Yuri Rubinsky

Well, that will require to generate the version of function with `samplerArray` parameter internally. Plus, the UV coordinates should also be handled somehow (they are `vec3` in that case). Maybe...

@paddy-exe OK, here is the instruction to how to test it: 1. Download: [godot-cpp.zip](https://github.com/godotengine/godot/files/11394431/godot-cpp.zip) - this version is updated to 8c729f0f34a92b92a1a8204e1d56b6ad05cfd27b: 2. Compile it with the following command: `scons arch=x86_64...

Currently, I lost my vision about it, too much I need to call to mind, maybe others may make it.

Note: However, the average performance cost (in milliseconds) of calculating the path is increased from 0.0005 to ~0.0025~ 0.0012 (on 1000x1000 grid, to find the path from (0, 0) to...

@lawnjelly ah, yes, it's simple - [TestAStarGrid.zip](https://github.com/godotengine/godot/files/10499494/TestAStarGrid.zip) So when it's compiled at current master, the result will be:  With this PR:  Which is worse that previous, but AStarGrid2D...

> Yes, I can confirm to start with, changing nbors to a LocalVector, reusing on each run and just calling clear() rather than using a linked list reduces the time...

Okay, that would require a lot more work, and currently I lost interest a bit. Maybe later I would touch this again (after 4.0 is released).

I guess you can change these hotkeys in the editor's shortcuts, as a temporary solution, but definitely this doesn't sound good and need to be fixed.

You need to squash the commits as required by our pipeline (https://docs.godotengine.org/en/latest/community/contributing/pr_workflow.html#modifying-a-pull-request).

Thanks for your contribution. You need to squash the commits as required by our pipeline (see [docs.godotengine.org/en/latest/community/contributing/pr_workflow.html](https://docs.godotengine.org/en/latest/community/contributing/pr_workflow.html)).