grid as well as rotate 90 degrees clockwise. It should collide with these blue walls, and when simply moving - it does. But if I try to rotate next to a wall that is to the right of the T shape (when looking from its perspective by taking into account the rotation), the T shape kinda glitches out instead of just not moving or rotating at all like it is supposed to.

Here's the script that's attached to the T shape:

Minimal reproduction project:
Rotation Collision Test.zip

[madmiraal]

This is also fixed with #35945 and its 3.2 version #37498.

[vesk4000]

Yup just compiled your commit (#35945) and tested it with my demo project - it works! Thanks so much for the fix! Do you have an idea when the commit might find itself in the stable version of Godot, because I think this is a very important commit and I don't feel that the current master is stable enough for me to use?

[Calinou]

Do you have an idea when the commit might find itself in the stable version of Godot, because I think this is a very important commit and I don't feel that the current master is stable enough for me to use?

We can't make any guarantees that it will be merged as-is, since reduz disagrees with the implementation.

and I don't feel that the current master is stable enough for me to use?

You can always compile the 3.2 branch from source and apply your own commits on top 🙂

[vesk4000]

You can always compile the 3.2 branch from source and apply your own commits on top 🙂

@Calinou Yeah that is actually a good idea. I see that @madmiraal has made a 3.2 version of the PR, so I think I'll just compile that (now that I actually know how to compile Godot, which really ended up being easier than I expected).

We can't make any guarantees that it will be merged as-is, since reduz disagrees with the implementation.

I haven't extensively looked at the implementation and from what I read it seems that it is a bit of a workaround until/if the physics engine is reworked. But the thing is, it is a massively important workaround in my opinion, given how basic of feature that is. Like I don't know how more people aren't having problems with this, but it's quite nonsensical (for the lack of a better word) how something as simple and basic as my demo didn't work. I really hope this gets fixed in stable as soon as possible, whether it is by #35945 or something else.

[vesk4000]

@Calinou By the way, I closed the issue, because it was fixed by #35945, but since you say that it is unlikely that that PR will be merged, should I reopen it until it is fixed in an official build, or should I keep it closed?

[pouleyKetchoupp]

Fixed by #42574, closing.