Update geometry related functions

[kymckay]

When merged this pull request will:

• Add a new getArea function to return the array form of any area construct
• Function randPosArea now accepts areas in the form of an array or location (same as the inArea command)
• Function taskSearchArea now accepts areas in the form of an array or location
• Function inArea deprecated in favour of inArea command

I'm not sure if CBA has a deprecation system we might want to use for the inArea function now that there's an equivalent engine command

[jonpas]

Yes there is a "system", reference this. Basically move it into a backwards compat file so it doesn't get included in documentation and prints warnings. CBA does not remove deprecated functions like we do in ACE3.

[commy2]

Suggestion for fnc_randPosArea.
Can't we simply make it create a random position with fnc_randPos and if the result is not inArea, then we discard it and try again?

[commy2]

[[10,10], 0] call cba_fnc_randPos

Before:
[10,10]

After:
[10,10,0]

While getPos supports 2d positions, the return value is always a 3d position. That is not true for our function.

I have no idea why they chose to make this the alternative syntax of getPos. It is completely different from the original purpose.

[commy2]

Has to be a single line. The format should probably be in L:09 f anyway though

[kymckay]

Can't we simply make it create a random position with fnc_randPos and if the result is not inArea, then we discard it and try again?

While rejection sampling would work, the reason I originally wrote this function was to avoid that since it requires looping an indeterminate amount of times. This way you're guaranteed to get a position in a single shot.

[commy2]

But that "single shot" could take way longer on average. I guess it also depends on the marker shape. For circles and hexagons it will on average be faster, but it could take a long time on narrow rectangles and ellipses.

[kymckay]

My thinking is that you'd use the function for a guaranteed result, whereas if you wanted to use rejection sampling for possibly quicker execution you'd just write it manually with while and getPos

[kymckay]

Trying to figure out if there's a way to optimise this part of randPosArea:

        private _rho = random (4*(_a + _b));

        private _x1 = (_rho min _2a);
        private _y1 = ((_rho - _x1) min _2b) max 0;
        private _x2 = ((_rho - _x1 - _y1) min _2a) max 0;
        private _y2 = ((_rho - _x1 - _y1 - _x2) min _2b) max 0;
        _posVector = [(_x1 - _x2) - _a, (_y1 - _y2) - _b, 0];

It's used to uniformly find a position along the perimeter of a rectangle, but I feel like there must some way to condense it down.

[Killswitch00]

This won't be compileFinal:ed, which, if I'm not mistaken, CfgFunctions functions are.

Also, there's a deprecation macro.

[commy2]

That's a general problem with this file and should probably be fixed in another PR

[Killswitch00]

Indeed, so we should not add to the problem.

Instead, change the old function file and keep its CfgFunctions entry.

[commy2]

The problem with that is, that the function will appear in the config functions viewer. Deprecated functions shouldn't appear there in my opinion.