[Lua][vJass] New Table

[Bribe]

The problem could be that you have TableArrays declared as Tables. You shouldn't be using typecasting between Tables and TableArrays.

 

[deathismyfriend]

This is what is happening.

itemDropTable[this][itemIdTable[this][L]]

The first part is the one table array.
This loads the the item type from the table array.

itemIdTable[this][L]

The item type is then used to load the weight.

itemDropTable[this][item type]

 

[Bannar]

From what I understood, problem occurs when there are multiple item tables declares, meaby even to the same unit type.

Problem may lie within: set unitTypeTable[ this.unitTypeId] = this since you are reassigning table index to be used on death event each time item table is declared for type T, meaning: with your currect create method, I can not have multiple tables for given unit type, like 'hpea'. Instead of allocating new instance you should have checked if such unitType table is already found and allocated. If so, just use recovered index instead.

Provide demo code with which you are testing your bugs, would help a lot.

B) there are many flaws within your code, like:
// where 'r' is random int from 1 to weight if r > weights[L1 - 1] and r < weights[L1] then what happens if we roll exactly 'border' number? (lets say 20 in your test map). There is no "=" operator within if statement, thus such item will never be found.
Searching and indexing could be improved a lot, however, this is not the case now.

 

[deathismyfriend]

From what I understood, problem occurs when there are multiple item tables declares, meaby even to the same unit type.

Problem may lie within: set unitTypeTable[ this.unitTypeId] = this since you are reassigning table index to be used on death event each time item table is declared for type T, meaning: with your currect create method, I can not have multiple tables for given unit type, like 'hpea'. Instead of allocating new instance you should have checked if such unitType table is already found and allocated. If so, just use recovered index instead.

Provide demo code with which you are testing your bugs, would help a lot.

There are no duplicates for unit type.

This is what I am using.

function addToTable takes nothing returns nothing
    local ItemDropSystem data = ItemDropSystem.createForType( 'hpea')
    call data.addItem( 'afac', 20)
    call data.addItem( 'ratc', 20)
    call data.addItem( 'rat6', 20)
    call data.addItem( 'rat9', 20)
    call data.addItem( 'rst1', 100)
    
    set data = ItemDropSystem.createForType( 'hfoo')
    call data.addItem( 'rag1', 100)
endfunction

Here are the methods with debug messages in them.

        method addItem takes integer itemId, integer weight returns nothing
            set this.index = this.index + 1
            set this.ttlWeight = this.ttlWeight + weight
            set itemIdTable[this][index] = itemId
            call BJDebugMsg( " Struct instance = " + I2S( this) + " Item Id = " + I2S( itemIdTable[this][index]) + " = " + I2S( itemId))
            set itemDropTable[this][itemId] = this.ttlWeight
            call BJDebugMsg( " Struct instance = " + I2S( this) + " Item weight " + I2S( itemDropTable[this][itemId]) + " = " + I2S( this.ttlWeight))
        endmethod
        
        private method dropItem takes real x, real y returns boolean
            local integer r
            local integer L = 1
            local integer L1 = 1
            local integer end = 0
            local integer array weights
            set weights[0] = 0
            if this != 0 then
                loop
                    exitwhen L > this.index
                    set weights[L] = itemDropTable[this][itemIdTable[this][L]]
                    call BJDebugMsg( " Struct Instance = " + I2S( this) + " Item # = " + I2S( L) + " Unit type id = " + I2S( this.unitTypeId) + " Item Id = " + I2S( itemIdTable[this][L]) + " Weight = " + I2S( itemDropTable[this][itemIdTable[this][L]]))
                    set end = end + 1
                    set L = L + 1
                endloop
                set L = 1
                loop
                    exitwhen L > this.multiDrop
                        set r = GetRandomInt( 1, this.ttlWeight)
                        loop
                            exitwhen L1 > end
                            if r > weights[L1 - 1] and r <= weights[L1] then
                                call CreateItem( itemIdTable[this][L1], x, y)
                            endif
                            set L1 = L1 + 1
                        endloop
                    set L = L + 1
                endloop
            endif
            return this.dropTypeTableAlso
        endmethod

Below are some pics of what I get with those messages.

B) there are many flaws within your code, like:
// where 'r' is random int from 1 to weight if r > weights[L1 - 1] and r < weights[L1] then what happens if we roll exactly 'border' number? (lets say 20 in your test map). There is no "=" operator within if statement, thus such item will never be found.
Searching and indexing could be improved a lot, however, this is not the case now.

Yes I did miss that = in the weights check. Thanks.
It now looks like this.
// where 'r' is random int from 1 to weight if r > weights[L1 - 1] and r <= weights[L1] then

 

[Bribe]

deathismyfriend, the main point is that you have the variables declared wrong. You can't set a Table to a TableArray index, and then not typecast that Table back to TableArray each time you use it. Just declare it as a TableArray.

 

[Bannar]

Because of your rep I'm representing 666 haxs

Honestly, when I entered your test map, I've immidiately replaced Table with TableArray where is was needed thus as of now, I still could not see the problem.

Meaby I'm just dumb but your example is more complex one, since its system implementation. Usually, the best way is to post smallest code possible, without anything (and no, add & drop Item methods solely wont work, meaning pasting just those 2 is incomplete ^)^ ) unrelated to the problem with statement what you'd exect from the code and whats the actuall output of the code is. Fine, you have added messages, but I dont know whats those should or should not shown.


However, you can not neclect this:

        static method createForType takes integer unitTypeId returns thistype
            local thistype this = thistype.allocate()
            set this.unitTypeId = unitTypeId
            set unitTypeTable[ this.unitTypeId] = this
            set this.specificUnit = false
            set this.dropTypeTableAlso = false
            call this.setValues()
            return this
        endmethod
       
        static method createForUnit takes unit u, boolean dropTypeTableAlso returns thistype
            local thistype this = thistype.allocate()
            set this.unitTypeId = GetHandleId( u)
            set unitIdTable[ this.unitTypeId] = this
            set this.specificUnit = true
            set this.dropTypeTableAlso = dropTypeTableAlso
            call this.setValues()
            return this
        endmethod

Lets assume user is tricky and/or mad and suddenly he adjust 2 tables for signle unit type?
Fix is quite simple:

        static method createForType takes integer unitTypeId returns thistype
            local thistype this
            if unitTypeTable[ this.unitTypeId] == 0 then // is allocated?
                set this = thistype.allocate()
                set this.unitTypeId = unitTypeId
                set unitTypeTable[ this.unitTypeId] = this
                set this.specificUnit = false
                set this.dropTypeTableAlso = false
                call this.setValues()
            else
                set this = unitTypeTable[ this.unitTypeId]
            endif
            return this
        endmethod

 

[deathismyfriend]

deathismyfriend, the main point is that you have the variables declared wrong. You can't set a Table to a TableArray index, and then not typecast that Table back to TableArray each time you use it. Just declare it as a TableArray.

Thanks Bribe. That was my problem this whole time. What a stupid mistake.

@Bannar
Thanks for your help as always. I will fix it to protect the creation.

Thanks to both of you +rep

Edit: I can't rep either of you atm sorry.

Edit 2: Fixed Bannars name lol.

 

[Almia]

@offtopic

@Bunnery
You should change your name, everyone is mistyping it,like deathismyfriend did.

 

[TriggerHappy]

You should change your name, everyone is mistyping it,like deathismyfriend did.

He just changed his name. Hence the "Formerly known as Spinnaker".

 

[Almia]

I know he is Spinnaker

 

[Bannar]

No I am not ;>
I've already stated, Spin, was not my real nick. Its my bros.

And btw, there is only 1 nickname change per lifetime on hive

 

[Almia]

No I am not ;>
I've already stated, Spin, was not my real nick. Its my bros.

And btw, there is only 1 nickname change per lifetime on hive

False.
I saw one member changed his name 2 times. I think it was Kimberly.

 

[Bannar]

This is what Ralle told me. I was an admin on few forums, and yes changing users nickname is easy - but it can also bring burden into forum with everyone getting confused who is he writing with.
Its good that Ralle set some rules, one of them being nickname limitation change.

Anyways, lets not go with offtopic any further.

 

[gorillabull]

an epic 1 cap'n murica!

 

[Geshishouhu]

Can anyone explain what are the top two lines used for? How are they useful?
Thank you.

set a[654321] = 'A'   //<-------
set a[54321] = 'B'  //<-------
set a.unit[12345] = GetTriggerUnit()
set a.unit[GetHandleId(a.unit[12345])] = GetSpellTargetUnit()
set a.real['ABCD'] = 3.14159
set a.integer[133] = 21

 

[PurgeandFire]

It is just assigning the integer 'A' and integer 'B' to particular keys in the table instance "a".

a -> key 654321 -> 'A'
a -> key 54321 -> 'B'

'A' and 'B' are integers. The apostrophe 'XXXX' notation is normally used for rawcodes. Rawcodes are just formed from ASCII characters, e.g. 'a' corresponds to 97. When you have a code such as 'abcd', it'll use an algorithm to generate an integer representation.

We use raw codes all the time for unit types, abilities, etc. But as for the individual 'a' or 'b', you won't find too many uses except in specific cases. For example, dialogs provide an option for a hotkey, and you can input 'a' or 'b', etc. to determine the hotkey (since it takes an integer, not a string). You can also use it for some weird array mechanic, if you want to be fancy:

local integer array i
set i['a'] = 24
set i['b'] = 25
set i['c'] = 26
// equivalent to
set i[97] = 24
set i[98] = 25
set i[99] = 26

 

[Geshishouhu]

It is just assigning the integer 'A' and integer 'B' to particular keys in the table instance "a".

a -> key 654321 -> 'A'
a -> key 54321 -> 'B'

'A' and 'B' are integers. The apostrophe 'XXXX' notation is normally used for rawcodes. Rawcodes are just formed from ASCII characters, e.g. 'a' corresponds to 97. When you have a code such as 'abcd', it'll use an algorithm to generate an integer representation.

We use raw codes all the time for unit types, abilities, etc. But as for the individual 'a' or 'b', you won't find too many uses except in specific cases. For example, dialogs provide an option for a hotkey, and you can input 'a' or 'b', etc. to determine the hotkey (since it takes an integer, not a string). You can also use it for some weird array mechanic, if you want to be fancy:

local integer array i
set i['a'] = 24
set i['b'] = 25
set i['c'] = 26
// equivalent to
set i[97] = 24
set i[98] = 25
set i[99] = 26

Thanks for the explanation. But why not just use this if 'A' and 'B' are considered integers:

set a.integer[654321] = 'A' 
set a.integer[54321] = 'B'

 

[PurgeandFire]

Thanks for the explanation. But why not just use this if 'A' and 'B' are considered integers:

set a.integer[654321] = 'A' 
set a.integer[54321] = 'B'

That option wasn't added until later requested by Spinnaker. Before, you had to save integers without the ".integer". It might've been for backwards compatibility/similarity with Vexorian's Table.

 

[Geshishouhu]

That option wasn't added until later requested by Spinnaker. Before, you had to save integers without the ".integer". It might've been for backwards compatibility/similarity with Vexorian's Table.

Oh, got it, thnx a lot! PnF, you seems the only active mod on Hive now.

 

[edo494]

oh it was hell to do Table stuff with textmacroes, having to separatly handle .integer. Im happy to see this added, even tho this is quite old news

 

[Bribe]

Oh, got it, thnx a lot! PnF, you seems the only active mod on Hive now.

I'm always trolling the forums I just don't have any time lately to get intellectual due to work sapping all my creativity.

 

[Geshishouhu]

I'm always trolling the forums I just don't have any time lately to get intellectual due to work sapping all my creativity.

Good to know that. Thank you!

 

[sethmachine]

This is such a great resource! It should be a standard in any map that uses vJASS. Many thanks for your work Bribe.

 

[Deleted member 219079]

This is really nice, only one hashtable makes it perfect

Here, local TableArray da = TableArray[0x2000] on your example, why do you have an x in integer?

 

[IcemanBo]

If a number starts with 0x, it's a hex value.

 

[Deleted member 219079]

Okay, hex value, cool (I've no idea what it is)

 

[IcemanBo]

decimal: 0-9
hexadicimal: 0-F (0, 1, ..., 9, A, B, C, D , E, F)
So 0xFF would be like 255 in decimal for example.

 

[edo494]

you have decimal, octal, binary and hexadecimal values(there are more obscure like duodecimal).

Hexadecimal allows you to have 1 character per 16 values, so valid hexadecimal literals are 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, a, b, c, d, e, f(and upper case equivalents), where a is equal to 10, b to 11, ..., f to 15 and finally 10 in hexadecimal is 16. This is the same as the rawcode system, but rawcode uses 256-base numeric system

 

[Deleted member 219079]

I'd be a total liar if I said I can do nxnnnn numbers now :/ But thanks for explaining anyway! Maybe some day..

What's the difference of flush and destroy? I wanted to destroy so I made flush then destroy but it nags at me, does destroy also flush? The api section only says destroy destroys it.. I'm so confused of this.

 

[edo494]

Flush merely reemoves everything within the table instance, whereas destroy flushes the table and recycles its instance so that when you call create next time it will give you that instance back

Call flush when you want to continue using the instance. If you no longer need it(such as inside your struct's destroy method), then you should call destroy

 

[Deleted member 219079]

Okay thanks dude, got it fixed now

 

[sethmachine]

Correct me if I'm wrong, but I believe Table is missing weathereffect.

Would appreciate if that could be added in, or I might do it myself if it's just a text macro that builds each .type syntax.

Edit: apologies. appears that Blizzard did not make LoadWeatherEffectHandle .

 

[chobibo]

I don't think Bribe would mind if you edit the code to your needs, as long as you don't redistribute it as your own work (no offense intended).

 

[sethmachine]

Well I'd rather not get messy.

But it would be nice (well nice for me) if Table supported weathereffect and possibly terraindeformation.

 

[Bannar]

For both handles, natives related to Save & Load are not available. Also, if you consider that it was Bribe who made this lib, it'd be unlikely that he had missed such thing.

 

[Bribe]

Non-hashtable supported agents are impossible to load without I2X typecast

 

[Deleted member 219079]

Is there a way to make array - based version? I mean in speed comparison cache < hashtable < variable arrays, so it could be blazing fast

I don't case if it'd only support 8190 instances, it'd still be super cool

 

[chobibo]

Why not just use normal arrays then? Or what you want is a jass hash function? I don't think it would be faster than the native hashtable though.

 

[Trigger.edge]

Sorry for my ignorance, but how to use HaveSavedHandle (in Table version)?

 

[Bribe]

boolean b = table.handle.has(someHandle)

 

[Trigger.edge]

lol thanks

 

[Kazeon]

If I got:

TableArray A = TableArray[5]
TableArray B = TableArray[5]

// First action
set A[1].integer[5] = 5
// Second one
set B[1].integer[5] = 6

will the second action overwrite A[1].integer[5]'s value from 5 to 6?

Currently, I think it does replace the value.

 

[edo494]

It shouldnt, since TableArray[xx] will create 5 Tables and return them to you, so they are completly different.

Try printing A[1] and B[1] and you will see it will not overrite it(it shouldnt at least), because they are different instances(they should be)

 

[Kazeon]

It shouldnt, since TableArray[xx] will create 5 Tables and return them to you, so they are completly different.

Try printing A[1] and B[1] and you will see it will not overrite it(it shouldnt at least), because they are different instances(they should be)

I tested it: it does replace.

 

[edo494]

can you post whole code, and did you actually teste the integers of A[1] and B[1]?

 

[Kazeon]

can you post whole code, and did you actually teste the integers of A[1] and B[1]?

Well, I don't have demo code, it's in my map. But I'm sure it does replace because since I replaced TableArray with my multi-dimension array snippet, the system works correctly.

 

[edo494]

scope s initializer i
    private function i takes nothing returns nothing
        local TableArray a = TableArray[5]
        local TableArray b = TableArray[5]
        
        call BJDebugMsg(I2S(integer(a[2])))
        call BJDebugMsg(I2S(integer(b[2])))
        
        set a[2].integer[4] = 8
        set b[2].integer[4] = 11
        
        call BJDebugMsg(I2S(a[2].integer[4]))
    endfunction
endscope

prints correctly 8 for me, and the indexes are -3 and -8(apparently TableArray will allocate Tables with negative indexes growing from 0 to -inf).

You have some problem on your end most likely

 

[Kazeon]

Hmm.. Perhaps I did some mistakes, but I can't check it anymore since I have replaced TableArray with MultiArray

Thanks for testing it

 

[Trigger.edge]

Lately I can't stop using Table, so it is my duty to say:

Some thinks of Table:

TableArray has many disadvantages:

- You have to know the index/size set ta=TableArray[size/index] when it is created .

- The point of using "TableArray" is that just not have limits... when set some value: set ta[myMissIndex].integer[this]=someValue

- TableArray if I am not mistaken does not work with arrays variables Table array ta because of static method operator []

It is tedious create a TableArray with a certain size... why not just use "TableArray" in Table struct as follows:

set ta.integer[parentKey][childKey]=someValue.

For me and I think for many, it would also be excellent if there was a new update of your Table fixing these disadvantages.

 

[edo494]

I dont even undertand the array thing, can you elaborate more?

Making TableArray resize dynamically is very hard to achieve, because every write would eventually turn into potential allocation