[System] Automatic Memory Leak Destroyer

[Element of Water]

For god's sake. I'm not saying it will inline. I'm saying I INLINED THE CATCH FUNCTIONS USING TEXTMACROS.

Why the hell can you never understand ANYTHING properly?

 

[MapperMalte]

Ok. But that doesn't help.

 

[MapperMalte]

Ok, I updated the code a bit.

 

[Troll-Brain]

I think Element Of Water want to mean that :

private function ASE takes location where, string modelName returns nothing
        //! runtextmacro CatchLocation("where")
        //! runtextmacro CatchEffect("GetLastCreatedEffectBJ()")
endfunction

instead of that :

private function ASE takes location where, string modelName returns nothing
        call CatchLocation(where)
        call CatchEffect(GetLastCreatedEffectBJ())
endfunction

But hooks are already .evaluated, it's not a simple call of function, which is slower, so in this case i dunno if it really does matter if you use inline the functions with textmacros or call them.

 

[Element of Water]

Hooks are .evaluated? Wtf? Vex has got hooks all wrong...

 

[azlier]

They are .evaluated to allow a function to have multiple other functions hooked to it. Or something like that.

The current hooks are not good enough for my needs.

 

[Troll-Brain]

Hooks are .evaluated? Wtf? Vex has got hooks all wrong...

Probably because it's easier to make like this way, no need to manage the position of the function inside the script.
Hooks are not for speed freaks guys.

 

[Element of Water]

TBH I'd much prefer hooks if the functions they hooked were basically replaced by them, so you have to call the function, but at least you get what it returns and it's much faster...

 

[Troll-Brain]

TBH I'd much prefer hooks if the functions they hooked were basically replaced by them, so you have to call the function, but at least you get what it returns and it's much faster...

But then it's define ?

 

[Element of Water]

Yeah it's define in cJass. But I like define in cJass...

 

[Troll-Brain]

Yeah it's define in cJass. But I like define in cJass...

I like it too, but i want mean that's not for the same purpose.

 

[Element of Water]

You can achieve the same thing with it + much more.

 

[HINDYhat]

I actually like this idea very much and I think it's a very neat way of using hooks. I haven't looked at the script yet, I just like the concept.

 

[aznricepuff]

I'll approve this, even though I personally wouldn't recommend its usage. But other people may want to use it.

 

[.Ragnaros]

This discussion is making me sick... really... I'm totally inside the opinion that this is a great system, GUIers often are too lazy to fix all memory leaks and this script cleans them very fast and without troubles, also saves me a lot of painful work. MapperMalte I must congratulate you because you made the most useful system I ever seen (in my opinion).

Thank you very much.

PS: I'm using it on my map and it dealt no damage to my GUI or JASS, for JASS scripters this system may look stupid but for GUIers it's useful.

 

[Troll-Brain]

You might want to know your system have minor leaks in the current state :
http://www.thehelper.net/forums/showthread.php?t=136608

 

[MapperMalte]

Well, I'm sure blizzard will fix that soon.
And I null all the things from the passer.
So it's probably still better than string leaks.

 

[Troll-Brain]

Well, I'm sure blizzard will fix that soon.

Let's me tell you the truth, your parents lie, Christmas father doesn't exist

Anyway it's easy to fix, and imho for a system which prevent leaks, leaks made by itself is kinda lame, even if it's minor.

 

[MapperMalte]

k, done.

 

[ShadowHunter291]

Gotta love the 4 page nerd fight.
Next time keep the useless banter for PMs and leave room for other comments on the program, please!

And FYI being a GUI user isn't always because your lazy.
As appealing as JASS was to me, it looked like brain melting jibberish until I took a class in C++. I had no idea what a memory leak was (other than being a flaw) or the difference between local and global variables.

And remember, many of the most historic games for Warcraft III were made in GUI, not JASS.

Examples: DoTA (initially), Footmen Frenzy, Wintermaul Wars, LoAP, NoTD (initially), etc.

JASS can be intimidating, and may snuff out a person's inspiration.

 

[Element of Water]

Gotta love the 4 page nerd fight.
Next time keep the useless banter for PMs and leave room for other comments on the program, please!

Discussions like that are criticisms of the script. They are the most useful comments, not like your "GUI rulez" comment.

 

[MapperMalte]

Update

Now stops giving very strange error messages if the number of caugth memory leaks gets too big.

 

[azlier]

Now we just need to somehow make a system that makes GUI leakless.

Maybe replacing Blizzard.j would do.

 

[MapperMalte]

How should I understand your post?

 

[PsycoMarauder]

When I try to protect some integer variables via call ProtectVariable(xxxx), jasshelper comes up with an error saying "cannot convert integer to handle"
Does this mean that I can't stop your system from deleting the information from my integer variables? Or are the integer variables safe, thus theres no reason to try and protect them?

>>Nevermind<<

 

[MapperMalte]

It doesn't affect integers at all, because integers don't cause memory leaks.
You only have to protect groups, effects and locations you want to keep longer than CLEAN_UP_INTERVAL seconds.

 

[azlier]

>How should I understand your post?

What I'm saying is that someone should invent leakless GUI, one way or another. It hasn't been done yet.

 

[YourNameHere]

>How should I understand your post?

What I'm saying is that someone should invent leakless GUI, one way or another. It hasn't been done yet.

Doesn't that cJass antiBJ thingy do that?

 

[azlier]

Never heard of such a thing. Linky?

 

[YourNameHere]

Well ok, it doesn't really make it leakless, but it removes all usage of BJ functions.
It's basicly a huge list of defines which replace all BJ calls with native ones.
Linky

 

[MapperMalte]

Yeah, BJs are evil! They are the reason for leaks!
And replacing BJs wont make code useless that uses things like last created unit!
Dude, no matter if you use BJs or not.

 

[YourNameHere]

Yeah, BJs are evil! They are the reason for leaks!
And replacing BJs wont make code useless that uses things like last created unit!
Dude, no matter if you use BJs or not.

lol. It's known that BJ's are in fact slower than using natives. Why would u need last created unit? Just store it in a varaible yourself if you really need it.
The only thing where I actually use BJ's are those multiboard thingies where you don't really have an alternative.

 

[MapperMalte]

Yes, BJs are slightly slower.
But they don't cause memory leaks just because they are BJs.

 

[YourNameHere]

I didn't try to prove anything else.
A mix of that antiBJ thingy and your system would make GUI alot more faster.

 

[MapperMalte]

Well, the BJ thing makes it slightly faster. BJs are nothing but function wrappers that add things like defining last created unit.
So it won't make gui much faster, but probably it sums so it'd be faster.

 

[azlier]

Many BJs also leak.

 

[elfian]

This is really useful, altho, who uses GUI for big projects now? I mean, after I learned JASS I stopped coding in GUI at all. No matter how simple the triggers are. But for example I had some old maps and projects that were in GUI and were leaking hard. And instead of bashing my head against the wall for hours I can just add this and everything is fixed. This "garbage system" is one of the most useful things on this site, come on guys, you can't say it's useless and stupid.

 

[Admiral Jack]

I think theres a problem in your Script:

    hook ForGroupBJ MemoryLeakHelper__FG // :D This should catch all GUI usages for groups.
    //hook ForGroup FG
    hook GroupPickRandomUnit CatchGroup
    hook CountUnitsInGroup CatchGroup

Its says it cannot Find the Function "CatchGroup" can you help me with this.

You are good Jass maker friend.....

 

[Magtheridon96]

This thing needs an update. Handle Tables? Really?

 

[PsycoMarauder]

*Bump for possible update? Super useful script.

 

[Bannar]

I'm still wondering if using such script provides more good or either kills the intelligence and responibility of user.

Is it so hard to destroy/remove leaks and write your OWN clear code? Even if you have troubles with that, with time you gonna learn it for sure. Making your script buggy, leaving everything for such stuff isn't good habbit and won't help you in future scripting.

 

[Magtheridon96]

This isn't useful. Anyone who uses vJass should know how to clear the leaks himself. Plus, in vJass, we don't have to worry about leaks.
We don't use locations (Except for a static one to find the Z of a point)
We don't use dynamic groups anymore. (Static ones for enumeration and FirstOfGroup loops to iterate)
We can easily clear special effects.

GUIers wouldn't use this.

 

[Bribe]

Well there is also the fact that this has a bigger learning curve than actually doing what you're supposed to with groups, locations, special effects, etc. by just removing them yourself.

It also adds a huge sum of KB size to your map. And makes everything perform a lot more slowly. Especially when you are adding a lot of effects on a rapid timer, or are a GUI user using lots of locations/groups per 0.03 seconds.

It requires NewGen and most people are strictly-WE users.

Why did this get approved in the first place. Just because it's "cool" that someone could do it? A JASS garbage collector is not practical unless it's built-in to the source code which Blizzard has made a point they have no intention of doing (demonstrated via the "bj_wantDestroyGroup variable").

 

[Zwiebelchen]

I don't know why this discussion about the legitimacy of garbage collectors is brought here? It's totally pointless. This thread is not about why this script should be useful or not. It's about the system itself. And it seems to work good, aside from some minor stuff that was already mentioned and fixed.

So far, there is no reason for me to think that this system should not be useful. Sure, Jassers won't need it and good GUI users shouldn't use it too ... and others suck at mapping anyway, but that doesn't matter. The system works and does what it is supposed to do. Wether you would use it or not doesn't matter.

 

[Bribe]

You are right, Zwiebelchen, if I were to say that practical usefulness is not a requirement for approval. In the Spells section if it is not useful we can just rate it low, but in the JASS section there isn't really such a system. That's one of the reasons I had been trying to push to merge the two sections but no one else seems to show interest in it.

 

[Dirac]

Wether you would use it or not doesn't matter.

Wether anyone would use it or not matters.

 

[Magtheridon96]

^true

I could make a system that creates a rect and moves it around for who knows what reason. It does it's job, but logically speaking, should it be approved? Hell no.

This system shouldn't destroy leaks, but instead, it should warn the user that he's leaking. GUI users would find that useful.

 

[Zwiebelchen]

This system shouldn't destroy leaks, but instead, it should warn the user that he's leaking. GUI users would find that useful.

Kinda like "give a man a fish and he can eat for one day. Give him a pole and he can eat every day."?
Seems legit.

 

[Magtheridon96]

^ Exactly :>

 

[Zwiebelchen]

^ Exactly :>

The problem is: most GUIers still prefer the fish over the pole.