JassHelper 0.A.2.B - A vJass and Zinc 2 Jass compiler

[Dark_Dragon]

Originally Posted by daPanda

I have a few questions about allocation if someone could clarify plz.


on another forum I was reading a post about global structure allocation. someone had posted the following:

JASS:

struct Branch   
//data
endstruct

globals   
Branch array STBRANCH[500]
endglobals

function test takes nothing returns nothing
   set STBRANCH[1] = Branch.create()   
   set STBRANCH[2] = Branch.create()
endfunction

someone replied that the calls in the function to create each array element wasn't necessary because global array elements don't need to be created or destroyed.

another person in the same thread also said that the following is also legit without allocating it each array element:

JASS:

struct name extends array
     real x
     real y
endstruct

function bla takes nothing returns nothing
     set name[1].x = 5.
     set name[2].y = 6.
endfunction

without even having to declare a global var or allocate each element.

but, when I try to do this, the data I store is unstable. I can't see anywhere in the output code where name[1] or name[2] is assigned a pointer (allocated) with the above code "as is".

In my mind, it seems necessary to do the following first before using the above:
set name[1] = name.create()
set name[2] = name.create()

I've always done it this way (allocating each array element). When I tested the above by removing the .create() from my global struct array elements, the data is unreliable.

are they correct in saying that the array elements don't have to be allocated manually if they are global struct arrays? I mean, I know they array itself is allocated (obviously) but each element is allocated as well?

their statements have me confused.

thanks for your input

the first part when someone said that array globals do not need to be allocated was wrong! why? coz that struct does not extend array... second example was correct and you dont need to allocate them since they extend array... which means there is no .create or .destroy methodes... thats coz array structs are usefull when it comes for player id works or simply if you need your own indexing system implemented.

and so its true that in array structs like u said! each element is allocated by you when you use it! no need for .create or anything like that since extending array means that it is automatically allocated when you insert specific index...

Line 62438: memory exhausted.
wtf?!

would u mind sending that line here?

[Eleandor]

Originally Posted by Litany

Your grasp of predication and inherence is not so good.

Will anyone ever create a shape? No. They'll create circles or squares, but never ever shapes. Therefor, shapes should not be structs. It's as simple as that.

[Flame_Phoenix]

Will anyone ever create a shape? No. They'll create circles or squares, but never ever shapes. Therefor, shapes should not be structs. It's as simple as that.

As I said before, it depends on your implementation. You are not aware of the system, therefore you do not have as much authority as Litany has to discuss it.

Please stop this arguments, we are here to talk about vJass, not about a Shape system.

And thx to dragoon for showing me delegate =D

[Dark_Dragon]

Originally Posted by Flame_Phoenix

As I said before, it depends on your implementation. You are not aware of the system, therefore you do not have as much authority as Litany has to discuss it.

Please stop this arguments, we are here to talk about vJass, not about a Shape system.

And thx to dragoon for showing me delegate =D

you are welcome ;)

[daPanda]

Thanks Dragon.

[Dark_Dragon]

Originally Posted by daPanda

Thanks Dragon.

np and btw just one thing i forgot!

JASS:

struct test extends array
   real x
   real y
   real z
endstruct

function testfunc takes nothing returns nothing
    local test t


    set test[1].x = 3.
    set test[1].y = 4.
    set test[1].z = 5.
    set test[2].x = 6.
    set test[2].y = 7.
    set test[2].z = 8

     // insted of that long stuff u can do this:
     set t = test(1)
     set t.x = 3.
     set t.y = 4.
     set t.z = 5.
     set t = test(integer(t) + 1)
     set t.x = 6.
     set t.y = 7.
     set t.z = 8.

     // will do same as array accessing! but its lesser to write :D
     // so pick one u like more
endfunction

Greets!
~DD

EDIT:

He said "A "shape" is not *real*." This is wrong.

hmm well to say the truth! both of you are correct! shape is real and is not... ofc it can be only one of this but from point of you two guys are looking... the result is that both of u are correct.

you are looking at shape as an "in programming" object which can be line, circle... which would be "real" coz we know what line or circle is and therefore we can know what shape would be... any of this ofc! since you look at it that way it would be real.

from point Eleandor is looking is well! shape is not known what it is at beginning... line, circle... so he is looking at them as we look at re+im or (x+yi)! well x would be real like line or circle and yi would be shape... well we can write yi! or y*sqrt(-1) but we cant immagine what it is other then that its on second dimension... so we can say that for shape we know it but we dont!

i cant explain this to good but i understand what both of ur are talking about... so yeah from my point of view both of u are correct since every one of u sees it different!

[Alevice]

Originally Posted by Alevice

I was thinking about that later on. I personally don't like the underscore notation for scoping and such, but it surely is a wrokaround. Pluse you don't have something like "using scope MyScope" and go the short way when you know your name might not collide somewhere else (not that it matters much).

Thanks for the input.

Alright, for the record, scopes and libraries can't be nested, so ditch that.

Also, I don't know if this is a bug or a deliberate design behaviour but I find it odd: it seems structs within libraries get their static members declared after global variables in goddamned scopes. COme the fuck on. Case in point

JASS:

library UnitIds
    private keyword KR_Units
    struct KalimdorRaiders
        static KR_Units Units
    endstruct
    
    private struct KR_Units
        static constant integer KOBOLD_MINER = 'kru0'
        static constant integer KODO_WAGON = 'kru1'
        static constant integer KODO_OFFSPRING = 'kru2'
        static constant integer KODO_OFFSPRING2 = 'krm2'
        static constant integer CENTAUR_DRUDGE = 'kru3'
        static constant integer CENTAUR_ARCHER = 'kru4'
        static constant integer OGRE_CRUSHER = 'kru5'
        static constant integer OGRE_CRUSHER2 = 'krm5'
        static constant integer HARPY_SCOUT = 'kru6'
        static constant integer HARPY_WITCH = 'kru7'
        
        static constant integer OGRE_CRUSHER_SIEGE = 'kru8'
        static constant integer OGRE_CRUSHER_SIEGE2 = 'krm8'
    
    endstruct
endlibrary

scope Marionette initializer Init

    globals
        private constant integer CASTER_ID = KalimdorRaiders.Units.HARPY_WITCH
    endglobals
    
endscope

It tells me KalimdorRaiders is an undeclared variable.

[Vexorian]

variables from structs are added after other global variables. No matter if it is a library or a scope or whatever. My solution? Assign the global post- globals init ...

I don't think there is a way to change this rule any time soon, as long as structs are compiled in a separate phase, it won't.

[daPanda]

thanks again dragon. one more question about that if you don't mine.

What if the extended struct contains another struct? is it also allocated?

JASS:

struct name1 extends array
   integer a
   integer b
endstruct

struct name2 extends array
     name1
     real x
     real y
endstruct

function bla takes nothing returns nothing
     set name2[1].x = 5.
     set name2[2].y = 6.

    set name2[1].name1[1].a = 5
    etc...
endfunction

is that legit?

[Dark_Dragon]

Originally Posted by daPanda

thanks again dragon. one more question about that if you don't mine.

What if the extended struct contains another struct? is it also allocated?

JASS:

struct name1 extends array
   integer a
   integer b
endstruct

struct name2 extends array
     name1
     real x
     real y
endstruct

function bla takes nothing returns nothing
     set name2[1].x = 5.
     set name2[2].y = 6.

    set name2[1].name1[1].a = 5
    etc...
endfunction

is that legit?

yes you can have struct inside the struct but this:

JASS:

struct name2 extends array
     name1
     real x
     real y
endstruct

function bla takes nothing returns nothing
     set name2[1].x = 5.
     set name2[2].y = 6.

    set name2[1].name1[1].a = 5
    etc...
endfunction

should be:

JASS:

struct name2 extends array
     name1 n1
     real x
     real y
endstruct

function bla takes nothing returns nothing
     set name2[1].x = 5.
     set name2[2].y = 6.

     set name2[1].n1 = name1(1)
    set name2[1].n1.a = 5
    etc...
endfunction

[Dark_Dragon]

Originally Posted by Vexorian

variables from structs are added after other global variables. No matter if it is a library or a scope or whatever. My solution? Assign the global post- globals init ...

I don't think there is a way to change this rule any time soon, as long as structs are compiled in a separate phase, it won't.

then how are static variables sorted? if i wrote one in library and one in scope will that one in library be declared before that one in scope? or are they randomly added after globals?

btw static variables should be really like they are now coz if they get messed with globals many of constants might not be accessable... and from view we are looking at structs in c++ or some other OOP language constant static variables should be allowed to always be assigned with other global constants...

btw structs here are more like classes in c++! so when i am saying structs i mean classes in c++...

[Element of Water]

Just a little problem with the syntax checker. If you do something like this:

JASS:

struct MyStruct
    local integer i = //syntax error
endstruct

JassHelper doesn't notice this; it throws up a syntax error in the generated allocator function, rather than saying there's an error in the struct. When this happened to me it took me about 10 minutes to work out what I did wrong. Please fix it.

EDIT: I have a request - please make the newline character ("\n") be considered in textmacros, so for example, I could do something like this:

JASS:

//! textmacro MyTextmacro takes NAME, VARS
struct $NAME$
    $VARS$
endstruct
//! endtextmacro

//! runtextmacro MyTextmacro ("Data", "real x\nreal y\nunit u")

which compiles to this:

JASS:

struct Data
    real x
    real y
    unit u
endstruct

[Barade]

Originally Posted by Dark_Dragon

then how are static variables sorted? if i wrote one in library and one in scope will that one in library be declared before that one in scope? or are they randomly added after globals?

btw static variables should be really like they are now coz if they get messed with globals many of constants might not be accessable... and from view we are looking at structs in c++ or some other OOP language constant static variables should be allowed to always be assigned with other global constants...

btw structs here are more like classes in c++! so when i am saying structs i mean classes in c++...

The only difference between structs and classes in C++ is that struct members are public by default. So structs in vJass are like structs in C++.
Btw. I can not assign some constant variables as array member size like bj_MAX_PLAYERS. A better global variable treatment would be useful.

[Dark_Dragon]

Originally Posted by Barade

The only difference between structs and classes in C++ is that struct members are public by default. So structs in vJass are like structs in C++.

i see... well i knew that classes are private (by default) and structs not! however can structs have methodes? in school we used structs only for group of variables. no methodes. anyway ty.

Btw. I can not assign some constant variables as array member size like bj_MAX_PLAYERS. A better global variable treatment would be useful.

well that might be true, well its just since i never needed it so i did not think to much about it. however for some people yeah could need some more work :P

[Alevice]

STructs in c++ can have methods.

Anyway, sucks to hear that Vex :( Oh well, time to redesign then :/