TerrainInfection v2.1

[Moderator]

TerrainInfection v2.1 | Reviewed by BPower | 07.07.2015 [COLOR="gray" [COLOR="gray" [COLOR="gray" [COLOR="gray" Concept[/COLOR]] Create "terrain infections" starting from a given point x/y. As time goes by the infection eventually spreads to nearby tiles. Maximum size and infection speed can be customized. Make use of infected tiles via isTileInfected(x, y) function. Definitly a useful, unique system. Due to it's innovativeness I will rate the concept with 5/5 Code[/COLOR]] The system is MUI, leakless and working. Nothing I could wish more. Demo Map[/COLOR]] The demo is done very well. Rating[/COLOR]] CONCEPT CODE DEMO RATING STATUS 5/5 5/5 4/5 5/5 APPROVED

 

[GywGod133]

interesting.....

 

[edo494]

you mention bunch of methods, but the API never states the name of the struct.

You also potentially generate TriggerEvaluation with your layout of TerrainInfection struct.

This is because method create is way above method callback, and you still start timer in create and try to call callback from it.

Also callback should be private.


Also maybe you could make it possible to have different timeouts for each infection instance, so you can have one spreading faster than another, but this is just suggestion.

 

[IcemanBo]

you mention bunch of methods, but the API never states the name of the struct.

Done.

You also potentially generate TriggerEvaluation with your layout of TerrainInfection struct.

This is because method create is way above method callback, and you still start timer in create and try to call callback from it.

Thank you, I did not know that inside structs. Changed.

Also callback should be private.

Done.

Also maybe you could make it possible to have different timeouts for each infection instance, so you can have one spreading faster than another, but this is just suggestion.

It actually is already. (interval) Thank you!

 

[Laiev]

OMG :O You just did the Creep Tumor from Starcraft :O I love it, definitely.

A little bit of adaptation and it become exactly how it is in Starcraft, based on range.

Nice work

Edit: If someone didn't know what is Creep Tumor from Starcraft, take a look at the image.

Tumor

 

[IcemanBo]

Ah yes, you are right. ^^ Thank you.

You could use this method, to create a limitation for a new builded structure, so it would only infect until a certain area:

/**      method setRegion takes region returns nothing
**          Infection will be limited to this region.

Of course the user would have to define the region by himself, for each building type individually.

 

[Zeatherann]

This is nice. I created a blight system myself, though it solely mimics Blizzard's blight spreading (the default spell can be edited to slow it down to look like Starcraft I's creep).

 

[BPower]

IsPointInfectedXY takes real x ,real y returns boolean

 

[Zeatherann]

Why do you have all the setter and getter functions?

 

[IcemanBo]

IsPointInfectedXY takes real x ,real y returns boolean

I would need to take track of each tile that was ever infected. :/
At the moment an instance gets destroyed, once there is no infectable tile around it, because it's not needed anymore.
Would you have an other idea?

But anyway I'm not very happy at the moment with not tracking them at all, because user has no controle of infected tiles and it's behaviour, once they are destroyed. So you can't post-change the behaviour, but only after creation. :S

Why do you have all the setter and getter functions?

Do you mean to change some readonly members to accessible members, instead of the setters?
What getters do you mean?

 

[Zeatherann]

method setChance takes real chance returns nothing
            set infectionChance = chance
        endmethod
        
        method setTerrainType takes integer t returns nothing
            set terrainType = t
        endmethod
        
        method setInterval takes real interv returns nothing
            set interval = interv
        endmethod
        
        method enable takes boolean b returns nothing
            set enabled = b
        endmethod
        
        method setMaxX takes real r returns nothing
            set maxX = r
        endmethod
        
        method setMinX takes real r returns nothing
            set minX = r
        endmethod

        method setMaxY takes real r returns nothing
            set maxY = r
        endmethod

These are what I'm referring to.

 

[Spellbound]

Preface: I don't understand JASS or vJASS

This looks really interesting. How would I use it to make a building spread sand in a 600 aoe around it, and then stop spreading sand once it dies/uproots?

 

[IcemanBo]

Preface: I don't understand JASS or vJASS

This looks really interesting. How would I use it to make a building spread sand in a 600 aoe around it, and then stop spreading sand once it dies/uproots?

Sorry, I completly forgot about your post and this resource. Updated and added this distance limitation. In the demo, there is an example.

I still don't infinitly keep tracking of all tiles and lose controle of the behaviour once it gets destroyed, it's kind of discouraging...

Well, but it still may be used in some scenarios.

 

[Heinvers]

Finish this please.

I want creep in wc3

 

[Kazeon]

• SquareRoot((xNew-rootX)*(xNew-rootX) + (yNew-rootY)*(yNew-rootY)) <= rootDistance
  =>
  (xNew-rootX)*(xNew-rootX) + (yNew-rootY)*(yNew-rootY) <= rootDistance*rootDistance
• You don't need this:
  

          private constant integer INDEX_START = -1
          private integer maxIndex = INDEX_START

• You can declare (2*bj_PI) once as "TAU". And another crap calculation like PI/2, rename it to whatever you want.
• You can shorten your infect method to:
  

          private method infectNearbyPoint takes real x, real y returns boolean
              local real angle = 0
              local real xNew
              local real yNew
              @local integer counter = 0@
              local real offset
              local boolean diagonal = true
              
              loop
                  if (diagonalInfection) then
                      // For 45/135/225/315 degrees the centre
                      // of the tile has an other distance, so
                      // we check if it's a digonal.
                      set diagonal = not(diagonal)
                      if (diagonal) then
                          set offset = TILE_DIAGONAL_LENGTH
                      else
                          set offset = TILE_SIDE_LENGTH
                      endif
                      set angle = angle + bj_PI/4
                  else
                      set offset = TILE_SIDE_LENGTH
                      set angle = angle + bj_PI/2
                  endif
                  
                  set xNew = x + offset * Cos(angle)
                  set yNew = y + offset * Sin(angle)
                  
                  if (isPointValid(xNew, yNew)) then
                  
                      //  A new tile gets infected.
                      return copyData(thistype.create(terrainType, xNew, yNew, infectionChance, interval, infectionEffect))
                  endif
                  
                  exitwhen angle > (2*bj_PI) // Full circular check was done.
              endloop
              return false
          endmethod

  And it doesn't need to take another x and y, it's a non static:
  

  private method infectNearbyPoint takes @real x, real y@ returns boolean

  EDIT: and you don't need local variable "counter".

•             if (interv < TIMEOUT) then
                  set interval = TIMEOUT
              else
                  set interval = interv
              endif

  Ermmm......
• You forgot to remove not null infectionRegion on destroy method.

You made this very effective. Max infection range and specific infection region was a very diligent feature, I like it.

 

[IcemanBo]

Thanks! I will think of it in the further update.

Sorry, I forgot what in the destroy method?

 

[Kazeon]

Sorry, I forgot what in the destroy method?

when infectionRegion is not null, you need to remove it: call RemoveRegion(infectionRegion)

Also I have edited the last review

 

[Deleted member 219079]

/*     Requires:
**    ----------
**
**       - nothing*/

Hehe I'm also working on vanilla JASS spell, on which I won't use any external scripts.

I think I have use for this, waiting for status change.. +rep

 

[IcemanBo]

^Thanks.

It can be used actually, but I'm not satisfied enough with functionality at the moment. So I will work on it some later. (probably not soon)

 

[IcemanBo]

Update.

Edit:

Another update is in progress.

Edit 2:

Updated

The infection can now shrink back again until completly removed.

 

[Almia]

ooooh!

IT's like the Amoeba Behavior in Game of Life!

[edit]

Haven't tested it yet. It seems that it is more of the Virus Powder in Dan Ball's Powder game

 

[IcemanBo]

Had a useless method in code. It is removed now, and some of doc is changed.

@Almia idk your named things - but I wanted it to be able to immitate things like this creep terrain in starcraft.^^

 

[Almia]

I'm just predicting things :v. You might wanna search game of life

 

[BPower]

Code is good, however does not completly follow the JPAG.

We didn't approve Vector from the JassSection yet, but apart from that
consider TerrainInfectio as already approved.

 

[BPower]

Wouldn't it be better to use a second struct with a linked list to store
all terrain types and terrain ids? Something like

private struct TerrainList extends array
    implement List// Whatever list module you prefer, or hand-written ...
    integer terrainId
    integer terrainType
endstruct

Vector<T> from the submission section does currently not even compile.

 

[IcemanBo]

In my callback I mistakely only make:
local integer counter = 0,
but should re-assign it with 0 for each instance in each iteration.

Can't update the map atm, so it probably should not be approved for now.
And anyway yes, if lfh seems to have potential issues with Vector, let's wait.

 

[BPower]

I would recommend to use a linked list over Vector, simply because we have approved
list modules.

private struct InfectionList extends array
    implement List //Any list module or write it yourself.
    integer originalTerrainId
    integer infectedTileId
endstruct

 

[IcemanBo]

Heyho!

I have my editor back, so updated it now.
So it uses list now over vector, 'cause not approved.

Also some changes here and there, and demo map changed a bit.

 

[BPower]

to exit the loop within a linked list you could also use
exitwhen element == 0 over i < 1

Normally it looks like

set element = this.list.first
loop
    exitwhen element == 0
    // do stuff
    set element = element.next
endloop

Everything looks good to me. I will check the demo map later.

 

[IcemanBo]

Usually, yes, but the list may extend during iterations, so
that means new list elements would instantly be iterated in
the very same timer call back which is not wanted.

 

[BPower]

Ok I see.

 

[Zwiebelchen]

What are the limitations on SetTerrainType, btw? I think you tested this already, right?

Does it require to use a tile that is loaded into the map or can it access any tile in the WC3 database?

 

[edo494]

any tile but only up to 16.

Each referenced tile instantly takes 1 space in the 4 bit table

 

[IcemanBo]

Achso, sorry for late response Zwiebelchen.

Yes, any tile, but do not use more than max tile limit. I thought 15, but yes 16 might be right edo.

 

[Statharas]

Finish this please.

I want creep in wc3

It's called



Blight.

 

[Lt_Hawkeye]

Since I'm not a jasser or vjasser, I am wondering how would I go about destroying a specific tile after checking that the tile is indeed infected?

 

[Pyrogasm]

Well, the TerrainInfection.isTileInfected method allows you to check if a certain tile is infected or not, but there's actually nothing you can do with that information. @IcemanBo has not provided a method to get the infection that 'owns' that tile or a method to clear the infection from a particular tile... and has made the Table objects used to store this information private (can't be accessed outside the library). It is possible to add such functionality to his system (as I see it) but rather than edit it and post it for you I'd simply ask that he do it himself. Bo is generally pretty good about updating his resources!

static method isTileInfected takes real x, real y returns boolean
     return thistype.table.has(GetTileId(x,y))
endmethod

 

[IcemanBo]

how would I go about destroying a specific tile after checking that the tile is indeed infected?

Yes, seems currently like Pyrogasm correctly said:

provided a method to get the infection that 'owns' that tile or a method to clear the infection from a particular tile... and has made the Table objects used to store this information private

.. iirc the plan was to provide some easy way backwards from infection, hence only the list with allowing to undo the last infection, or just all. 1 tile can be multiply infected, so allowing to uninfect 1 tile only by coordinates one still would need know, or to specify the infection that started that now has to be undone, and this would require new listing for each tile for all infections.
Long story short, we can make some sloppy approach for removing one specific tile from a infection list by coordinates:

        method undoInfection_XY takes real x, real y returns boolean
            local InfectionData element = this.list.first
            local integer id = GetTileId(x, y)

            loop
                exitwhen element == 0
                if (element.tileId == id) then
                    call SetTerrainType(GetTileCenterXById(id), GetTileCenterYById(id), element.originTerrainType, -1, 1, 1)
                    set this.size = this.size - 1
                    call thistype.table.remove(id)
                    call element.remove()

                    set thistype.current = this
                    set thistype.currentTileId = tileId
                    set thistype.infection_state = 3
                    set thistype.infection_state = 0
                    set thistype.current = 0
                    set thistype.currentTileId = 0
                    return true
                endif

                set element = element.next
            endloop
            return false
        endmethod

^.. add this method to the main code API, and then call it like call this.undoInfection_XY(100, 100) (for coordinates x/y = 100 for example)
But as said, the list, here the this is required. You can make the list itself public, too, but I believe this is more dangerous for you in case you don't know how the list logics internally works.

I could do some update, allowing to maintain all infections for a tile in a list, or vector, providing it to the user. But for this, I would wait until new wc3 patch, as I currently don't use some external program to code, and editor is atm not too cool to code with.

In case you lean more on the "GUI" demo, there is currently not a too nice way to achive the same thing without custom script, I'm afraid. But if you need GUI, I probably can switch the system code for your map to only allow single infractions for 1 tile, and then you could maybe just define 1 tile in GUI, and then run some un-infection trigger, if you want.

===

Actually after more reading the system... the code seems not perfectly consistent with single, or multiple infections, and I maybe should by default just allow only 1 infection, and just map the infection directly to a tile. Then there can't also a problem occur with losing "original" terrain type. I will maybe do this, when editor is updated -- but still maybe something from above helps.

 

[Spellbound]

So I've been attempting to use this system and while it works as advertised, there are a few issues I'm running into. For one, with high enough numbers, new TerrainInfection instances just seem to stop working. I don't have a specific number, but in a small map like Turtle Rock, two players building a bunch of food-providing structures that also spread the Infection will quickly run into that problem with one 1 expansion each. I'm not discounting user error, but I'd mention it here just in case.

Additionally, undoing infection will cause tiles that are being infected by two or more sources will also get removed. Ideally, overlapping Infections should only clear once all contributors are destroyed. This would allow me to stop the spread when it completes so that it stops checking nearby tiles when it doesn't have to.