Small Code Snippets

[Berb]

I can't really say, because I don't really have any idea where this notion of adding extra zeroes increases the precision of a floating point.

 

[Magtheridon96]

It's not about adding zeroes, you add any digit (maybe).
With more digits, the Jass Interpreter would probably assume that accuracy is more important for a certain real. I don't know why it doesn't assume that 4 or 5 digit places indicates the need for accuracy. Why 9? Maybe it was that n00b coder's lucky number or something ;o (lol)

 

[Berb]

But how can you prove that? How do we know that our timer is really off by 0.0000001 seconds? How do you know timers even maintain a strict accuracy based on the value of the real? If the coders were noob I'd assume that they would assume that all timers should have the same accuracy, because that would be easiest.

 

[Magtheridon96]

I don't know, that's why I would test things out ;/
I wonder if timers just execute things multiple times at once when the given interval is less than a certain value :/ (meaning that it would execute a code twice every 0.001 seconds instead of doing it once every 0.0005 seconds.)
I can't think of ways to test that >:O

 

[Berb]

It just seems rather gimmicky that some reals are more precise than others. I'm pretty sure they're all either half-precision or single-precision (floats) values.

 

[Nestharus]

It just seems rather gimmicky that some reals are more precise than others. I'm pretty sure they're all either half-precision or single-precision (floats) values.

That's exactly how it works >.>

.031250000 or w/e => .03125.
.03125 => .03125001 or something crazy like that.


I have tested it and if you add 0s to the end, it will keep its accuracy. Don't ask me why, JASS is just weird. Reals are not floats, they are crappy tripped out floats ;o.

 

[Berb]

How did you determine that it was in fact 0.03125001?

 

[Nestharus]

0.03125001?

I displayed it with R2SW? If isn't .03125001, then it might also be .0312500001. I'm doing it off of memory here :\. I tested this a very long time ago, heh.

 

[Berb]

Wow I didn't know the WC3 API had a function that printed reals to a certain precision.

 

[Bribe]

The "unit leak" will occur whether the unit decayed, was removed or exploded.

For dummies, the death time should be set to 0 and set to "can't raise, does not decay".

 

[Troll-Brain]

It seems that i'm not the only one who thinks that timer periods use the first 8 numbers under the decimal.
I i believe my old test, the period was even fucked up with 9 numbers or more under the decimal.

I mean this should be investigated.

 

[Berb]

Under the decimal you mean?

If you have 10 numbers under the decimal you've destroyed your precision. Even if they are 0s. A number like 0.0123456789 printed out as 0.018 or something weird like that. For 0.0123000000 it came out to a similar result, 0.018 or something weird like that.

 

[Troll-Brain]

Under the decimal you mean?

Yep, fixed.

If you have 10 numbers under the decimal you've destroyed your precision. Even if they are 0s. A number like 0.0123456789 printed out as 0.018 or something weird like that. For 0.0123000000 it came out to a similar result, 0.018 or something weird like that.

I'm not only talking about reals in general, but the period of timers which seem to don't have the same behavior (8 numbers)

 

[Berb]

So the precision of the timer itself?

 

[Troll-Brain]

Yes.

 

[Alain.Mark]

just a simple function that checks how far a point is from a line segment

	function DistanceFromLineSegment takes real px, real py, real x1, real y1, real x2, real y2 returns real
		local real r =SquareRoot((x2-x1)*(x2-x1)+(y2-y1)*(y2-y1))
		local real a1=Atan2(y2-y1,x2-x1)
		local real a2=Atan2(px-x1,py-y1)
        if(a1>a2)then
            return Sin(a1-a2)*r
        endif
		return Sin(a2-a1)*r
	endfunction

 

[Troll-Brain]

Heh, you have just broken a chain of off-topic !

 

[Alain.Mark]

lol, but i broke it legally.

 

[The Reborn Devil]

What is this? Codes in the "Small Code Snippets" thread?!

BLASPHEMY!

 

[Troll-Brain]

Exactly.

What will be the next step, serious topics in the off-topic section ?!

 

[Berb]

Everything posted on this forum should inherently be in

 tags.

 

[The Reborn Devil]

Exactly.

What will be the next step, serious topics in the off-topic section ?!

That would be the end of the world! D:

Berb, you like?

 

[Alain.Mark]

// that would cause a syntax error, double "/" to be safe

 

[Magtheridon96]

I'm going to save this thread now.

library IsUnitDead

    function IsUnitDead takes unit u returns boolean
        return IsUnitType(u, UNIT_TYPE_DEAD) and GetUnitTypeId(u) != 0
    endfunction

    function UnitAlive takes unit u returns boolean
        return not IsUnitType(u, UNIT_TYPE_DEAD) and GetUnitTypeId(u) != 0
    endfunction

endlibrary

This is a solid library.
No one should ever make the mistake of only using IsUnitType(u, UNIT_TYPE_DEAD) anymore.
We really need a standard way of checking whether a unit is dead or not, so why not have a library for it?

 

[watermelon_1234]

    function IsUnitDead takes unit u returns boolean
        return IsUnitType(u, UNIT_TYPE_DEAD) @and@ GetUnitTypeId(u) @!=@ 0
    endfunction

You might as well as make UnitAlive be not IsUnitDead too.

 

[overcold_ice]

@mag:

should be like this:

library IsUnitAlive
    
    function IsUnitAlive takes unit u returns boolean
        return not IsUnitType(u, UNIT_TYPE_DEAD) and GetUnitTypeId(u) != 0
    endfunction
    
    function IsUnitDead takes unit u returns boolean
        return IsUnitType(u, UNIT_TYPE_DEAD) and GetUnitTypeId(u) != 0
    endfunction
    
    function IsUnitDeadEx takes unit u returns boolean
        return not IsUnitAlive(u)
    endfunction
    
    function IsUnitRemoved takes unit u returns boolean
        return GetUnitTypeId(u) == 0
    endfunction
endlibrary

or just simply:

library IsUnitAlive
    
    function IsUnitAlive takes unit u returns boolean
        return not IsUnitType(u, UNIT_TYPE_DEAD) and GetUnitTypeId(u) != 0
    endfunction
    
    function IsUnitDead takes unit u returns boolean
        return not IsUnitAlive(u)
    endfunction
endlibrary

 

[Bribe]

IsUnitDead will return false for removed units, so much for bug fixing.

Honestly 99% of the time you just need to check IsUnitType, UNIT_TYPE_DEAD. The other 1% is when you are storing non-hero units for prolonged periods.

 

[Magtheridon96]

It's not supposed to return true for removed units :/
What if you had a spell that only affects dead units?
Maybe we can return null or we can crash the thread =p

 

[Bribe]

If that was the case, then all you need to do, Mag, is to check IsUnitType, UNIT_TYPE_DEAD

 

[Bribe]

Considering approving:

DistanceFromLineSegment: http://www.hiveworkshop.com/forums/2098605-post716.html

The reason why I am considering it still is because the description could be more clear (variable names for example). There is also similar functionality in the AdvancedLoc library by Dirac.

UnitPropWindow: http://www.hiveworkshop.com/forums/2052422-post676.html

The reason why I am still considering it is because it's my resource and there's been no comment on it. I don't approve just because I like it.

 

[Magtheridon96]

function Xor takes boolean a, boolean b returns boolean
    return a != b
endfunction

function Nand takes boolean a, boolean b returns boolean
    return not (a and b)
endfunction

function Xnor takes boolean a, boolean b returns boolean
    return a == b
endfunction

function Nor takes boolean a, boolean b returns boolean
    return not (a or b)
endfunction

Wrappers and functions that will never be used by any sane man are what makes the world go around :3

 

[Bribe]

function Nor takes boolean a, boolean b returns boolean
    return not (a or b)
endfunction

Other than that, good reference table for people looking to do certain channels.

 

[Magtheridon96]

/*******************************************
*
*   UnitRallyPoint
*   v1.0.0.0
*   By Magtheridon96
*
*   - Provides GetUnitRallyPointX and GetUnitRallyPointY
*
*   API:
*   ----
*
*       - function GetUnitRallyPointX takes unit u returns real
*           - Returns the x coordinate of the rally point for a given unit
*       - function GetUnitRallyPointY takes unit u returns real
*           - Returns the y coordinate of the rally point for a given unit
*
*******************************************/
library UnitRallyPoint
    
    globals
        private location loc
    endglobals
    
    function GetUnitRallyPointX takes unit u returns real
        local real x
        set loc = GetUnitRallyPoint(u)
        set x = GetLocationX(loc)
        call RemoveLocation(loc)
        return x
    endfunction
    
    function GetUnitRallyPointY takes unit u returns real
        local real y
        set loc = GetUnitRallyPoint(u)
        set y = GetLocationY(loc)
        call RemoveLocation(loc)
        return y
    endfunction
    
endlibrary

Moved from the Jass section.

 

[Berb]

library Comparable
interface comparable
    method compare takes comparable c returns integer
endinterface
endlibrary

 

[Bribe]

What

 

[Magtheridon96]

I spent 10 minutes trying to figure what Berb implied when he posted that :|

 

[Nestharus]

I spent 10 minutes trying to figure what Berb implied when he posted that :|

oh, I knew what he meant immediately ; ).

 

[Bribe]

But it is completely useless

 

[Berb]

Comparisons are useless?

 

[The Reborn Devil]

function Xor takes boolean a, boolean b returns boolean
    return a != b
endfunction

Should be this:

function Xor takes boolean a, boolean b returns boolean
    return (a or b) and (a != b)
endfunction

False and True -> True
True and False -> True
True and True -> False
False and False -> False

 

[Garfield1337]

The one Magtheridon posted is fine, check all the cases yourself.

 

[Berb]

No, Maghtheridon's is right.

His yields the same results yours does.

 

[The Reborn Devil]

The one Magtheridon posted is fine, check all the cases yourself.

No, Maghtheridon's is right.

His yields the same results yours does.

Are you kidding me?

Magtheridon's code will return the wrong answer if both a and b are false.
(False != False) = true.

However, False Xor False should return false. Xor stands for Exclusive Or. It returns true iff one of the arguments are true.

Edit: See below post

 

[WaterKnight]

(false != false) == false

 

[The Reborn Devil]

Wait, wat, nvm

I mixed up == and != (or something. I dunno :>)

I think my brain tried to compensate for "a != b" being too easy. Kinda made Xor seem redundant. (Edit: It is redundant in fact)

Edit: However that means Mag's code snippet is useless, and using it is not recommended at all, as it will degrade the performance of your code.

 

[WaterKnight]

This seems inline-friendly.

 

[The Reborn Devil]

This seems inline-friendly.

Only if you use Jass NewGen Pack. Not all do, and as a result people should stay away from it. It also makes your code longer, both before compiling and after (the compiler sucks).

 

[WaterKnight]

It gets compressed again when going through optimizer. Having shorter code does not mean that it would be better readable or structured. If it was for the ultimate performance, I would not use any foreign code. I see the posting of snippets more like a basic reference how things can be solved, you just learned how xor can be written for example.

 

[The Reborn Devil]

It gets compressed again when going through optimizer.

The compiler I spoke of also includes the optimizer. It sucks.

For example, the following code:

function Xor takes boolean a, boolean b returns boolean
    return a != b
endfunction
    
function HaiThar takes nothing returns nothing
    if Xor(true, false) then
        call BJDebugMsg("WOOO")
    endif
endfunction

Is turned into this:

function Xor takes boolean a,boolean b returns boolean
    return a != b //No, this function is not removed
endfunction
    
function HaiThar takes nothing returns nothing
    if ((true ) != ( false)) then // INLINED!!
        call BJDebugMsg("WOOO")
    endif
endfunction

Which is much longer (and also slower) than only having this:

function HaiThar takes nothing returns nothing
    if true != false then
        call BJDebugMsg("WOOO")
    endif
endfunction

Having shorter code does not mean that it would be better readable or structured.

It could be confusing. In this case it would confuse some people who don't know what Xor is, and they would have to look up the function.

If it was for the ultimate performance, I would not use any foreign code.

Neither would I, but that is personal preferences. I prefer making everything myself, but some people will just include shit they see being posted into their own code.

you just learned how xor can be written for example.

I already knew what Xor was and what it meant, but I was made aware of it being utterly redundant when dealing with booleans.

Unnecessarily long post is long.

 

[Magtheridon96]

They were only for future reference