Fixes some instances of groups shooting into air

[nk3nny]

Fixes Suppress, Flank and Assault group level actions -- where sometimes Units would shoot into air
Fixes units shooting UGL sometimes not partaking in assault.

[rekterakathom]

In other functions, instead of 'fixing' the height to 0.5, they just exit. Would that be a smarter thing to do here too? If a target is over 6 meters from the ground, what does it help to assume the height to be 0.5 instead and miss the target by at least 5.5 meters?

[jokoho48]

In other functions, instead of 'fixing' the height to 0.5, they just exit. Would that be a smarter thing to do here too? If a target is over 6 meters from the ground, what does it help to assume the height to be 0.5 instead and miss the target by at least 5.5 meters?

no, that would filter out the functions early as _target can be an "expected" position of a target as those are often Extrapolated based on last seen velocity

[rekterakathom]

In other functions, instead of 'fixing' the height to 0.5, they just exit. Would that be a smarter thing to do here too? If a target is over 6 meters from the ground, what does it help to assume the height to be 0.5 instead and miss the target by at least 5.5 meters?

no, that would filter out the functions early as _target can be an "expected" position of a target as those are often Extrapolated based on last seen velocity

I can see how that's useful, but can we be certain that it is always just extrapolation errors? What if a target really is 6 meters (or more) above ground? In my humble opinion, it's better to sometimes not act even if a target is valid (extrapolation error) than to always act at the risk of unintended behaviour (shooting at least 5.5 meters below an actual target).

For example, could these checks run for air vehicles, paratroopers or infantry rappelling from a helicopter? If yes, then this would result in basically malfunctioning AI.

And like I said, in other functions, this exact same condition is used to exit, rather than to just correct the position.

• fnc_tacticsContact L120
• Used to be the case in fnc_brainEngage L70 but it was changed last week.

[rautamiekka]

it's better to sometimes not act even if a target is valid (extrapolation error) than to always act at the risk of unintended behaviour (shooting at least 5.5 meters below an actual target).

Inclined to agree that in terms of realism (and thus less immersion-breaking) it's better to ignore in this case than shoot God knows where; frozen from fear, a broken down gun, an impaired vision ... Any of just those would be a better flimsy excuse than shoot meters off the mark and claim a huge "Can't shoot straight" problem.

^ Still, if someone wanna play 'Arma', they can quickly disregard such immersion breakage if the mission still functions as intended, so I'm fine with either.

[nk3nny]

Nice catch with TacticsContact. I'll update that too! As @jokoho48 notes the position passed to these functions are usually derived from getHideFrom. This is the position the AI thinks the enemy is occupying, but it unfortunately sometimes reads far above ground.

The benefit is that it gives, what I call in the design philosophy, the AI a degree of fuzzy predictability. The AI seems to be affected by the fog of war. The AI is operating not on perfect knowledge. And so on. You don't have to participate in many platoon level engagements with players, particularly PvP, to experience a lot of fire expended uselessly. ;)

that said: I wouldn't be opposed with a function to replace our use of getHideFrom with one that allows the option of giving the AI perfect getPosATL knoweldge(!).