Ideas for a tactical battle system.

ID:153331
 
May 20 2004, 3:24 pm
By "tatical battle system" I mean a turn based battle between two teams on a grid(ex: Shining Force, Final Fantasy Tactics, and more recently Gladius[which I would recommend to anyone who's a fan of this battle system]). I only call it a tactical battle system because all the games I've seen with the word "tactics" in the title uses one. Anyway, I was toying around trying to make one(without planning ahead as usual) and while it probably would have worked if I kept at it, it wasn't very desirable.

So I need some ideas on how to go about making this work cleanly. In my first try I just had one big looping proc that was called to start a battle. That seemed like a good way to go until I got to turns. When a players turn starts I want to give \him some commands, and allow them to move around in a designated area without a time limit. I did that in a proc(TakeTurn(), for future reference) and had the main loop wait for a return value, but finding out when the turn was over is what seemed wrong to me. The only way I could think of to do it, was to have TakeTurn() loop every second and check a flag variable that is set when they finished there turn.

That's my main concern about this, but anything related to making this type of battle system should help out. I'm limiting the players movement to a diamond shape instead of a square(thanks to a snippet posted by Shadowdarke a while ago) and it's working as expected. I haven't tried to do anything to adjust the movement range to account for densities yet, so I might as well get some opinions on that as well. Examples:
 Fig.1    Fig.2    Fig.3
   o        o        
  ooo      ooo      o o
 ooooo    ooDoo    ooDoo
ooooooo  ooooooo  ooooooo
 ooooo    ooooo    ooooo
  ooo      ooo      ooo
   o        o        o

o = Movement range
D = Density
This is with a movement range of 3.


This should explain what I intend to accomplish. I'd like to take any turf that is dense(or has something dense occupying it) out of the movement range.(Fig.2) Then I would like to remove any other turfs they couldn't get to with a certain number of steps(no diagonals).(Fig.3) Should I try to use some math(wich I'm not great at =p), a mix of get_dir() and get_dist(), or something else entirely to achieve this? Thanks for reading, and replying if you see fit. Now, I leave you with another example of the movement range idea. :p

    o          o         
   ooo        ooo        o o
  ooooo      ooooo      ooooo
 ooooooo    oooDooo    oooDooo
ooooooooo  ooooooooo  ooooooooo
 ooooooo    ooooooo    ooooooo
  ooooo      ooooo      ooooo
   ooo        ooo        ooo
    o          o          o

This is with a movement range of 4.
May 20 2004, 7:07 pm (Edited on May 20 2004, 7:22 pm)
I made a tactics style range system before. It more or less traced the steps while keeping a list of checked turfs in a list. I'll see if I can dig it up.

[EDIT] Here it is in it's full, uncommented form.

turf
    var
        cost = 0

    proc
        queryCost()
            return queryElevation()

        queryElevation()
            return 0
datum
    pathobj
        var
            cost
            x
            y
proc

    cardinal(atom/ref)
        return (list(get_step(ref,NORTH),get_step(ref,SOUTH),get_step(ref,EAST),get_step(ref,WEST)))
        
    tactics_dist(atom/ref, atom/targ)
        return sqrt( (ref.x-targ.x)**2 + (ref.y-targ.y)**2)

    tactics_orange(dist as num, downhill=FALSE, atom/ref = usr)
        return tactics_range(dist,downhill,ref)-tactics_range(0,downhill,ref)
        
    isdense(atom/A)
        var/retval = 0
        retval += A.density
        for(var/atom/B in A.contents)
            retval += isdense(B)
            
        return retval
                

    tactics_range(dist as num, downhill=FALSE, atom/ref = usr)
        var/list/turfs = list(ref.loc)
        var/datums[world.maxx][world.maxy]
        var/datum/pathobj/refloc = new()
        refloc.cost = 0
        refloc.x = ref.x
        refloc.y = ref.y
        datums[ref.x][ref.y] = refloc
        for(var/x=1, x<=dist,x++)
            for(var/turf/T in turfs)
                for(var/turf/t in cardinal(T))
                    if(isdense(t))
                        continue
                    var/cost = 1+t.queryCost()
                    var/datum/pathobj/p = datums[T.x][T.y]
                    cost += p.cost
                    var/elevdiff
                    var/turf/tref = locate(p.x,p.y,ref.z)
                    elevdiff = t.queryElevation()-tref.queryElevation()
                    if(!downhill)
                        elevdiff = abs(elevdiff)
                    cost += elevdiff
                    if(cost <= dist)
                        var/datum/pathobj/np = new()
                        np.cost = cost
                        np.x = t.x
                        np.y = t.y
                        if(datums[t.x][t.y])
                            var/datum/pathobj/test = datums[t.x][t.y]
                            if(test.cost > cost)
                                datums[np.x][np.y] = np
                                turfs.Add(t)
                        else
                            datums[np.x][np.y] = np
                            turfs.Add(t)
        return turfs


    xrange(dist as num, atom/ref = usr)
        var/list/turfs = list(ref.loc)
        for(var/x=1,x<= dist, x++)
            for(var/turf/T in turfs)
                if(tactics_dist(ref,T) < dist)
                    for(var/turf/T2 in cardinal(T))
                        if(tactics_dist(ref,T2) <= dist)
                            turfs.Add(T2)

        return turfs


If some of the procs look familiar, they are. The procs cardinal() and xrange() are pulled straight from Abyssdragon.Basicmath, with minor to no modifications. I found it easier to use a couple of procs rather than use the full library as I was planning on releasing this as a library at some point (with full credit mind you). I believe that the formula for tactics_dist() was also pulled from Basic Math.
May 20 2004, 10:02 pm
An easy way to advance turns is to just call the turn advancement proc whenever a player's turn ends.
May 21 2004, 7:30 am (Edited on May 21 2004, 9:19 am)
In response to sapphiremagus
sapphiremagus wrote:
var/list/turfs = list(ref.loc)

I never thought of doing something like that, go fig.
May 21 2004, 8:32 am
In response to Goku72
Why do you use "Reply with quote" and leave the entire original post in there when your comment pertains to only one line, and even then you resaid the line in your post? I have seen others do that too, and I am just curious why. Is it habbit to reply with quote all the time so you can see it while you are typing the reply, then you just leave it there?
May 21 2004, 9:18 am
In response to Loduwijk
Meh, I some times hit it on accident and don't even realize it. >.>
May 21 2004, 2:48 pm
I, too am working on a "tactical" battle system, though I haven't quite got around to movement. I have procrastination problems, so the hardest things always seem to come last. I was planning on simply restricting movement to a square, but your efforts have encouraged me to try a little harder. I hope you have fun building your game, YMIHere.
May 24 2004, 8:02 am
Thanks for the input guys, I put this off for now to work on some Single Tile Mini-games for the contest, but I'll get back to it and upload a demo when I'm finished. =)
May 24 2004, 8:06 am
In response to MetalMan42
I'm glad I could be of assistance. =)