Snippet LinkedListModule

[Sevion]

You should post the errors you got from Alloc in the thread so I can take a look at them.

Sure, you can deallocate whole lists without iterating through it, but I'm still not sold on this as opposed to writing my own Linked List when I need it.

Any Linked List I write can be leaner than this should I ever need to utilize a Linked List.

 

[Dirac]

The errors I get from i Alloc i can solve myself, and thats the beauty of this.

Disclaimer: I don't recommend you to use this system if you plan on having a struct with non-dynamic list (0 being the head) instead code your own.
The whole purpose of this resource is to be able to create multiple linked lists inside the same struct and be able to iterate through them (in this case writing your own list would be difficult)

 

[Sevion]

If there are errors with Alloc, then I need to fix them. Just because you can fix them, doesn't mean I should leave it be.

 

[Dirac]

What i meant is that Alloc helps me see errors i reproduced myself that lead, for whatever the reason, to deallocate the same instance twice.

Updated to 1.2.0
-Renamed the LinkedList module to DynamicLinkedList
-Added the new LinkedList module to support static linked lists

Updated to 2.0.1
-Removed all the modules except for LinkedList
-API overhaul
-Added the "base" static head

 

[tooltiperror]

globals
    constant boolean HEAD  = true
endglobals

local thistype ta = thistype.createNode(HEAD)
local thistype tb = thistype.createNode(not HEAD)  // Awesome verbosity.

> headp
> nextp
> prevp
What does the "p" prefix mean?

I think some ambiguosities should be addressed in the documentation. Namely, how the default head works, and how is [LJASS]createNode[/LJASS] different from [LJASS]allocate[/LJASS]? You're alright for not going over what a list is, people should know if they want to use an implementation of it.


I actually really like this. It's nicer than a regular Linked List in some ways. However, LinkedListModule is a shitty name. It's like naming games for Nintendo 64 stuff like "Mario64" ... "LinkedList" is a better name.

> Disclaimer: I don't recommend you to use this system if you plan on having a struct with non-dynamic list (0 being the head) instead code your own.
Don't you think that should be mentioned in the doc?

 

[Dirac]

Actually since 2.0.1 this snippet suports static lists using thistype(1) as the head and you can refer to it as "base"
[ljass]call base.insertNode(this)[/ljass]
[ljass]local thistype this=base.next[/ljass]

the "p" prefix stands for private, the method operators are there to make those members readonly (inside modules readonly members act as if they were public inside the struct)

I'm not sure about that head thing, the documentation does say "isHead" the user should know what is it about.

The great thing about this version is that it supports 3 types of lists: dynamic, static and circular.

 

[Dirac]

v2.1.0
-"base" now points towards 8190, this is because having (1) as the base made the static list incompatible with other allocators.
-createNode now takes no arguments, nodes created this way are considered heads.

 

[tooltiperror]

>Actually since 2.0.1 this snippet suports static lists using thistype(1) as the head and you can refer to it as "base"
I wasn't really asking that. I was kind of telling you to add in a boolean called HEAD equal to [LJASS]true[/LJASS] so that [LJASS]insertNode[/LJASS] was more verbose.

>the "p" prefix stands for private
It could just as easily stand for "public". I was kind of trying to get you to add a comment.

I think some ambiguosities should be addressed in the documentation. Namely, how the default head works, and how is createNode different from allocate? You're alright for not going over what a list is, people should know if they want to use an implementation of it.


I actually really like this. It's nicer than a regular Linked List in some ways. However, LinkedListModule is a shitty name. It's like naming games for Nintendo 64 stuff like "Mario64" ... "LinkedList" is a better name.

> Disclaimer: I don't recommend you to use this system if you plan on having a struct with non-dynamic list (0 being the head) instead code your own.
Don't you think that should be mentioned in the doc?

 

[Dirac]

Updated to v2.1.2
-"next", "prev" and "head" are now public (v2.1.1)
-The recycle stack now uses prev instead of next as the pointer because this way the user is allow to know which is the node's "next" pointer even after removed and deallocated, and since "next" is the most common way to iterate through a linked list this is very useful. DISCLAIMER: if you pretend to iterate using "prev" remember to use a variable to store the node's previous "prev" pointer before removing and deallocating it.

 

[Dirac]

v2.2.0
-The insertNode method now inserts the node behind the given node (append)

This update "sorta" breaks backwards compatibility, make sure to do the proper fixes.
(Same reason why i'm updating MergeSort just now)

 

[Dirac]

v2.3.1
-Added the new LinkedListLite module to avoid many lines of code from being generated when implementing the module into a struct and not using the methods it provides, it's also slightly faster (but you have to use the textmacros)
-Added a list of textmacros useful to handle linked lists safely when using the LinkedListLite module.

 

[Nestharus]

You forgot to mark B's first node's head to false ;p

    //! textmacro LINKED_LIST_MERGE takes nodeA, nodeB
        set $nodeA$.next.prev = $nodeB$.prev
        set $nodeB$.prev.next = $nodeA$.next
        set $nodeA$.next = $nodeB$
        set $nodeB$.prev = $nodeA$
    //! endtextmacro

And what is this about???

            set thistype(8190).next = 8190
            set thistype(8190).prev = 8190
            set thistype(8190).head = true

 

[Dirac]

Thanks for the fix on the merge i'll implement it right away.

The 8190 instance will never be used (or at least it shouldn't in debug mode) and since 0 is already being used for recycling, i'm turning it into a "base" preloaded head (see the base static operator) for users to have without having to initialize it themselves.
So in case you don't need a dynamic LinkedList you can simply insert nodes into 8190

 

[Nestharus]

I have a better idea... get rid of it ; 0. It's really silly.