I have an issue that seems to bear some relation to what is described in this thread.
For a project I am currently working on, I have a configuration where I can enable freezing of 'updates' to move variables immediately after a c_move call for the applicable move mode. Essentially, I can single step c_move calls using the [ enter ] key. (I'm not using freeze_mode or breakpoints to do this. ) So the sequence might be:
calculate movement
c_move
freeze next calculate movement and next c_move until [ enter ] key press
look around and look at debug values
[ enter ] key
repeat
player: defined as the moving entity of interest with polygon flag set to off
poly-structure: defined as a static non-moving entity such as building, surface to move on, etc. with polygon flag set to on
It seems, for me, there is a bit of c_move indecisiveness, when the player moves over certain types of poly-structures. It appears that the player 'flickers briefly' between an 'old position' (or maybe an undetermined position A) and a 'new position' (or maybe an undetermined position B) when c_move is invoked 'on' those types of poly-structures. It appears the greater the distance c_moved the more obvious or prenounced the issue becomes. Note that in the scripts, there are checks for collisions, and it seems that the issue even occurs when there are no direct collisions with the applicable poly-structure entity. Perhaps, there are merely mistakes in the scripting, but I have spent hours upon hours trying to determine the cause of the issue. If it is not a scripting mistake, then my current substandard, half-baked theory is that the issue occurs primarily when the player penetrates the bounding box area of the applicable poly-structure to deeply, like when the player walks a good distance into a poly-structure house, and perhaps when this 'deep' bounding box penetration occurs, there might be a built-in, hidden mechanism that tries to push or move the player back out or correct the penetration. (If I recall correctly, I think that I directly observed and/or read about similar mechanisms.) If such mechanisms exist, while they might be practical in some instances, there may very well be instances where they should be 'disabled'.