Trigger leak question

[lovexylitol]

I have a map under development and found the following code leaks.

function xylitolGetUnitNumber takes player p, integer u returns integer
    local integer i
    set udg_xGroup = GetUnitsOfPlayerAndTypeId(p, u)
    set i = CountUnitsInGroup(udg_xGroup)
    call DestroyGroup(udg_xGroup)
    return i
endfunction

To be exact, the handle count is increasing once per time when I
call this code. I called this code once every 0.01, and the handle count
after an hour was about in the order of several hundred thousands, and
memory war3.exe was over 600mb; therefore the code does leak.

This function is supposed to get a count(i) of a type of unit(u) owned by a player(p). I have also tried using CountLivingPlayerUnitsOfTypeId, and got
the same results. Is there a way to do this? Or is it supposed to leak?

Thank you for your replies.

 

[Troll-Brain]

I don't see any leak in your code.
Also if you really need to check it every 0.01 s (i doubt really) inline the functions you actually use.

 

[lovexylitol]

Oh the 0.01s was done to see if it really leaks.
(In the real map I only use it about 12*5 times every minute, so its not really a problem, but still for perfection)
It was done on an empty map with several(a little more than several) footmen.
When the body of the function is replaced with
a simple 'return 1' the counter does not increase.

It is a mystery to me why this leaks.

 

[Romek]

All the GroupEnum functions can have the same effects as leaking can if used at very high frequencies. The lag clears up though once the function's aren't called any more for a short period. (Unconfirmed, I read this somewhere)

Is it essential for you to use such a high frequency?

 

[phyrex1an]

Many blizzard.j functions have these kinds of leaks (in your case GetUnitsOfPlayerAndTypeId) , they never null local variables in blizzard.j.

Here is an alternative version:

globals
     group dummyGroup = CreateGroup()
endglobals
function CountUnitsOfTypeEnum takes nothing returns boolean
    if GetUnitTypeId(GetFilterUnit()) == bj_groupEnumTypeId then
         set bj_groupCountUnits = bj_groupCountUnits + 1
    endif
    return false
endfunction

function xylitolGetUnitNumber takes player p, integer u returns integer
    set bj_groupCountUnits = 0
    set  bj_groupEnumTypeId = u
    call GroupEnumUnitsOfPlayer(dummyGroup, p, Condition(CountUnitsOfTypeEnum))
    return bj_groupCountUnits
endfunction

 

[lovexylitol]

Everybody thank you for your replies. :thup:
I'll try that code.

 

[Troll-Brain]

Oh the 0.01s was done to see if it really leaks.
(In the real map I only use it about 12*5 times every minute, so its not really a problem, but still for perfection)
It was done on an empty map with several(a little more than several) footmen.
When the body of the function is replaced with
a simple 'return 1' the counter does not increase.

It is a mystery to me why this leaks.

Try to add these line :

set udg_xGroup = null

But it should'nt do anything.

Oh also use a bigger period like 0.1 s, maybe the memory have not the time to be free, that's not a leak but the memory grow up like one, while you don't stop the execution of the function.

 

[lovexylitol]

Hmm, I tried to give reputation, but not sure if it worked
(Clicked the add to sombody's reputation and submited the form)
(This was my first post here)

Anyways thank you!

 

[AceHart]

Anyone ever tried "GetPlayerTechCount(<player>, <unit id>, true)" ?