Bad collision with rotated trains at speed exceeding 72 (and angles near 270 or 90)

[Baker7]

Bad collision with rotated trains (func_train) at speed exceeding 72 (and when angles near 270 or 90) == "angles" "0 270 0" or "angles" "0 90 0" or very close to those -- maybe within 30 degrees.

Speed 72 you stay in the train (velocity '0 -72 0')
Speed 73 you fall out of the train (velocity '0 72 0') - in this case you can walk through the back of the train but not the front.

Does not happen with angles '0 0 0' or '0 180 0' as far I can tell.

I suspect a tracebox move function may not divide up the moves enough ... (although I have had falling through the floor of the train issues with train angles '0 90 0' and speed > 72 so not sure how that works)

[Baker7]

Increasing the size of the wall of the train helps greatly. (FTE has no issues and you don't need to do that).

Also the engine physics code seems to allow things that "on ground" of the func_train to block the func_train.

(a func_pushable chair on the train should not impede movement of the func_train, the chair certainly has groundentity = train)

Anyway, I hope I can track down the weakness. Since the speed and the angle matters and the physics for the player run before the physics for the train ....

[Baker7]

It appears one solution is adjusting sys_ticrate to a lower number like 0.010 (Quake is 0.01388889)

[Baker7]

Adjusting the ticrate does not solve everything. For brush models, I am going to experiment taking the yaw out of tracepoint by rotating the mins, maxs to angles 0 0 0 and see what happens.

(I have issues regardless of ticrate where at a func_train rotated to angles 0 90 0 where if I stand at the front of the train it feels like the player is not receiving the train's velocity near the front and near the back -- as if collision against it is unrotated. This particular train is longer than wide.)