[Lua][vJass] New Table

[Dirac]

TableArray[array index][hashtable index]
The first parameter tops at 8190, the second one goes up to 2^32
You're just using it rong

 

[mckill2009]

Bribe posted it like TableArray[Parent][ChildIndex] in post #93...

 

[Dirac]

And he's correct, but the order you use them is irrelevant because you would reach the var in the same way anyways...
tb[1][2] = 3
tb[2][1] = 3
However the Parent can't go over 8190 because it's a struct array allocator, but the child is a hashtable index, so it can go beyond.
Anyways, i've never ever had to use TableArray anyways, you can always avoid it

 

[Laiev]

struct S extends array
     static TableArray ta
     /* Is basically 'the same' as */
     Table t

     method m ....
          set ta[this]['A000'] = 1
          set this.t['A000'] = 2

          if ta[this]['A000'] == 1 then
          endif
          if this.t['A000'] == 2 then
          endif
     endmethod
endstruct

 

[Magtheridon96]

Well, there's only a small difference between Table arrays and TableArrays

TableArrays need no initialization after setting them to TableArray[...]
Table arrays need you to check if the Table is equal to 0 or not before creating it ;o

TableArrays just make things cleaner

But you know what would be better?
A TableArray array 8)

struct Hash extends array
    
    private static constant integer DEFAULT_SIZE = 0x2000
    private static TableArray array H4X
    
    static method operator [] takes integer i returns TableArray
        if H4X[i] == 0 then
            set H4X[i] = TableArray[DEFAULT_SIZE]
        endif
        return H4X[i]
    endmethod

endstruct

Demo

// ...
set Hash[4680][1212]['A!8*'] = 9
// ...

And this would work 100% ^.^

 

[Bribe]

You could always store Tables within Tables and have syntax like tb[2424234][0xFEFE][22][-69][101010101]

Of course you pretty much never need that level of complexity.

@Dirac, not so much that 8192 is the "limit" of TableArrays, it's just typically the most someone will ever need.

To prevent clashes, it would be impractical to create a large enough TableArray to support handle ID's and rawcodes for the first index. Which is why I wrote on the first post "if you need such a large index, store a Table within a Table, or even a TableArray".

 

[Magtheridon96]

I'm using a 3D array to store unit-unit data

This is an example:
3DArray[GetUnitId(unitOne)][GetUnitId(unitDeux)][DATA_TYPE_THREAT]

 

[Bribe]

So it's effectively a Table per unit per unit?

 

[Laiev]

I always asked 'why Blizzard implement a 5D array at Starcraft II', but I see the answer while coding the Cooldown System ;P a 3D array is so common in general.

 

[Magtheridon96]

So it's effectively a Table per unit per unit?

Yeah
With 600 units, I'd have 360000 Tables and that's not even close to 0.1% of the limit ;D

 

[Bribe]

It's not close to the limit but you should also be considering RAM at that point. Ideally, not every unit on the map is going to interact with every unit on the map on an all-for-one, one-for-all basis, so you wouldn't end up needing a direct units-on-map^2 setup.

 

[Magtheridon96]

I'm filtering units out so that I only create space for Hero-unit data in fact
I was just giving an example above, plus, I'm only storing integer data =p

 

[mckill2009]

And he's correct, but the order you use them is irrelevant because you would reach the var in the same way anyways...
tb[1][2] = 3
tb[2][1] = 3
However the Parent can't go over 8190 because it's a struct array allocator, but the child is a hashtable index, so it can go beyond.
Anyways, i've never ever had to use TableArray anyways, you can always avoid it

even if it's like this set ai = TableArray[2], parent ID's can still be reduced by this set ai[id-id+1][1] = 5...although you're right about the order...
one problem though, I cant destroy nor flush it...

 

[Bribe]

What do you mean you can't destroy nor flush it?

call ta.destroy()
call ta.flush()

If you have a TableArray[2], you can access slots 0 and 1, not 1 and 0. It works exactly like a normal JASS array where it starts at 0 and ends at 8191. Actually, pretty much all arrays in computer science start at 0 and stop at "size - 1".

If you look at what I have going in the Retro library, I use what is basically a TableArray (but I did it with a hashtable at the time) with offsetting so I actually have 3-D arrays:

http://www.hiveworkshop.com/forums/2050808-post39.html

 

[mckill2009]

I can't destroy nor flush it coz it causes debug message errors...
But the problem is fixed now, coz I put it inside the loop...suppose to be outside...
I have same results even though I reverse the order like Dirac said...

set ai = TableArray[2]
    loop
        set t = CreateTimer()
        set id = GetHandleId(t)
        set ai[id-id+1][1] = 88
        set ai[id-id+1][2] = 999
        set ai[id-id+1].string[1] = "hello"
        set ai[id-id+1].string[2] = "table"
        call BJDebugMsg(I2S(ai[id-id+1][1]))
        call BJDebugMsg(I2S(ai[id-id+1][2]))
        call BJDebugMsg(ai[id-id+1].string[1])
        call BJDebugMsg(ai[id-id+1].string[2])
        //TableArray Error: Tried to double-free instance:
        //but this is OK now, suppose to be ourside
        call ai.flush()
        set i = i-1
        set t = null
        exitwhen i==0
    endloop

 

[Bribe]

Yeah because you are still saving data to the same place twice, you'd have to change it to this:

        set ai[id-id+1][1] = 88
        set ai[id-id+1][2] = 999
        set ai[id-id+1].string[@3@] = "hello"
        set ai[id-id+1].string[@4@] = "table"

 

[Dirac]

You didn't reverse the order.
[id-id] is equal to 0, don't know where are you getting at
You're storing a string over an integer instance

ai[1][1] = 88
ai[1][2] = 999
ai[1].string[1] = "hello"
ai[1].string[2] = "table"

You're confusing the TableArray feature with common Table functionality.
ai[TableArray index][hashtable index] not the other way around.

 

[mckill2009]

@Dirac
i know that id-id=0, Im just experimenting dude if a small size can work that way with
a huge handleID of a timer is implemented, "1" as parentID is fine with the loop but what
if its a looping timer in another function?, that's why in the scipt there is a timer...

 

[Bribe]

I recommend TimerUtils + structs for things like that. A timer is predictable and data attachment is an equally predictable utility. SetTimerData/GetTimerData has worked perfectly for years.

Generally, I recommend to use Table instances for more complicated storage needs. Tables are often used for the longer-term or if there is some crazy data structure going on, like an item recipe system.

 

[Magtheridon96]

like an item recipe system.

An indexed array of stacks

 

[Bribe]

set TB.force[GetHandleId(p)] = f

There is no "method operator []x=", there is only "method operator []="

 

[deathismyfriend]

y does having 2 TableArrays[0x2000] in the same library cause them to bug up ? heres how im doing this
this is in the init function

set itemIdTable = TableArray[0x2000]
            set itemDropTable = TableArray[0x2000]

here is the problem

set this.index = this.index + 1
                    set this.percent = this.percent + percents[L]
                    set itemDropTable[this][index] = this.percent
                    set itemIdTable[this][index] = itemIds[L]

when i call this

call BJDebugMsg( I2S( itemDropTable[this][L]))
                    call BJDebugMsg( I2S( itemIdTable[this][L]))

it displays the same values as the itemIdTable but when i use the
call BJDebugMsg( I2S( itemDropTable[this][L])) in between the

                    set itemDropTable[this][index] = this.percent
                    set itemIdTable[this][index] = itemIds[L]

it displays the correct numbers

 

[Bribe]

When it comes to using Table, if you are getting errors, I advise checking over your code to make sure things match up the way you wanted them to. Table, by design, is read and write protected as long as you are following the rules.

 

[deathismyfriend]

so from what i posted it looks like it should work ?

heres the full init method if that helps

private static method onInit takes nothing returns nothing
            local trigger t = CreateTrigger()
            set unitTypeTable = Table.create()
            set unitIdTable = Table.create()
            set itemIdTable = TableArray[0x2000]
            set itemDropTable = TableArray[0x2000]
            call TriggerRegisterAnyUnitEventBJ( t, EVENT_PLAYER_UNIT_DEATH)
            call TriggerAddCondition( t, function thistype.deathEvent)
            set t = null
        endmethod

and heres the full function that writes the values. they are not written anywere else

method addItem takes integer itemId1, integer percent1, integer itemId2, integer percent2, integer itemId3, integer percent3, integer itemId4, integer percent4, integer itemId5, integer percent5 returns nothing
            local integer L = 1
            local integer array itemIds
            local integer array percents
            set itemIds[1] = itemId1
            set itemIds[2] = itemId2
            set itemIds[3] = itemId3
            set itemIds[4] = itemId4
            set itemIds[5] = itemId5
            set percents[1] = percent1
            set percents[2] = percent2
            set percents[3] = percent3
            set percents[4] = percent4
            set percents[5] = percent5
            loop
                exitwhen L > 5
                if itemIds[L] != 0 then
                    set this.index = this.index + 1
                    set this.percent = this.percent + percents[L]
                    set itemDropTable[this][index] = this.percent
                    set itemIdTable[this][index] = itemIds[L]
                    call BJDebugMsg( I2S( itemDropTable[this][L]))
                    call BJDebugMsg( I2S( itemIdTable[this][L]))
                endif
                set L = L + 1
            endloop
            if this.percent > 100 then
                call BJDebugMsg( "|c00ff0303WARNING|r |cffffcc00Please Change The Percents of the Items for unit with Item Type Id = |c00ff0303" + I2S( this.unitTypeId) + "|r 
                |cffffcc00as the items are over 100% which will cause bugs|r |c00ff0303WARNING|r")
            endif
        endmethod

edit: i got it to work another way just want to know y it doesnt work this way?

 

[Bribe]

You are debugging [this ][ L ], but it looks like you need to debug [this ][ index ].

 

[deathismyfriend]

You are debugging [this ][ L ], but it looks like you need to debug [this ][ index ].

sry when i edited the top post i didnt change that back to index. in this case tho index and L are the same values.

i attached the map itself so u can see the problem the one up top is the fixed version. the one labeled problem is the problem if u just start the test map u will see that the values r the same. thx for ur help

 

[Nromk]

What does the 0x2000 mean? Is that related to the size?
(I am referring to your example of TableArray)

 

[Magtheridon96]

0x2000 is 8192 in base 16.

You are correct, it refers to the size of the TableArray.

 

[Nromk]

So there are 8192 Tables inside the TableArray? That is quite a lot...

 

[Magtheridon96]

Considering the fact that the max Table count is 2.147 billion, it's totally fine.

 

[mckill2009]

Found a nasty bug...

scope BUG

struct Bug
    static Table t
   
    static method onInit takes nothing returns nothing
        local effect sfx = AddSpecialEffect("Abilities\\Spells\\NightElf\\Starfall\\StarfallCaster.mdl", 0,0)
        set t = Table.create()
        set t.effect[123] = sfx
        set t.player[123] = Player(1) //if I disable this line, it will work
        call DestroyEffect(t.effect[123])
    endmethod
endstruct

endscope

so the player and effect share the same childKey?

EDIT:

BTW, normal hashtable works;

scope BUG

struct Bug
    static hashtable h = InitHashtable()
   
    static method onInit takes nothing returns nothing
        local effect sfx = AddSpecialEffect("Abilities\\Spells\\NightElf\\Starfall\\StarfallCaster.mdl", 0,0)
        call SaveEffectHandle(h, 123, 1, sfx)
        call SavePlayerHandle(h, 123, 2, Player(1))
        call DestroyEffect(LoadEffectHandle(h, 123, 1))
   endmethod
endstruct

endscope

 

[Almia]

Question.
How do i compare integers saved in the Table?
I tried something like this:

if someTable[id] > SomeConstantInt then
    set someTable[id] = someTable[id] + 1
endif

But,errors says that I cannot compare Relative Type to Native Type.

 

[kStiyl]

Question.
How do i compare integers saved in the Table?
I tried something like this:

if someTable[id] > SomeConstantInt then
    set someTable[id] = someTable[id] + 1
endif

But,errors says that I cannot compare Relative Type to Native Type.

(integer) (someTable[id]) > SomeConstantInt

 

[mckill2009]

Question.
How do i compare integers saved in the Table?
I tried something like this:

if someTable[id] > SomeConstantInt then
    set someTable[id] = someTable[id] + 1
endif

But,errors says that I cannot compare Relative Type to Native Type.

if someTable[id] > SomeConstantInt then
==>
if (someTable[id] > SomeConstantInt) then

 

[Almia]

didn't know that.

 

[Bribe]

That is because Table drops Tables instead of integers, so due to Vexorian's compilation priorities it thinks you are trying to do an operator overload of an integer comparison. As others stated, the fix would be to either do a typecast "integer(table[data])" or simply set an integer variable to the Table variable (which you might do anyway, in case you need to repeat reference it).

 

[Maker]

BTW, normal hashtable works;

struct Bug
    static hashtable h = InitHashtable()
   
    static method onInit takes nothing returns nothing
        local effect sfx = AddSpecialEffect("Abilities\\Spells\\NightElf\\Starfall\\StarfallCaster.mdl", 0,0)
        call SaveEffectHandle(h, 123, 1, sfx)
        call SavePlayerHandle(h, 123, 2, Player(1))
        call DestroyEffect(LoadEffectHandle(h, 123, 1))
   endmethod
endstruct

Why would that not work? I think you should compare to this

struct Bug
    static hashtable h = InitHashtable()
   
    static method onInit takes nothing returns nothing
        local effect sfx = AddSpecialEffect("Abilities\\Spells\\NightElf\\Starfall\\StarfallCaster.mdl", 0,0)
        call SaveEffectHandle(h, 123, 1, sfx)
        call SavePlayerHandle(h, 123, 1, Player(1))
        call DestroyEffect(LoadEffectHandle(h, 123, 1)) // You're screwed :(
   endmethod
endstruct

 

[mckill2009]

Why would that not work? I think you should compare to this

You're screwed

Mind if I ask why?, coz as I said the normal HT works, not the Table...

 

[Troll-Brain]

mckill, check again your code you don't use the same keys for the hashtable native way, use '1' two times instead and it will fail just the same.
That's how hashtables work (and that should be the same with a gamecache)

 

[Magtheridon96]

And no internal errors pop up because these things are stored as 4-byte pointers

 

[mckill2009]

@Troll, do you think I dont know how to use keys in HT's?, please read my post carefully, I'm NOT using the SAME key '1', Maker did that:

I did this:
call SavePlayerHandle(h, 123, 2, Player(1))
He did this:
call SavePlayerHandle(h, 123, 1, Player(1))

and BTW Im not complaining about hashtables anyway, but the Table made by Bribe, look at my post before that...

 

[Bribe]

If you are getting errors with Table, it's all but 100% of the time usage error. That's where you look for the source of the problem - use Table as a reference and learn what part of your code could be doing something wrong.

 

[mckill2009]

it's all but 100% of the time usage error.

The code in post #131 is very short so it's so easy to check it, anyway, here's a sample map to prove it...

 

[Troll-Brain]

Found a nasty bug...

scope BUG

struct Bug
    static Table t
   
    static method onInit takes nothing returns nothing
        local effect sfx = AddSpecialEffect("Abilities\\Spells\\NightElf\\Starfall\\StarfallCaster.mdl", 0,0)
        set t = Table.create()
        set t.effect[123] = sfx
        set t.player[123] = Player(1) //if I disable this line, it will work
        call DestroyEffect(t.effect[123])
    endmethod
endstruct

endscope

so the player and effect share the same childKey?

EDIT:

BTW, normal hashtable works;

scope BUG

struct Bug
    static hashtable h = InitHashtable()
   
    static method onInit takes nothing returns nothing
        local effect sfx = AddSpecialEffect("Abilities\\Spells\\NightElf\\Starfall\\StarfallCaster.mdl", 0,0)
        call SaveEffectHandle(h, 123, 1, sfx)
        call SavePlayerHandle(h, 123, 2, Player(1))
        call DestroyEffect(LoadEffectHandle(h, 123, 1))
   endmethod
endstruct

endscope

Pride is such a pity ...

As said for the table way you use the same keys but you don't with the native hashtable way, you just can't compare them ...

EDIT :

so the player and effect share the same childKey?

Obvioulsy yes, not hard to check anyway, the code is available.
So it's not a bug just an expected behavior, but yes you have to we aware about this.
If you really need to store different types of handle just use a struct.

 

[mckill2009]

Obvioulsy yes, not hard to check anyway, the code is available.
So it's not a bug just an expected behavior, but yes you have to we aware about this.
If you really need to store different types of handle just use a struct.

Truth to be told, I really didnt examine that the 'key' is set as a child, that's why it keeps on overwriting, well I guess TableArray is just fine to use...

 

[Troll-Brain]

I suppose it's worth to write this behavior in the Table documentation though.
That's not that obvious, i mean if you didn't use hashtables that way before.
And anyway, let's face it, people don't always read/understand a code before using it, you're a perfect example ^^

 

[Trakya]

Hello everyone. I need help with creating a table for map called Heroes and Armies x1.49 (example: Table who can shows numbers of creeps all allies/enemy hero icons, kills assists deaths and creep, kills and game time if possible)

 

[Almia]

This is Table,not a Multiboard or somesort of.
Please use the World Editor Help Zone.

 

[Cokemonkey11]

/*
    | method flush takes nothing returns nothing
    |     flush all stored values inside of it
    |    
    | method remove takes integer key returns nothing
    |     remove the value at index "key"
    |    
    | method operator []= takes integer key, $TYPE$ value returns nothing
    |     assign "value" to index "key"
    |    
    | method operator [] takes integer key returns $TYPE$
    |     load the value at index "key"
    |    
    | method has takes integer key returns boolean
    |     whether or not the key was assigned
*/

Just chiming in here to let you know that this is absolutely unacceptable API design. You, under no circumstances, should change or remove behavior from the library you updated.

If your Table is meant to be completely different, then Table is an unacceptable name.

I suggest you take a look at this: http://www.youtube.com/watch?v=heh4OeB9A-c

 

[Bribe]

Cokemonkey11, this has been discussed already. You are like 2 years too late to be recommending changes to this resource. The original Table resources pales in comparison to this, anyway.