fix false positive when being inside the edge of a block

[maxomatic458]

closes #2000 (Happens with all blocks that the player can collide with, not just gravity blocks)

Currently tested on 1.8.8/1.8.9 (same version as in issue).
I will do some more testing on other versions before marking it as ready.

let me know if there is anything important i need to consider when using
isNearGlitchyBlock, i just used it here to increase the collision threshold when the player is partially inside a block.

[ManInMyVan]

using uncertainty seems wrong, this should be fixed properly

[maxomatic458]

@ManInMyVan

using uncertainty seems wrong, this should be fixed properly

So should grim be able to accurately predict movement like that?
Im not sure if that is easy to implement as thats some really glitchy mc behavior
(the difference between the predicted and actual movement is also pretty different / seems pretty random)

Do you have any specific idea in mind?

It would also be possible to just force push the player out of the block (But im not sure if thats "intended" as that would change vanilla behavior)

[ManInMyVan]

I have a feeling this will false at high speeds

[ManInMyVan]

Do you have any specific idea in mind?

Are we detecting if the player is in a block properly? If not, fix that

[maxomatic458]

changing the uncertainty value from 0.1 to 0.17 fixes the false positive. (0.17 was the value where the values where consistently green)

But at high speeds (speed lvl ~50+; im assuming more than 1m/tick) the game client lets the player pass through the block.
How should this be handled?
Should this be reported (as its not really valid movement) or not (because its possible in vanilla).

output.mp4

[maxomatic458]

currently testing this on every version.

and im having trouble getting the plugin to work on 1.13+

gradle tells me i need java 16+ to build the plugin, but the papermc 1.13 server requires java 13 or older
(so far the fix is working with every version up to 1.12)

any tips for properly building the plugin for those versions?

[SamB440]

See "other versions": https://github.com/GrimAnticheat/Grim/wiki/Updating-to-Java-17#other-versions

[maxomatic458]

I tested this so far with the following versions:

• 1.8.9
• 1.9
• 1.10.2
• 1.11.2
• 1.12.2
• 1.13.2

In versions 1.14+ This issue does not exist (the player doesn't "bug around" like in the older versions)

But at high speeds (speed lvl ~50+; im assuming more than 1m/tick) the game client lets the player pass through the block.
How should this be handled?
Should this be reported (as its not really valid movement) or not (because its possible in vanilla).

(see clip above)

Let me know how we should handle this and then the PR would be ready.

[maxomatic458]

@ManInMyVan
let me know if/how this edge case (being able to pass through the block at high speeds) should be handled.

using uncertainty seems wrong, this should be fixed properly

i believe it's fine to do it here, as the original implementation for handling pushing did so too (the uncertainty was just a bit to low)

[Axionize]

Is there any reason for the uncertainty to be 0.17 in 1.14.4+ ?