System Agent Stats

[Romek]

You'd be able to disable the board, in that case.

 

[Hatebreeder]

You'd be able to disable the board, in that case.

That would kill the use of leaderboard, no?

Meh, it's OK like this, as it really is for debugging single objects. if you really want to keep track of everything at once, then there is no use of using a handle counter?

I don't know, i find it quite useless to have all handles shown at once.

 

[Romek]

> That would kill the use of leaderboard, no?
No, no. You'd disable the leaderboard if it got in the way.
Don't forget that not everyone who would use this would have a leaderboard already.
(Besides, the leaderboard would be removed when not in debug mode or for final releases).

I personally wouldn't use anything like this unless it had a leaderboard or a multiboard. People use systems like this to see if they're leaking where they didn't know. If you knew what had problems, you wouldn't use a system like this to... Check if something had problems.

It'd be a nuisance if this doesn't get a leaderboard. I personally think it makes this that much more useful. I would even use this if it had a leaderboard, and it would save me the trouble of writing my own version of this to use privately (I wouldn't submit it, because I'd basically be 'copying' J4Ls system - And I hate it when people do that, hence I won't do it myself).

> I don't know, i find it quite useless to have all handles shown at once.
Not all. Total, and top 5 individual ones, possibly.

 

[Renendaru]

I think it'd be cool if you could use a Timer window and set its time to the number of handles, or even have separate Timer Windows for each statistic.

 

[Jesus4Lyf]

>I would even use this if it had a leaderboard
If it really matters that much, I can add it...

But I have to warn you, it is O(n) complexity over every handle id that exists in the game, and enabling the leaderboard will screw up the "incremented by this much since last scan" values, and leak significantly.

I'll get around to it.

 

[Narks]

Instead of using a leaderboard, you could make an -autocount on command, which periodically does what -count currently does. Maybe you can set the interval?


This is still very cool though.

 

[weaaddar]

So I'm reading over this and I'm not really sure why you need to use the preload file. Your abusing a simmilar trick that I abuse here

Since all you really want to do is shove types into the hashtable you can use this much simpler function without the obtuse typecasting::

function SaveHandleHandle(hashtable ht,int pk,int ck,handle h)->boolean
{
     integer hid = GetHandleId(h);
     fogstate f  = ConvertFogState(hid);
     return SaveFogStateHandle(ht,pk,ck,f);
}

And trollbrain, while I never actually thought about it I suppose commonW if it ever gets finished could be used as a handle counter.

 

[weaaddar]

Blizzard borked this with the 1.24c patch.

 

[Troll-Brain]

Because of the preloader ?

 

[Jesus4Lyf]

Because of the return bug (finally).
But I can use fogstate typecasting to fix it. :thup:

 

[Troll-Brain]

The return bug is a phoenix or an hydra, he will never die

 

[weaaddar]

Yep. I'm glad I found it, here is the one lineable return bug exploitor SaveHandleHandle which will inline (because obviously performance is a requirement on something that loops through all handles...)

Spoiler

function SaveHandleHandle takes hashtable ht,integer pk,integer ck,handle value returns boolean
return SaveFogStateHandle(ht,pk,ck,ConvertFogState(GetHandleId(value)))
endfunction

This works because of the way blizz stores thing in hashtable. Each type is stored in a different bucket so you can do something like::

Spoiler

SaveInteger(table,0,0,1);
SaveReal(table,0,0,94);

And pull out different results by asking for a real or integer at index 0,0. This is how my CommonW was going to work, I was going to store the object in a hashtable, and its type which would allow easy typecasting and asking of its type, but since CJass is considered the devil, I stopped working on it.

 

[Troll-Brain]

Personally, i'm not against cJass, i just don't like the C syntax, i don't find it really readable, but i love cJass' defines.
In my opinion, cJass is perfect for personal codes but quite ugly for shared ones.
You can make really silly things with vJass, but cJass has more potential to do them, in wrong hands

 

[13lade619]

offtopic.. sry.

The return bug is a phoenix or an hydra, he will never die

remember the rumors that warcraf 4 was codenamed project Hydra .

 

[weaaddar]

I have nothing against the C syntax. I actually like it. my biggest beef was the fact that you can type Vjass and Cjass right on top of each other without it freaking out. I feel Vex went the right way when he designed Zinc by forcing you to stay in Zinc the whole way through. I'd like Zinc to introduce a "nothing"/"void" keyword so that its possible to get rid of the keyword method and function, and to fix this horrible looking thing::

interface MyInterface
{
     method someMethod() = null;
}

What the hell does that mean? Assign someMethod to null? So now someMethod is assignable?
This would be so much better in my mind

interface MyInterface
{
    SomeMethod()->nothing = nothing;
}

Which at least is cleaner.

 

[Narks]

hey does this work anymore?

 

[Jesus4Lyf]

hey does this work anymore?

Yep. It does.
But I somewhat question how valuable it is.

 

[Narks]

I imported it into my map, and it works, but it keeps saying that there are 0 agents all the time, but still lists all the other agents or handles or w/e. I'm not sure if it's working properly.

 

[Jesus4Lyf]

Should work fine. It's a straight copy/paste import now (off memory), and only runs in debug mode. Have you updated it (from the first post)?

 

[Narks]

Documentation is misleading... I thought I still needed to import that preloader file.

Anyway, yeah, it works, but I get this:

Zero agents?