Entities: inconsistent collision outcomes

[Termos]

When a 'physical' entity collides with a node, the perpendicular velocity component goes 0, while parallel components remain unchanged, and this is cool.

The problem is, sometimes, not always, on the very next step after the first step following the collision, parallel components become zero as well for no apparent reason.

Is this a known problem?
This is v 4.17, local game.

[sorcerykid]

That is a "feature" in the CPP collisionMoveSimple() function, which seems to overcompensate for collisions against a node face that is not actually exposed, and therefore reporting false positives. I ended up overcoming the problem by sending information about collisions and movement state to Lua for post-processing. In this video, the "glitch" is fairly perceptible, since its prior to the workaround.

viewtopic.php?p=343126#p343126

[Termos]

Can you tell me more about this workaround? Is it something that you did on your copy or there's a chance of fixing it in a future MT release?