Global Group standard?

[Troll-Brain]

I don't use structs all the time ?

 

[Jesus4Lyf]

Troll's modified GU is not good for detecting double frees, actually.

You understand? :thup:

 

[Troll-Brain]

No, since it's true for you but false for me.
It's just an opinion anyway, we should stop to talk about this.

 

[Romek]

I don't see you removing that dummy unit from the group, Troll-Brain.

 

[Troll-Brain]

I don't see you removing that dummy unit from the group, Troll-Brain.

GroupClear on release.

 

[Jesus4Lyf]

Romek is right - it is added after the group clear? <_<

Should be removed when the group is returned, shouldn't it?

 

[Troll-Brain]

The dummy exist to know if the group is released or not, but for some odd reason i've removed one debug line inside the function NewGroup :banghead:
Now, it's fixed.

 

[Vexorian]

I OPPOSE tO standArds in jaSS

anyway, this thing is silly, adding 1 line of code is easier than adding 5 lines of code. Besides, The name GROUP is among the lamest in history.

(PS. My background is mapper, I mapped in GUI for about 4-5 yrs before looking at JASS.)

Sorry, but wasting 4.-5 years in GUI makes you not a mapper guy but a mentally-challenged one.


GroupUtils uses hashtables now, if you want I can make dusk make it so the double free protection is only available when a constant is set to true, or maybe debug mode. I can't think of a good reason to release a group that was not allocated by NewGroup. Besides, recycling shouldn't be in use at all anyway...

 

[Troll-Brain]

I can't think of a good reason to release a group that was not allocated by NewGroup.

You're probably right, i was thinking about gui functions which returns group but i had forgotten these ones are pretty useless.

It's cleaner now, but anyway i still will do all the checks and hashtable usage only on not debug mode.

 

[Jesus4Lyf]

Sorry, but wasting 4.-5 years in GUI makes you not a mapper guy but a mentally-challenged one.

Hahaha... Yea. Well, I was 13-->16 yrs old (and not a programmer). It was fun.

>GroupUtils uses hashtables now, if you want I can make dusk make it so the double free protection is only available when a constant is set to true, or maybe debug mode.
What people want is a a 5 line function when debug mode is off. GroupUtils can't do that without breaking backwards compatability, I'm pretty sure. It just happens that it is built for extreme safety, to the point where the final map's efficiency is reduced instead of just in debug mode. Not saying it's wrong, I just wouldn't use it myself.

>I can't think of a good reason to release a group that was not allocated by NewGroup.
I can't think of a reason to restrict it.

>Besides, recycling shouldn't be in use at all anyway...
People said that about attaching to timers, iirc. Why not recycle?

 

[Troll-Brain]

I think Vexorian is talking about groups recycling, because of possible ghost units for groups used not instantly like GROUP_ENUM.

Finally i see one good reason why to forbid the release of a group which wasn't in the stack.
If you want to create a private group which should never de recycled, use CreateGroup instead of NewGroup, and you will protect from yourself (still a silly case though)

 

[Jesus4Lyf]

There is a better reason (kinda) as Nestharus found out - you can theoretically use offset attachment if you can't free groups that didn't come from NewGroup.