Structs for Dummies

Structs For Dummies

In this tutorial, I'm going to try to explain the concept of exactly What a struct is and why we use structs.

Table of Contents

  • What is a struct
  • Why do we use structs
  • Important methods
  • Difference between Static Methods and Methods
  • Struct extends array
  • Struct Variables
  • Thistype Keyword
  • Operators
  • Allocation/Deallocation
  • Modules
  • Initialization
  • Prevent Variable Modification
  • Keyword Keyword
  • Misc Tips

What is a struct
I'm going to be as basic as possible:
  • A struct defines a new type
  • A struct can have 8190 instances (Will explain later)
  • A struct can be treated like any other type (unit, location, etc...)
Just think of structs as virtual objects you can create and destroy.

Why do we use structs
Structs make everything easier.
I'm assuming you're a GUIer. Have you ever created a spell that uses index recycling methods?
If yes, you'd know how unreadable and ugly index recycling methods are.

With structs, you can think of spell instances as simple "objects".
These objects can contain data, and they can manage that data with
functions known as "methods".

Important methods
There are only 5 methods I want you to know:

The Allocate method is used to allocate an instance:
JASS:
static method allocate takes nothing returns <Name_of_Struct>
    // bla bla bla
    // This method is automatically generated in each struct by JassHelper
    // You just need to call it once inside the "create" method.
endmethod

The Create method is like an extension to the allocate method. It calls allocate, then initializes some data.
You have to create this method yourselves.
JASS:
static method create takes nothing returns <Name_of_struct>
    local <Name_of_Struct> this = <Name_of_Struct>.allocate()
    // do some init stuff
    return this
endmethod

The onInit method can be very useful
I'll explain later
JASS:
static method onInit takes nothing returns nothing
    // anything you put inside here will run at map initialization
endmethod

The deallocate method is... I already explained that below :p
JASS:
method deallocate takes nothing returns nothing
    // bla bla bla
    // This method is like allocate, but it deallocates an instance
    // it's automatically generated by JassHelper, so you don't need
    // to write it. Only use it in the "destroy" method
endmethod

The destroy method is a must. You should write it yourselves.
It's like an extension to deallocate. It nulls all the handles in the struct, and deallocates the instance
JASS:
method destroy takes nothing returns nothing
    // null your variables
    call this.deallocate()
endmethod

These are very basic concepts.
destroy is always a method
create is always a static method
onInit is always a static method

There's another known method, but please, if you ever see someone use it, you
should flame them give them a link to this tutorial:

JASS:
method onDestroy takes nothing returns nothing
    // do stuff when you destroy
endmethod

This method is bad because it generates useless functions.
Instead of creating this method that runs when you call destroy,
why don't you just write the destroy method?
Isn't that logical?

As you've noticed, there are 2 types of methods: static methods and methods
If you're curious to know the differences, scroll down.

Difference between Static methods and Methods
Static methods and methods have some differences:

Take the following example:

JASS:
struct Pos
    real x
    real y
    real z
    method moveUnit takes unit u returns nothing
        call SetUnitX(u,this.x)
        call SetUnitY(u,this.y)
        call SetUnitFlyHeight(u,this.z,0)
    endmethod
    static method create takes real X, real Y, real Z returns Pos
        local Pos this = Pos.allocate()
        set this.x = X
        set this.y = Y
        set this.z = Z
        return this
    endmethod
    method destroy takes nothing returns nothing
        // nulling
        call this.deallocate()
    endmethod
endstruct

Difference #1:
The way they're used:
JASS:
function hi takes nothing returns nothing
    local Pos somePos = Pos.create(3,5,2)
    call somePos.moveUnit(GetTriggerUnit())
    call somePos.destroy()
endfunction

As you can see, these are the formats:
Call static method:
call <NameStruct>.<NameStaticMethod>
Call methods:
call <Instance>.<NameMethod>

These are very simple concepts.

Difference #2:
How they work.
Static methods can be used directly referring to the name of the struct.
This means that they have to find the instance they are required to work with INSIDE the struct.
Methods are used referring directly to the instance. This means you don't need to specify the instance since it's already given.

While working with the instance inside a method, always assume it's called this

Static methods are not required to work with instances. They can just perform
simple procedures. On the other hand, since methods must take
instances outside the struct, they are REQUIRED to do something concerning
that instance.

Struct extends array
Here's a simple concept:
If a struct has no methods, just make it extend an array.
You can make absolutely ANY struct extend an array, but that would
require you to implement a dynamic indexing and recycling snippet.

Example:
JASS:
struct someObj extends array
    static unit u
    static unit v
endstruct

Structs that extend arrays are good because normal structs generate lots of trash code.
A struct that extends an array doesn't have an allocate/deallocate method.

To learn about writing efficient structs with dynamic indexing and recycling, click here.

Struct Variables
There are 2 types of variables in a struct:
Static variables and normal variables.

The differences between static variables and normal variables are exactly
the same as the differences between static methods and normal methods.

One new and important difference is that static variables are not unique to
each struct instance. In the following struct:

JASS:
struct bla
    integer i
    static integer j
endstruct

Changing the variable j will affect ALL instances of the struct, while changing the variable i will only affect one instance.
You can think of j as a global that can only be accessed through the struct. (Because it is)
It isn't associated with any specific instance, but the actual struct itself.

Thistype keyword
What?​
Inside a struct, instead of referring to the name, you could just use the keyword thistype
Why?​
Because highlighting is awesome :ogre_datass:
Also, when you read about Modules later in this tutorial, you'll know that using this keyword​
is important because you're not always going to know the name of the struct that the module​
is going to be implemented into. If you don't get it, then come back to this section after you​
finish reading about modules, and I can guarentee that you'll understand it.​



All of the very basic concepts have been covered.
If you're not THAT much of a dummy, you may continue to the bottom.




Operators

I'm sure you've seen people do something like this:

JASS:
static method operator [] takes <something> returns thistype
    return <Some_data_reffering_to_an_instance_of_the_struct>
endmethod

Static method operators have 4 types:

[] -> Takes a variable inside the brackets
[]= -> Takes a variable inside the brackets and a variable after the "=" sign
== -> Takes a different instance, and allows you to return a boolean for the evaluation
< -> Same concept as the above

Assume the above static method operator is used:
JASS:
function kill takes nothing returns nothing
    local <Name_of_Struct> something = <Name_of_Struct>.allocate()
    call <Name_of_Struct>[GetUnitId(GetTriggerUnit())].killHim()
endfunction

GetUnitId is nothing really important for you to know about right now (It's a function in the UnitIndexer library)
killHim happens to be a method. You may have noticed that we referred to it using the Name of the Struct. This is because the static method
operator returns an instance of the struct.

These operators allow us to use not only integers, but units, locations, handles, etc....

This makes them VERY useful.

Method operators are the same as static method operators.
The differences between method operators and static method operators
are the same as the differences between static methods and methods.

Here are some examples:

JASS:
struct Unit
    unit u
    method operator hp takes nothing returns real
        return GetWidgetLife(this.u)
    endmethod
    method operator hp= takes real r returns nothing
        call SetWidgetLife(this.u,r)
    endmethod
endstruct

You're probably asking this: Why don't we just use variables instead of method operators?

Keeping a variable like hp inside the struct and constantly updating it is really stupid.
The method operator allows you to directly return the "GetWidgetLife" native.

"hp=" is a useful method operator too.

The above example is basically a simple API (Application Programming Interface)
Using things like "hp" or "hp=" makes your code look pretty good :p

Static variables along with static method operators are VERY useful:

JASS:
struct someOtherThing
    unit u
    static integer array index
    static method operator [] takes unit v returns thistype
        return index[GetUnitId(v)]
    endmethod
    static method create takes unit v returns thistype
        local thistype this = thistype.allocate()
        set index[GetUnitId(v)] = this
        set this.u = v
        return this
    endmethod
endstruct

Just assume the presence of GetUnitId
This small template can be very useful when you're referring to instances
in a struct that are unique to each unit on the map.

[]= is also a very important operator.
Look at this Unit Data Struct:

JASS:
struct UnitData
    unit u
    integer data
    static integer array index
    static method operator [] takes unit v returns thistype
        return index[GetUnitId(v)].data
    endmethod
    static method operator []= takes unit v, integer i returns nothing
        set thistype[v].data = i // As you can see, thistype[v] returns an instance of the struct :)
    endmethod
    static method create takes unit v returns thistype
        local thistype this = thistype.allocate()
        set index[GetUnitId(u)] = this
        set this.u = v
        set this.data = 0
        return this
    endmethod
endstruct


Allocation/Deallocation
This information can be very useful.
As I've said before, JassHelper generates 2 methods:
static method allocate takes nothing returns Name_of_Struct
and
method deallocate takes nothing returns nothing

Sometimes, we want to know if a certain struct instance already exists.
If yes, most of the time, we wouldn't want to create it again.
But how can someone detect a null instance?

Easy: A null struct instance equals 0.

Consider the following struct:

JASS:
struct Blekh
    static integer array index
    static method operator [] takes unit u returns Blekh
        return index[GetUnitId(u)]
    endmethod
    static method create takes unit u returns Blekh
        local Blekh this = Blekh.allocate()
        set index[GetUnitId(u)] = this
        return this
    endmethod
endstruct

Assume each unit has a specific instance of this struct.

In order to check if an instance for that unit exists, we do this:

JASS:
if Blekh[unit] == 0 then // is it null?
    call Blekh.create(unit)
endif

Simple, huh?

You can deallocate instances by setting them to 0, but that's really bad since the instance won't be recycled :)
Never do that. I repeat, NEVER do that!

Sometimes, instead of deallocate/allocate, you could create your allocation/deallocation methods :)
Here's an example: Alloc by Sevion

Modules
This concept is very simple:
A module is just a piece of a struct...
It can contain methods, static methods and variables.
Example:
JASS:
private module SomeMod
    unit randomDude
    unit otherRandomDude
    static method onInit takes nothing returns nothing
        set randomDude = ...
        set otherRandomDude = ...
    endmethod
endmodule
private struct someThing
    implement SomeMod
    static method create takes unit u, unit v returns <Name_of_Struct>
        local <Name_of_struct> this = <Name_of_struct>.allocate()
        set randomDude = u
        set otherRandomDude = v
        return this
    endmethod
endstruct

As you can see, modules are 'implemented' into structs.
You're probably asking yourself: "So?"
Well, when a module is implemented into a struct, all of it's contents are copied into it.

Modules have several uses. One very important and common use is Initialization:

This is the format of Module Initialization.

JASS:
private module Init
    private static method onInit takes nothing returns nothing
        // some Init stuff
    endmethod
endmodule
private struct Inits extends array
    implement Init
endstruct

Initialization
Take:

JASS:
library temp initializer Init
    private function Init takes nothing returns nothing
        // bla
    endfunction
endlibrary

and

JASS:
private module Init
    private static method onInit takes nothing returns nothing
        // some Init stuff
    endmethod
endmodule
private struct Inits extends array
    implement Init
endstruct


What's the difference?
These 2 initialization methods run in different orders.
Using the second sample as a template is a requirement if you're initializing important variables because it is executed first.
In libraries like Ascii, using the second method is important because if the functions included were called
during map initialization and the library used a common initializer, null values would returned. Anytime you have a library
that requires you to configure variables like in Ascii, use the second method. The first method can be useful for initialization
that involves creating a trigger, registering an event, registering a condition, etc...


Preventing Variable Modification

If you want to prevent people from modifying variables outside of a given struct, just use the readonly keyword

This will prevent variables from being modified OUTSIDE the struct. I repeat, OUTSIDE!!

JASS:
struct HI
    readonly static integer i
    static method create takes nothing returns thistype
        local thistype this = thistype.allocate()
        set thistype.i = 2 // this is ok
        return this
    endmethod
endstruct

function hi takes nothing returns nothing
    set HI.i = 4 // This will popup a compile error
endfunction

The difference between readonly and private is quite simple:
- readonly prevents modification outside the struct
- private prevents both modification and access outside the struct

Keyword Keyword
Did you know there's actually a keyword keyword?
It's important because it allows us to access functions and variables regardless of the position in the library.

For example:

JASS:
function hi takes nothing returns nothing
    call A.create()
endfunction

struct A
    static method create takes nothing returns thistype
        return thistype.allocate()
    endmethod
endstruct

This would cause a syntax error since the struct A is positioned below the function.
But, if we were to do this:

JASS:
private keyword A

function hi takes nothing returns nothing
    call A.create()
endfunction

struct A
    static method create takes nothing returns thistype
        return thistype.allocate()
    endmethod
endstruct

Then the syntax error would no longer popup!

Misc Tips
  • Always destroy struct instances since they're limited
  • Structs can extend other structs, but if you do that, then those structs will share the same instances and thus, you have a greater chance of reaching the 8190 limit
  • If Struct A extends struct B, then struct A will be able to use methods and variables inside struct B (except private ones)
  • Struct instances can be treated EXACTLY like integers
  • Always use thistype inside a struct instead of the name
  • Timers can only have static method callback functions, making TimerUtils a very useful library. You can use SetTimerData to store the "integer"/"instance" in the timer to load in the static method.
  • Always use module Initializers for standalone libraries with important data.
  • Always use "this" to refer to the instance inside a method
  • Always remember to deallocate instances in your destroy methods.
  • Structs are compiled to arrays. This is why they can only have 8190 instances at once.
  • 0 refers to a null struct instance


********************************************

There are lots of other things for you to know about structs.
This tutorial only covers the basic concepts. There are other
tutorials involving structs, but this one is particularly aimed at
Dummies.

If you feel that this tutorial isn't sufficient, feel free to post.

Here's a list of things I'm intending not to cover:
- Delegates
- Super
- Interfaces
 
Last edited:
Level 31
Joined
Jul 10, 2007
Messages
6,306
Hmm.. you should cover allocation/deallocation and checking for allocation techniques as well as how custom allocation can avoid trash code generated by vJASS. Put this towards like more advanced techniques.

Linking to Sevion's Alloc module could be helpful

Explain why library initializers are horrible. They aren't horrible =P.

Just cover order of initialization >.>.

You also forgot to mention readonly and delegates

And this isn't true
Method operators can never be [] or []=

You also forgot about the like >, +, <, etc operators. Generate an operator vJASS error to get a list of all of the operators.

That's all for now.
 
You also forgot to mention readonly and delegates

DUMMIES don't need to know about those =P
This tutorial covers what a person should know before he starts.
He'll come across the rest through trial and error.
I'll add "readonly" though :)

Thanks for the comments :)
And btw, ->

Hmm.. you should cover allocation/deallocation and checking for allocation techniques as well as how custom allocation can avoid trash code generated by vJASS. Put this towards like more advanced techniques.

That can be added in by saying: How to detect a null instance: Struct_Instance == 0 -> it is null :p
I'll link to some resources and improve this tutorial :)


edit
Updated
 
Last edited:
Nice tutorial. However, I have a few small comments on certain parts.

Changing the variable j will affect ALL instances of the struct, while changing the variable i will only affect one instance.

To clarify, the variable j is essentially a global when it has the static keyword. It isn't associated with any instances (opposed to the variable i), but rather it is associated with the struct itself. (meaning, you have to use structName.var to access it [at least outside the struct])

For the keyword thistype, you may want to mention that it is mostly useful for modules. Since modules are implemented into a struct, and you don't necessarily know the name of the struct that it is being implemented into, you can use the keyword thistype to refer to the current struct. Although, it is true that it became a habit in normal structs, most likely because of the highlighting. :D

For modules, you probably want to mention that it will also copy the module's contents into the struct. You jump into talking about how it is used for initialization before saying how to use it in general. ;P

If you want to prevent people from modifying variables outside of a given struct, just use the readonly keyword

Well, there is also the keyword private. You might want to explain the difference between private and readonly. (private prevents both access and modification outside the struct, while readonly only prevents modification)

Otherwise, it seems like a nice tutorial. I like how it is very light, and gives a nice introduction to structs. :) Fix those stuff, and it should be ready for approval. Sorry about the delay.
 

BBQ

BBQ

Level 4
Joined
Jun 7, 2011
Messages
97
There is no such thing as onCreate method, so you should remove that.

You can also mention the use of keyword for getting library-private struct members.
 
There is no such thing as onCreate method, so you should remove that.

Really? I always see people using it :L
Well, If you say so... I'll remove it..

You can also mention the use of keyword for getting library-private struct members.

Yeah, I'll add that in.


edit

UPDATED!
 
Last edited:
I wrote this in 2011.

It's 2015 and I can't believe some of this bullshit I put together.
You should really rewrite that. There's just a lot of "absolutes" in this tutorial that you should either explain or simply not cover in your tutorial.

For example, using the onDestroy method over destroy. You describe that as if using onDestroy is ALWAYS dumb, but there is a reason why it exists, which is struct extension. Using onDestroy will work properly on extended structs, whereas overwriting destroy will break inheritance control, unless you implement that manually.

In fact, you should teach newbies to structs to use onDestroy over destroy, simply because onDestroy universally works, whereas rewriting destroy can break some stuff if you don't know what you're doing.
 
Top