Multiple instances of one script or one monitoring global array?

[lendrom]

Sorry for non-informative topic title but I didn't know how to say what I want to ask in such a short sentence.
Suppose I'm writting a script which will run on a lot of units and the units which are affected by the script change many times durring the mission. I want to ask which is beter if it comes to performance:

Starting a new script that monitores a specified condition for every unit:
[_unit, _cover] exec checkForMovement.sqs

Code: [Select]

_unit = _this select 0
_cover = _this select 1
@ (not(currentCommand _unit in acceptableCommands)) OR not(alive _unit)
_unit setunitpos "auto"
?_unit HideBehindScripted false
IgnoreList = IgnoreList - [_cover]
exit

Or having one script that is running for the whole mission and has monitores every unit from a global array to which I add the unit:
UnitsToMonitor = UnitsToMonitor + [[_unit, _cover]]

Code: [Select]

acceptableCommands = ["STOP", "WAIT", "FIRE"]
#mainloop
;sending global variable to local to avoid unwanted interference from global space
_unitsToMonitor = UnitsToMonitor
_numOfUnits = count _unitsToMonitor
~0.01
? _numOfUnits == 0: goto "mainloop"

#subloop
_i = 0
_currentArray = _unitsToMonitor select _i
_currentUnit = _currentArray select 0
_currentCover = _currentArray  select 1
? not(currentCommand _currentUnit in acceptableCommands)) OR not(alive _currentUnit): _currentUnit setunitpos "auto"; _currentUnit HideBehindScripted false; IgnoreList = IgnoreList - [_currentCover]; unitsToMonitor = unitsToMonitor - [_currentArray]
~0.01
_i = _i + 1
? _i < numOfUnits: goto "subloop"
goto "mainloop"

The presented scripts can have some bugs because I wrote them straight at the forums but I wanted to ask about the principle. Is it better to have multiple instances of the first script or one script that monitors global array and runs during the whole mission?

[Mandoble]

Basically you want to do the same in two ways, first in parallel and second in sequence. At first glance I would say the parallel mode would have a bigger impact in the performance if you have many many units, but the constant loop with only 0.01 delay time will be worse, and I understand that unless you have very few units you cannot increase that delay in the loop (#subloop).

[lendrom]

Thanks for your answer. I'm glad my explanation of the problem was understandable. Yes, I fear that in both cases the performance hit could be big but the I need to almost instantly know if the condition is met. And what I was thinking was if one quite fast loop would be better than many '@' conditions.

[Mr.Peanut]

I don't know if it has changed for ArmA, but in OFP using an @ was better for performance than a very fast looping script, as long as the @ condition was simple which yours is.

If you are really worried about performance you should use sqf style scripting and the execVM command instead of sqs.

[lendrom]

That's interesting Mr. Peanut. Thanks for reply.
And about sqf. I'm trying to get used to it since OFP but I think my brain just works in a different way. Sqs is simply much more intuitive for me. And the only experience with programing I have is using PROLOG which is much different than enything else and is absolutely performance-unfriendly .