Patch 1.24 - Return bug false positives.

08-12-2009, 03:38 PM

Please post your examples if you find any:

Does not work:

JASS:

//===========================================================================
private function Charge takes nothing returns boolean
    local unit target

    if IsUnitType(target, UNIT_TYPE_MECHANICAL) then
        return true        
    else
        call AddSpecialEffectTarget(FX_UNIT, target, "chest")
    endif

    return false
endfunction

Works:

JASS:

//===========================================================================
private function Charge takes nothing returns boolean
    local unit target

    if IsUnitType(target, UNIT_TYPE_MECHANICAL) then
        return true        
    else
        call AddSpecialEffectTarget(FX_UNIT, target, "chest")
        return false
    endif    
endfunction

Conclusion: if you have return in if you need a return in else.
(or remove else branch)

EDIT:
This seems to be the only error, at least in my maps.

Second solutions:

JASS:

//===========================================================================
private function Charge takes nothing returns boolean
    local unit target

    if IsUnitType(target, UNIT_TYPE_MECHANICAL) then
        return true        
    endif
    call AddSpecialEffectTarget(FX_UNIT, target, "chest")
    return false
endfunction

08-12-2009, 05:35 PM	#2
Alevice	is this 1.24a or 1.24b

08-12-2009, 05:36 PM	#3
TKF	To fix the return compile error you get when saving old maps again with codes, you just move all endifs after return to fix all compile errors? I'm right? Or is there more? ________________ Sry i got no examples, but is there also a problem with endfunction or is it the editor being wrong when i get compile errors upon saving?

08-12-2009, 05:36 PM	#4
Alevice	just usse latest jasshelper, and be done with it.

08-12-2009, 05:42 PM

cohadar

I use latest NewGen (but not latest jasshelper)
I got wc3 patch from BNET today.

Quote:

Originally Posted by Alevice

just usse latest jasshelper, and be done with it.

Slows things by a function call (not negligable for filters and conditions)

Besides blizzard has a tendency to fix errors once in a millenia.

This bug does not create compile errors, your map simple does not load.
Stupid blizz detects this as H2I function (idiotic algorithm)

08-12-2009, 08:02 PM

Vexorian

Quote:

This bug does not create compile errors,

Yes it does.

Quote:

your map simple does not load

which is war3.exe's way to tell you there is a compile error.

Quote:

Slows things by a function call (not negligable for filters and conditions)

In comparison to not running a map at all, an extra function call is incredibly irrelevant.

The thing is, that we won't have a retail patch with these things fully fixed in a while... I thought 1.24b fixed them, but it seems not. If so then the real problem here is that blizzard will keep changing the algorithm. And thus we'll have to figure things out with each release (i.e. on 1.24 there were so many variations of it that it was crazy). It is just not worth it.

Anyway, if you want to find the reasons for the bugs, just use the blizzard.j trick.

Edit: Only functions with multiple returns "slow down", the reality is that from now on, we'll just have to get rid of double returns, begin refactoring.

Let's ask again:

Quote:

is this 1.24a or 1.24b

08-12-2009, 08:27 PM

ToukoAozaki

Quote:

I thought 1.24b fixed them, but it seems not.

Would you be more specific?

08-12-2009, 09:16 PM	#8
Vexorian	This thread?

08-12-2009, 09:38 PM

Vexorian

... Actually:

Quote:

I got wc3 patch from BNET today.

From westfall ?

08-12-2009, 09:39 PM	#10
Alevice	hence my question

08-12-2009, 09:58 PM	#11
Mr_Saturn	His example compiles and runs in the new westfall patch.

08-13-2009, 06:14 AM

#12

cohadar

Quote:

Originally Posted by Vexorian

Anyway, if you want to find the reasons for the bugs, just use the blizzard.j trick.

I don't know, explain?

No I did not use westfall.

And it is not possible to use remove all functions with multiple returns.
(I muself don't plan on using a variable for return value of every filter I ever made.)

@Mr_Saturn
Are you saying that 1.24b has fixed the bug?

08-13-2009, 08:46 AM

#13

Karawasa

Quote:

Originally Posted by cohadar

@Mr_Saturn
Are you saying that 1.24b has fixed the bug?

Yes.

08-13-2009, 02:13 PM

#14

Vexorian

Quote:

And it is not possible to use remove all functions with multiple returns.
(I muself don't plan on using a variable for return value of every filter I ever made.)

Me neither, hence the reason I added that to jasshelper.

Anyway, is it confirmed everything is fixed in 1.24b? I'll open a new thread... and close this as it was basically a fad...