User:Duerig/Archive2

From RogueBasin
Jump to navigation Jump to search

How to write a roguelike gameengine - Esa Ilari Vuokko [eivuokko@mbnet.fi].txt

Some ideas that could be useful ? As shared by Esa Ilari Vuokko. Comments and bugfixes ;) to eivuokko@mbnet.fi


     I've played roguelike games for 7 years now. First I hacked Moria,
  which I got from friend (no modem or net connection :(). Then I got
  Omega which I still play (newer version, thought ;). At that time I
  got rid of Basic I was programming with and got Turbo Pascal. Well,
  I tried to do some pathetic games myself but they were just dead
  as they were real mode programs. Then (not earlier) I got nethack
  which I didn't like at that time. I hated djgpp (it's not bad, it's
  ugly in msdos) so I was bounded to real mode. Then I got Linux and
  installed it few times with no success (one time I installed it on
  broken hd, guess did it work, no - it just paniced suddenly ;). And
  I played ADOM, a lot. And then little Nethack and Omega.


     Because Linux I got all so mighty gcc and make. After playing for a
  while with graphics stuff (in linux and in win) and in winenv doing
  some accounting program (with delphi, thought I quitted after it
  started compiling wrong - I mean(debugged) it) I got an idea. In an
  accounting program we move money from one account to another (I'm sorry
  I know finnish terminology better than english). Well those accounts
  must be linked list because : they must be easy to create, modify and
  delete. Nothing new - a linked list. Of course all objects in rg should
  be in linked lists. But the what if even attributes were linked lists ?
  So that we had a linked list which can store data (some 16 bytes is enough)
  and a name (string) and children list.


     Listing attributes, skills, and everything that describes object with
  such presentation would be quite flexible. But if one does such game,
  and many people contribute to it ? It can go way off road as everybody
  add something new. A bloated memory use, and unneeded complexity are
  real threats. As normally, say ADOM, player char needs more memory
  (more variables) than npc. But in this approach we give npc only those
  skills and attributes it needs and we are still able to use same routines
  on both, player and npc.


     Of course above system can be optimized. I've defined it like this :
  Okey, I've got a bit more complex.
        class List {

private: List *child,*next,*parent; int id;

                int data[4];
                Link(List *);
                Unlink(List *);
                Query(int,int,int,int,int);
          public:
                int *Data();
                char *Name();
                int *Data(char *);
                List *Name(char *);
                int *Data(int,int,int,int,int);
                List *Name(int,int,int,int,int);
                Item *Index(int);
                int Index(Item *);
                int Count();
                List *Add();
                List *Remove();
        } ;
     Quite clear, I think. Id can be converted (by another class) to char.
  Almost all ints are for quick query. I use dot as delimiter between child
  and parent and I don't have plurals.  "skill.weapon.blade.short".
     Then I wanted an engine that doesn't need special cases. It would be
  (sorry for saying this) stupid to do special levels, which have special
  if of switch statement in level code. How could one have such a really
  flexible system ? Well I read Thomas Biskup's ideas of JADE. I don't know
  whetever he meant the same as I with the mentioning everything to be
  event based. So what I'm chasing here is that every object should have
  message handler list, which would handle requests to do something.
    Say we have a character A. A has handler list of next functions
  (stacked):
      creature_shield_deflector,
      creature_ro_poison_res,
      creature_basic_poison,
      player_non_ai,
      basic_creature_handler.
  First a monster(mage) B would shoot an firebolt to A. Firebolt code would
  send a message to A that some fire damage is  coming. First this message 
  would reach first handler in list, creature_shield_deflector. That would 
  say "no,no fire damage" and that would be it, no damage to creature. Then
  B would throw a acidbolt. Well this time deflector would say nothing as 
  wouldn't other before basic_creature_handler. That would make some damage 
  to creature and spare some to inventory too (and send approciate messages 
  to items). Then would engine send TIMEUPDATE to creature, which would be 
  handled first by creature_basic_poison. Well it seems this creature has 
  poisoning and handler would notice that it just ending. First it sends to 
  creature A (self) damage message of poison, which would end to 
  creature_ro_poison_res, and no damage. Then it would return 
  UNLINK_AND_CONTINUE so that it would be removed from list and message 
  (TIMEUPDATE) would be sent on. Then message would reach basic_handler which
  would add some speed to energy (ADOM) or add a timepulse (nethack,angband).
  If it would be creature's turn to move, it would send message ACTION to 
  itself (creature A). That would be caught by player_non_ai which would wait 
  for keypress and then do whatever is defined and player wishes to do.
     Because paragraph above seems a bit unclear ;) I'll do an easier
  table here :
     1 creature_shield_deflector,
     2 creature_ro_poison_res,
     3 creature_basic_poison,
     4 player_non_ai,
     5 basic_creature_handler.
  Messages :
    Fire damage
      1 - take message (do nothing) return KEEP_AND_STOP -> that's it.
    Acid damage
      1 - no action, return KEEP_AND_CONTINUE
      2 - no action, return KEEP_AND_CONTINUE
      3 - no action, return KEEP_AND_CONTINUE
      4 - no action, return KEEP_AND_CONTINUE
      5 - Share damage to creature and inventory, send damage message to
          inventory. Return KEEP_AND_STOP.
    TIMEUPDATE
      1 - no action, return KEEP_AND_CONTINUE
      2 - no action, return KEEP_AND_CONTINUE
      3 - Check if it's poison time. If it is send poison damage to self:
             Poison damage
               1 - no action, return KEEP_AND_CONTINUE

2 - Take message (do nothing) and return KEEP_AND_STOP. If poison is diluted enough return UNLINK_AND_CONTINUE else return KEEP_AND_CONTINUE

      4 - no action return KEEP_AND_CONTINUE
      5 - Check whetever it's time to move or not (some way to count time
          compared to time). If it is send ACTION to self.
             ACTION
               1 - no action, return KEEP_AND_CONTINUE
               2 - no action, return KEEP_AND_CONTINUE
               3 - no action, return KEEP_AND_CONTINUE
               4 - Normal player's turn in roguelike games.

return KEEP_AND_STOP.

     KEEP_AND_CONTINUE, UNLINK_AND_STOP etc mean whatever function which
  handles handler lists should keep sending message on in list and if
  handler should be removed from list.


     Yes, I know, this is really heavy way to do things. But if we can count
  on fast 486, it is REALLY flexible. I think that with handlers lists and
  description lists we can mimic any game we want or make just about anything
  we want (well world conquest is a _bit_ hard maybe, but as rg game anything)
     I'm sorry if there's too many errors in my english. I didn't mean to
  be offensive either but I needed to make my point clear (if it can be
  done at all). If someone has implemented this allready, good, sorry to
  say that I were to late.
     I'm sorry if there is something that have been published before and
  I don't want to claim that I've been first to thing this kind of things
  but I haven't seen anything like this before (I haven't read too much).
  Anyway Copyright 2000 Esa Ilari Vuokko. I don't (of course) take any
  responsibility of anything you do or don't do with this text. It can be
  freely copied and published electronically as long as it isn't modified.

Heap Space Conservation - Steve Segreto [ssegreto@titan.com].

Or How to have LOTS of monsters and LOTS of treasure in your Roguelike.

   Here are some tips for conserving precious heap space, so that you

will be able to populate each of your dungeon levels with as many monsters and items as you want! This is a good alternative to having to write a protected mode program, and while it may be a little too slow for some 386s, the algorithms can be tuned as needed. The basic concept is that you will only store in RAM as little as you possibly need to know about all the monsters and items. All the rest of the stuff (specific information about a monster or item) you store on the player's hard drive. The speed versus memory usage trade-off is obvious. You will use a lot less RAM by only storing the indices into the disk files in a linked list in memory, but your game will be slightly slower because you must frequently return to the hard drive and read/write information from it (this is VERY slow compared to reading/writing from RAM).

Anyway, here is some sample code for storing indices for monsters and items using ANSI C and assuming that each of your dungeon levels is 80 cells by 25 cells, for a total of 2000 square cells (this may be smaller or larger than what you want for your roguelike). My ANSI C is pretty rusty (I prefer Pascal) so please be aware that this code does contain syntax errors.

/////////////////////////////////////////////////////////////////////////////////

// BEGIN CODE FRAGMENT /////////////////////////////////////////////////////////////////////////////////

   #define MAX_ROWS 25
   #define MAX_COLS  80
   typedef unsigned int word;   // 16-bit quantity
   typedef unsigned char byte; // 8-bit quantity
   typedef enum NodeTypes = { MONSTER, ITEMS }; // Different sorts of

data files you will have


/////////////////////////////////////////////////////////////////////////////////

   //  A singly linked list of indices.

/////////////////////////////////////////////////////////////////////////////////

   typedef struct index_list_node;
   typedef index_list_node *index_list_node_ptr;
   struct
   {
        word                       index;
        NodeTypes             node_type;
        index_list_node_ptr link;
    } index_list_node; // Size = 7 bytes
    // Related list functions are new_list(), free_list(),

insert_before(), insert_after(), is_empty(), is_full()

    //  advance_list(), reset_list(), read_list_from_disk(),

write_list_to_disk()


/////////////////////////////////////////////////////////////////////////////////

   //  Records to hold monsters and items on diskette.

/////////////////////////////////////////////////////////////////////////////////

   typedef struct
   {
         byte row, col, depth;
         char symbol; // Whatever data you will have to represent a

monster: name, hit points, AC, etc.

    }monster_record;
   typedef struct
   {
         byte row, col, depth;
         char symbol; // Whatever data you will have to represent an

item: weight, damage, etc.

   }item_record;


/////////////////////////////////////////////////////////////////////////////////

   // Global array of index_list_nodes for each square of the dungeon.
   // Initiallly this array will be MAX_ROWS * MAX_COLS * sizeof

(index_list_node) bytes large

   // 80 * 25 * 7 = 14,000 bytes plus 7 bytes for every monster/item

added.

   // Assuming about 150 monsters and 200 items per level, this gives

you 14,000 + (7 * 350)=16,450 bytes

/////////////////////////////////////////////////////////////////////////////////

   index_list_node object_array[MAX_ROWS][MAX_COLS];


/////////////////////////////////////////////////////////////////////////////////

   // Because the above list will only contain INDICES into permanent

disk files, deleting elements from the list

   // such as when an item is picked up or a monster slain, will be

extremely slow (because the entire level file

   // on disk will have to be re-written minus one single record). To

alleviate this, simply keep a second linked

   // index list of all the indices which need to be deleted from the

permanent disk file when the player leaves this

   // dungeon level (these indices have already been deleted from the

object_array linked lists.) Remember to

   // insert indices into this list EVERY time a monster is killed or

an item is picked up (you might want to

   // have a function delete_monster(which_row, which_col, what_index)

which removes the specified node

   // from the object_array[] linked list and also inserts it into the

deleted_list [Do you see why you need to

   // pass in what_index? That's right, so you can go to the

appropriate (what_row, what_col) element in object_array

   // and traverse that linked list until currPtr->index == what_index,

then you can delete this node and insert the

   // what_index into the deleted_list!] ).

/////////////////////////////////////////////////////////////////////////////////

    index_list_node deleted_list;


/////////////////////////////////////////////////////////////////////////////////

   // You will also have two disk files per dungeon level, one for

MONSTERS and one for ITEMS.

   // You must devise a naming scheme for each (LEVEL01.MON and

LEVEL01.ITM for example)

   // LEVEL01.MON is a random-access file of monster_records and

LEVEL01.ITM is a random-

   // access file of item_records, one record per monster on that level

and one record per item on the

   // level. Note that these disk files may be quite large

(sizeof(monster_record) * num_records bytes).

/////////////////////////////////////////////////////////////////////////////////


/////////////////////////////////////////////////////////////////////////////////

   // stock_depth()
   // Call this function at the start of the game or whenever the level

needs to be completely re-stocked:

/////////////////////////////////////////////////////////////////////////////////

   void stock_depth ()
   {
       // 1. Open the appropriate monster level file (LEVEL01.MON for

example)

       // 2. Read each monster_record one at a time into a local

variable, m

       // 3. Based on the m's (row, col, depth) triple, use the

appropriate element of the global object_array[].

       //     For example, if m.row = 10 and m.col = 10 then

object_array[10][10] would have a singly linked list

       //     of index_list_nodes.
       // 4. Insert the index into the linked list (use insert_after()

from the basic list routines above).

       // 5. Do the same thing for the item level file.
       item_record i;
       monster_record m;
       FILE *infile;
       word index = 0;
       // Add all the monsters to our array of linked lists.
       infile = fopen ("LEVEL01.MON") // This line is not syntactically

correct

       while (!eof(infile))
       {
           fread (infile, m); // Wrong also!
           insert_after (object_array[m.row][m.col], index, MONSTER);

// The index is the record number and the linked list is

// at (m.row, m.col)

           index++;  // Move on to the next record
        }
       // If there are not enough monsters on this level, ADD monsters

until there are enough.

       // Add all the items to our array of linked lists.
       index = 0;
       infile = fopen ("LEVEL01.ITM") // This line is not syntactically

correct

       while (!eof(infile))
       {
           fread (infile, m); // Wrong also!
           insert_after (object_array[i.row][i.col], index, ITEM); //

The index is the record number and the linked list is

// at (m.row, m.col)

           index++;  // Move on to the next record
        }
    }


/////////////////////////////////////////////////////////////////////////////////

   // change_depth()
   // Call this function every time the player changes dungeon levels,

including at the end of the game:

/////////////////////////////////////////////////////////////////////////////////

   void change_depth ()
   {
       // 1. Open the current monster level file (LEVEL01.MON for

example)

       // 2. Open a temporary file
       // 3. Read each record one at a time into a local variable, m.
       // 4. If this record number (index) is NOT in the deleted_list

(see above) then write this

       //     record to the temp file (otherwise DON'T write the record

to the temp file).

       // 5. Close and delete the monster level file.
       // 6. Rename the temp file to the same name as the monster level

file.

       // 7. Now the temp file contains all the records except those

which have been deleted (dead monsters).

       // 8. Do the same with the item level file.
       // 9. Call free_list() to destroy the linked index list.
       // 10. Based on the new depth, call stock_depth() with the new

depth number to load/build the new index_list.

    }


/////////////////////////////////////////////////////////////////////////////////

   // square_has_monster()
   // Simply scan the linked list at object_array[which_row][which_col]

and return TRUE as soon as one of

   // the nodes has a node_type of MONSTER.
   // You can devise a similar function for square_has_item().

/////////////////////////////////////////////////////////////////////////////////

   boolean square_has_monster (byte which_row, byte which_col)
   {
       // List at this square is empty (square has neither monster nor

items)

       if (object_array[which_row][which_col] == NULL) return (FALSE);
       index_list_node currPtr = object_array[which_row][which_col];
       while (currPtr != NULL)
       {
            if (currPtr->node_type == MONSTER) return (TRUE);
            // Move down the list
           currPtr = currPtr->link;
       }
       return (FALSE);
   }

/////////////////////////////////////////////////////////////////////////////////

// END CODE FRAGMENT /////////////////////////////////////////////////////////////////////////////////

Handing Out Experience - Rick Carson.txt

Roguelikes are part of a family of games in which the participant builds something over time. Relatives are of course roleplaying games and other storytelling games. Distant relatives are games such as Civilization.

Since players often fail to achieve the primary goal (e.g. retrieve the artifact of lint removal which will cleanse the bellybuttons of everyone in the universe thereby ending world hunger) the secondary rewards (or payoffs (or utility for all you economists out there in RL land)) become very important.

Too much advancement too quickly (kill a kobold - get to 35th level) devalues 

the payoff, whereas too slow an advancement (kill everything in the whole game and maybe get halfway to level 2) means that they never even get to feel rewarded.

Of course experience might be handed out for a range of activities, not just killing things. such a list is certainly not limited to: solving quests; receiving damage; causing damage; casting spells; using skills; getting treasure; idle gossip; donations to the (un)worthy cause of your choice; doing nothing or resting; healing; travelling, mapping, making other players laugh....

So lets consider the case of a fairly generic dungeon bashing roguelike.

The dungeon has sixteen levels, a predetermined number of monsters per 

level, and unique monsters. There are no 'random' monsters. We'll call our theoretical roguelike Jiavlo and code it in Java. (Stop me if I infringe any copyrights ;-)

Experience is only handed out when we generate a monster killing event.

Our first realization is that in this scenario there are a fixed number of monsters, and that this means that there is an upper limit for how much experience the player will ever get.

We need to decide how many levels we want the players to be able to gain.

So what constitutes a good number of levels that could be gained?  Most 

of the audience will have at least a passing familiarity with the D&D game family, in which you start at level one, and progress is technically unlimited,

but which is less meaningful once you get into double digits.  And certainly 

once you get into the low twenties you are in danger of being accussed (unjustly of course) of being a 'power gamer'. So most of the target audience is going to be happy with a top level of around 15 to 30. Twenty is a nice round number in that range which should give a decent payoff in terms of achievement.

Note that the point is not to mirror any particular published system for which one runs the risk of being sued and losing the shirt off ones back,

but rather to recognise and take advantage of this subconscious assumption 

that most people are going to make, ie that getting to level 20 is a superhuman feat and that getting to level 12 (say) is pretty spectacular.

The other thing to do is to make the experience scale exponential (this means that each level requires more experience to get to than the level before it). Mathematically there is no reason for this, but people will expect it, and because of their expectation they will not feel as much of a payoff if they get the same amount of experience for killing a dragon as a kobold. Because of this, I recommend using a spreadsheet to play around with the numbers till you some that you like. Hopefully you won't need a spreadsheet to follow this article though!

It is useful to consider the linear case here though, because it forms a basis from which you can build the exponential formula, and if you do so you will be much more likely to get a well balanced game. But before we look at the linear model, lets examine the constant model. This is even worse in terms of payoffs, but much simpler again to balance properly.

For every monster you kill you get 1 xp. There are 50 monsters per level. (Times 16 levels is 800 xp) We can now say something like, for every 40 xp you have, you go up a level.

(800 divided by 20).

Note that you start at level 1, so you would get to level 20 after killing 760 creatures, ie you max out your levels shortly before you kill the UberLintDaemon which drops the game ending artifact, so that you have time to enjoy maxing out on levels before the game ends.

If the monsters on deeper levels get harder to kill, then logically players would try to clear a level before progressing to the next in order to get the 'easiest' xp first (minimise risk, maximise payoff).

Note that we are requiring them to kill 40 monsters in order to advance to second level. This is far too harsh, and for many first time players this will take too long for them to get hooked. Ten is a much more reasonable number. And we want them to be able to get to say level 5 or 6 fairly quickly before it becomes hard slog. So lets break it down, lets say that by the end of dungeon level 15 they are level 20, and they gain a level per level they clear back till say the player is level 10. That means that we then need to plot the xp from 0 (at level 1) to 250 (which they have after clearing 5 levels at level 10). We could go back and rewrite our assumptions, and have them start at level 0, and then have them gain a level every 25 points.

That makes it nice and easy for us in terms of maths.

The effect this has on the game though, is to make the clearing of each level a hard thing at first, and then get easier halfway through each level.

This is actually not too bad in terms of payoffs, because the player begins 

to notice that the monsters are easier to clear after going up the level.

So lets modify how players go up levels. Here is the table (level, xp required): 1 0 2 5 3 15 4 30 5 50 6 75 7 105 8 140 9 180 10 225 (and 50 per level after that).

So at the end of clearing level 1 (50xp), they are fifth level, clearing level 2 puts them at sixth level verging on seventh, etc. This uses a clear progression, in order to go up a level, you need as much extra xp as the last time you went up a level, plus five, until you get to needing 50 a level where it tops out).

Despite the plain maths behind this, I would be tempted to fiddle with it even more, because fifth level after clearing only one level of monsters seems 'too much'.

But lets take this 'constant' model, and see what happens when we make it 'linear'. We assume that all the monsters on level x of the dungeon are 'level x monsters' and we give x points for killing them. The maximum amount of experience in the game is now: 6800. The maths for working this out is starting to get more complicated, but the formula is: (((n squared) plus n) divided by 2) times the number of monsters per level). Where n is the number of levels, and the number of monsters per level is still 50.

Obviously we cannot now just multiply the cost of going up levels by 8.5

(6800 divided by 800) because that would mean we would need 85 xp to get 

to second level, which means clearing all of level 1, and a third of level 2. All of a sudden things look a lot harder! One way of solving this, is to try to match monsters killed to levels gained, assuming that all the low level monsters are killed first. This is easier than it sounds for levels 10 through 20, because we already know that we want the player to advance at the mid point of the level. And this is not too hard to work out (especially with a spreadsheet).

&gtFrom 10th level and up (level, xp required): 10 900 11 1225 12 1600 13 2025 14 2500 15 3025 16 3600 17 4225 18 4900 19 5625 20 6400

(The formula for this is: (((n squared) plus n) / divided by 2) plus (25 times (n plus 1)) where n = level required minus 5. The minus five is because we have five more experience levels than the last level of the dungeon that we cleared)

Now if we want to keep the same progression, that is, getting to level five after clearing the first level of the dungeon (yuck!) we leave that bit of the table alone, and figure out a progression from 5th level to 10th level. Or, we can choose another formula to govern low level advancement,

such as: (n times n) plus (14 times n) minus 1, where n is the current 

level which yields: 1 0 2 14 3 45 4 95 5 166 6 260 7 379 8 525 9 700

Which is a much nicer fit to the lower levels of the dungeon and how fast they should advance. Unfortunately I've just noticed that it bears a striking similiarity to the xp curve in Crawl. What a shame :-)

But of course noone wants to get 12 xp for killing a Dragon, so we need to think about how much xp we want to hand out for killing a monster of each level. I suggested at the start of the article that making it exponential is probably more in fitting with the players expectations. Lets say that on level one you get 1 xp per monster killed, and on level 16 you get 10, 000 xp per monster (so level 16 has half a million xp of monsters on it - suitably impressive amount :-)

A 'basic' formula for this would be: (n to the power of ((n minus 1) divided by (15 divided by 4)))

Which gives these results: 1 1 2 1.847849797 3 3.414548874 4 6.309573445 5 11.65914401 6 21.5443469 7 39.81071706 8 73.56422545 9 135.9356391 10 251.1886432 11 464.1588834 12 857.6958986 13 1584.893192 14 2928.644565 15 5411.695265 16 10000


You then need to figure out how many xp are needed for each level (use a spreadsheet!), and split the difference between levels, and then somehow figure out a formula for advancing. I came up with (2 to the power of n) plus (n cubed) minus (25 times n). Which is ok, but given more time we could come up with something better. In particular, I'd reduce the base number of the power to something in the range of 1.97 to 1.99 so as to get closer to the 'ideal' value for that last level (you would get to 20th level with three creatures remaining given the formula above).

Some things to think about: do the rewards actually match the difficulty.

For instance if a level 12 monster isn't really much harder to kill than 

a level 8 monster, why hand out ten times as much xp for killing it?

In fact, the difficulty of balancing even a limited scenario as outlined above means that you are far better off having a 'complicated' function for handing out experience, than aiming for some mathematically pure function.

An example: Offense * Defense is a very good measure of combat capability (ignoring the effects of swarming, assume that the character can always retreat such that they only fight one monster at a time). Offense is the average damage they do (ie chance to hit times average damage times number of attacks) and defense is how much damage they can take on average (ie chance to avoid damage times hit points).

Example 1: a rat which does 1-2 points of damage, has a 1 in 3 chance of hitting, has no chance of dodging and 2 hitpoints would be worth 1 xp. ((1.5 times (1 divided by 3)) times 2) Example 2: a balrog which does 10-20 points of damage, has a 100% chance of hitting, gets three attacks a round, avoids (dodges/deflects/counterspells/whatever) 50% of attacks and has 100 hit points would be worth (15 times 3) times ((100 divided by 50) times 100) or 9000 xp. Actually, that might not be enough :-)

You can increase the spread by raising the amount of xp from the above formula by some power greater than 1, try 1.1 or 1.2. 9,000 to the power of 1.2 is 55,602 which should keep the players happy. Whereas 1 to the power of anything is still 1.

Other factors to consider are the dungeon level the character has gotten to, what resistances the monster has, whether the monster has ranged attacks (these are really nasty in roguelikes, you may wish to reward the players accordingly) and whether it drops items that help the player (which are a reward in themselves in that they improve the character) when it dies.

Of course killing an Elder Dragon and having the experience it gives you 

reduced by 10% because it dropped a +1 dagger, would suck.

In terms of experience to go up levels, something to avoid is requiring as mush xp to go from 1 to 2 as from 2 to 3. This can happen when you use an exponential curve that doubles for a while. Just as a wild totally random example, take a fighter that needs 2000xp to get to level 2, and 4000xp to get to level 3, and 8000 xp to get to level 4 etc. Now think about this,

in order to get to level 2, the fighter had to gain only 2000xp, but this 

is also the amount they need to gain in order to get to the next level, but now they have an extra levels worth of hitpoints, better chances to hit etc, which means that it is easier to get from 2 to 3 than it was to go from 1 to 2. And the game shouldn't get easier as they go up levels.

cheers,

Rick

Allocating Experience and Character Advancement - Matthias E. Giwer [mgiwer@ij.net].

First, a brief bit about my experience in this area:

I've been an active player of "pencil and paper" role playing games since 1987, encompassing more than thirty (truthfully, I've forgotten all the games I've played) different systems, as well as a number of "tactical" games such as BattleTech, Warhammer, etc.

I've spent (wasted :) hundreds of hours in roguelikes such as Larn, Moria, Nethack, (A-Z)ngband, ADOM, etc. I attempted to develop my own on several occasions, the most recent attempt being Saga, a Perl roguelike.

I do not claim to be the worlds most serious gamer, nor the worlds best software engineer, but I do feel I have enough experience on this subject to be able to offer some useful information.


Now, on to the meat:

How you implement character advancement will depend heavily on your choice of game mechanics. For example, is it a level based system where a 10th level being will be ten times more capable than a 1st level being? a hundred? Or is it entirely skills based, where facility with a particular weapon or ability is to be developed independently of any other capability?

The majority of current roguelikes seems to be dominated by levels and level/skills hybrids. This is fine, though while I personally advocate an entirely skills-based approach, it is truly a matter of taste.

The base idea is that individual actions performed by the characters will return (on success, failure or both) some measure of "experience", either generic or towards the skill governed by the action. In order to develop a mental yardstick, I usually try to get a firm idea of what a brand new, baseline character is capable of. In most roguelikes today, that would be a human fighter.

Next, try to imagine what this character would be capable of, if his abilities were taken to their absolute limits, without regard to items and equipment that may boost her capabilities. This should give you a continuum of capability that will let you look at, and gauge, difficulty of things like quests, unique creatures, and so on.

It also usually helps me to think of creatures that this character may face as based on this brand new beginning fighter. In other words, a Grey Kobold is about half the power of a beginning fighter, but an Elite Orc Guardsman is six times more powerful than the baseline.

Yes, there is a LOT of tweaking and experimentation involved to get it right. Also, many abilities can make combat orders of magnitude simpler (teleportation, ranged attacks, etc) so this has to be taken into account when trying to judge anything other than a raw fighter type. Again, there is no perfect measure, experimentation is the key to balance here.

If you wish to model the real world, most performance type skills (dancing, martial arts, typing, etc) are developed in terms of plateaus that may take months or years to break through, usually on a sliding scale of time. The first plateau might take only a few weeks to reach, but the next may take a few months, etc. Roguelikes (and many traditional RPG's) model this with each "level" requiring progressively larger amounts of experience points to reach.

However, because most games give out larger amounts of experience for more difficult and more powerful creatures, the result is that most characters end up on a hyper-accelerated track, and can reach quite amazing levels of skill in relatively short amounts of game time. Now, as a fantasy game goes, that is perfectly acceptable, but it does make it more difficult to judge when a player is ready to face the "Deep dwelling Thing from Beyond". Then again, people don't play roguelikes because they are easy. ;)

I would have no issue with this, if it was a one-time event. For example, the first time you wipe out a Grey Ooze, get the full experience value for it. For each one after, only bestow a standard action credit that doesn't change, for all actions of that type. New things, and new challenges help you grow a lot, practice helps you a little (which is why you have to keep at it).

An extension of this idea is experience for, say, casting spells. You get full experience value the first time you cast a new spell, but only a small "cast spell" value each time thereafter. The idea works for any ability that is action based, rather than intrinsic or "always-on".

For added realism (ha :), it may be worth investigating a "decay" function for skills and abilities that go unused for long periods of time, with the time before decay increasing the higher the level of skill.


Like the design of any game system, this article is two parts experience and one part prejudice. I hope that, at least, mentioning these issues will prompt roguelike designers to devote more thought game mechanics issues, and to question some accepted standards.

I do apologize if you were looking for code examples, but the entire issue of character development is so tightly bound to the specific system and other game mechanics to make even pseudo code examples just about worthless.

Feel free to write me with questions, comments, etc. Please keep the flames constructive, thanks! :)

-Matthias E. Giwer mgiwer@ij.net

Creating A Player - Brian Bucklew [bbucklew@inteletek.com].

Section 1 : The Definition

[Note : I will be using C++ Class definitions]

       So, let's pretend we have a nice endless dungeon filled with

vile demonic beasts, heaps of gold, and piles of flaming swords of instant monster to magic item transmutation. Mabey we even have an engine to make the ubiquitous '@' saunter around in this great dungeon.

       Now, what we really need is some way to make that little '@' into

a daring adventurer, with Herculean strength and an intelligence to rival Hawking (or perhaps the lack of the same...).

       First off, we need to have some basic statistics and attributes.

perhaps we need a Name, a Race, and a Class... We also need some numbers to represent the Hero's strength, and intelligence as well as his other important attributes.

       Now, in any RL, a player's Statistics are going to go up and

down during the course of play, whether it's from going up a level or getting hit by a Lecherous Walking Carrot Of Charisma Sapping. So we should really represent each score by two numbers, a current score and a base score:

class Stat {

       public:
       int     Base;
       int     Current;

}

       All of our calculations done in the game should be based off of the

current score, but the base score will allow us to revert to the Player's origional score or hit-point total after a magical-effect or damage wares off.

       Having the basic stat class, let's go ahead and derive a class to

contain a basic RL character:

class Player {

       public:
       char    Name[256];      // A String to hold the player's name
       int     Race;           // let's say, 0=Human 1=Elf 2=Frog...
       int     Class;          // hmm... 0=Fighter 1=Mage 2=Tourist...    
       Stat    Str;            // How Strong
       Stat    Int;            // How Smart
       Stat    Dex;            // How Fast
       Stat    Hlt;            // How Healthy
       Stat    Cha;            // How Well-Liked
       Stat    HP;             // How much damage you can take
       Stat    AR;             // Your armor-rating
       Stat    SP;             // Spell-points... How many kobold you can toast

};

       That should be a good starting point for developing a unique character

representation for your RL...

Section 2 : The Creation

       The actual creation of the character is accomplished by somehow

assigning a score to each of the stats... There are two main ways to accomplish this; Random Rolling and Point Distribution.

       Random Rolling would display all of the statistics, and allow

the player to hit a key to "Roll" (or randomly assign numbers, say between 1 and 100) to each main statistic. The player would then continue rolling until they have a set of numbers that is agreeable to them.

       Point Distribution would give the player's a number of points, say

100, and allow them to distribute those points to their attributes in any way they wish; Thus allowing the player to design a character that they want to play.

Usually, the player's chosen race affects the statistics in some way. For instance if our player picked:

A Human - No change; Humans should be pretty standard fare. An Elf - +2 Dex, -2 Str; Elfs are quick, but weaker than humans A Frog - +2 Int, -2 Cha; Frogs are obviously intelligent, but ugly...

Classes are generally restricted by attributes...

Fighters - No basics; Everyone can pick up a stick and hit something... Mage - At least a 12 Int; A Mage has to read, at least... Tourist - At least a 12 Dex; You have to move fast, only 36 days left of

                             vacation...

The player might be given some basic equipment based on his class, and that's about it... That '@' has everything it needs to thump some Kobolds!

The Author: Brian Bucklew, 18 bbucklew@inteletek.com Current RL Project : Dungeon or "This game needs a new name"... :) Projected Beta Release : Early 98

Pretty Pictures - Darren Hebden [rogue@skoardy.demon.co.uk].txt

Some thoughts on Graphics in Roguelike Games. by Darren Hebden [rogue@skoardy.demon.co.uk]


Traditionally, if you suggested graphics to a roguelike- stalwart, you would have found yourself out on your ear for speaking such heresy. Many would argue that being text-based is a defining feature of the roguelike genre. Others want more and look upon graphics as a natural progression.

Many people (purveyors of quality knee-jerk reactions and 'The end is Nigh' flavoured rantings) warn of the old-style being phased out or graphic-based roguelikes lacking the same depth of text-based. Personally, I like to believe that both can exist quite happily and that the addition of graphics in roguelike games is a bonus, not a replacement. It is a new feature and it's use within the roguelikes is still being explored.

This document talks a little about the uses of graphics in roguelikes, the different styles and some graphic techniques. Not being an expert on all systems (or the one I own, come to that), I'll try to make my article non-machine-specific. If you know a little about the packages you own, you should be able understand my texts.

                             ----//----      
  1. Why Graphics.

PROS. 1) It's a "draw". The visual style attracts the eye to games that many of the uneducated heathens (sorry, I mean -- people who haven't encountered roguelike games before) would ignore and pass over because of their archaic text-editor style appearance.

2) It's expected. Unfortunately, the expectations of todays game player have been adversely influenced by big budget, multi-cd, flashy productions. I don't exactly find this a pleasant situation but ignoring it or "taking a stand" won't help you or your game.

3) So what's a 'Furgikan-bug'? Although a great deal of people are up on Tolkien lore and AD&D monster theory 101, but an unknown creature shown as a 'F' character on-screen may leave several of your players scratching their heads. A well-drawn image, however, is instantly recognisable and gives a better impression of the creatures appearance.

CONS. 1) Graphics kill the imagination. The original roguelike style counts on the players imagination to transform a simple red 'D' into the terrifying bulk of Fire Dragon flesh. No matter how detailed your graphics are, they may not be able to live up to the visions swirling around an over-active imagination.

2) Bigger programs. The storage of large amounts of graphics data all adds to the overall archive size for your game. It's very difficult to produce a competent roguelike game with a small selection of graphics without compromising the colour-depth or tile size. Whichever way you even things out, graphics mean an increased game footprint.

3) Bad graphics drag a game down. Good graphics can really add to the atmosphere of a game but in the same way, a bad set of graphics can seriously hamper a game which is technically competent and full of depth. Will you be able to supply good graphics for your game or would the game be better served staying with a text only display? Don't treat graphics like the current band-wagon to climb aboard.

                             ----//----      
  1. Graphic Styles.

Although all through this document I speak of older roguelikes being text based, simple texts characters _are_ a form of graphics. A 'g' from a gremlin may be missing a few limbs and that mad glint in his eye but people soon become lost within the game world and accept that a horde of crazed 'g's bearing down on a weakened adventurer is not a good sign.

There are several different styles that you can use when representing your dungeon on-screen.

1) Flat Tiles. The most typical use of graphics in roguelikes is a one-for-one replacement of the ascii-characters themselves. A wall character is replaced with a brick pattern, a water character by a lump of blue.

A little work is needed re-assigning what is displayed or how they are arranged on screen but often, this method restricts the artist in getting the best from the tile size and colour depth available. Due to the forced nature of the way the tiles may be arranged, the artist could be unable to introduce some details that would improve the overall look.

This style doesn't _have_ to suffer from badly arranged tiles if the artist uses the way the tiles are placed to his benefit and the programmer is willing to make slight alterations to get the best out of the tiles provided. The perfect solution is to be programmer and artist :-)

Badly drawn example... XXXXX

                      XXXXX
                      XXXXX
                      XXXXX 
                      XXXXX


2) Pseudo-3D Flat Tiles. The next style is very similar to the first but instead of flat patterns for tiles, the impression of depth is given by drawing front edges on lower half of the tile for the walls. It's a cheap trick but is quite effective. The illusion is only spoiled by the fact that the player cannot walk behind the tiles as they occupy the same space as a normal wall tile. A little 'walk-behind' effect can be gained if the tiles are overlapped. It okay but you also begin to lose a little of the floorspace of the tile above.

All in all, a popular choice and for tiles that need a little more work when it comes to designing them to fit together. It means more graphics but generally, it's worth it.

Badly drawn example... XXXXX

                      XXXXX
                      XXXXX
                      ||||| 
                      |||||


3) Isometric Tiles. These tiles create the impression that the level is viewed from an angle of 45 degrees to the left/right. As they go, this style gives a very good illusion of depth (apart from the total lack of perspective, ofcourse).

A problem with this style is that these tiles really need to overlap. I've tried some experiments with not overlapping them and they looked very bizarre. The overlap causes a bit of lost floorspace of the tiles drawn above the walls but this can be reduced by having shorter walls.

As with the overlapping from the previous style, don't go over the top with the shortened walls though or you'll have levels looking like a stunted hedge-maze.

A solution would be to make any wall that overlaps a floorspace semi-transparent. This effect can be achieved by either some rather clever palette mixing tricks on the programmers side of things (which is rather a bit of overkill for a roguelike) or by a chess-board style hashing of the pixels over the overlapped area -- one pixel of the wall, one of the floor, etc. It is a bit of a cop-out but quite effective.

Badly drawn example... X

                       XXX  
                      XXXXX
                      |XXX|
                      ||X||
                       |||
                        |


4) Tunnel Vision (or That-Style-From-Dungeon-Master). The player is presented with a full perspective view of a corridor/tunnel of the dungeon. Items in the distance are smaller and tunnels lead off from the original at right angles. Movement is generally made in steps and switching the direction viewed usually snaps round ninety degrees.

This style allows for more detail on the wall/floor tiles as the tiles closest to the player are quite large on screen. I've yet to see this used in a roguelike game although it has been suggested quite a few times. Other games have used this style and in some ways, you could argue that Dungeon Master, Eye of the Beholder, etc. are in essence roguelike.

One reason for it not being implemented is that the player can only see in one direction at a time. Usual roguelikes have forces coming at the player from all sides. This may put several players off due to the confusing nature of this 'realistic' feature. Other windows containing the left/right/back views may help this situation.

Badly drawn example... .........

                        |.......|
                        ||.....||
                        |||   |||
                        |||   |||
                        ||.....||
                        |.......|
                        .........


5) 3D Texture Mapped. The last style I'm going to cover is texture mapping. So far (without making a massive leap with the definition of the genre), no roguelike game has ever used texture mapped 3D graphics. I'm sure if you visualise a game like Quake, or whatever the current splurge of the month is, you'll realise what I'm getting at here.

One problem with texture mapping is that to get a good definition on the walls, floors, etc. the textures need to be of a fair size otherwise, close inspection of a wall reveals a horrible level of pixelisation. 3D accelerator cards (and a certain console) can provide techniques to smooth the textures but now we're getting into silly-land. Roguelike game with texture-mapped polygons needing a 3D accelerator card? Hmm. One day maybe...? ;-)

Anyway, it's really up to you as to what looks right and at what point it stops looking like a wall pattern and more like a bunch of large square pixels. Working towards high quality textures means bigger textures and even more space taken up by the graphics.

Obviously, programming such a view for just a roguelike game is no small fear either when compared to the original text-mode (or even the other styles mentioned in this document). How you present your 3D environment is up to you. Possible options are the fully submersing environments of Quake/Mario64. Or you could go for an isometric display such as Bullfrog's Dungeon Keeper.

Badly drawn example... Yeah, right.

                             ----//----      
  1. Issues

Some problems you might want to take into consideration when working on your graphics...

  1. Colour Depth.

Colour depth is essentially how many colours you're allowed to create your wondrous and vivid roguelike graphics. Basically there is two camps. The programmers want the less number of colours possible and the artists generally want the most. The exception to the programmers is when the programmer in question doesn't care how much space the game takes or already has the code to handle higher colour depths. The exception with artist is when you find an artist who knows they can get the same effect with less colours.

A set of tiles that only use a palette of 16 colours but have been stored as in a graphic format that utilises 65,000 colours uses up more memory than exactly the same tile set in a format that only allows for only 16 colours needed. If your tiles only take up a small range of colours, that 16bit (or 24bit) mode could be a waste of time. And if you can get beautiful tiles with 16 colours, programmers will love you. :)

Again, it's a balance. Try some test tiles first. If you've got a list of the tiles needed, experiment with some fairly mis-matched sets and see how they stand up to the lower colour depths. If you're not happy with how they are working out, move up to the next depth. Remember though, an artist may have to compromise the graphics he/she creates to work within limits. Do the best you can with the tools available and the boundaries you are set.


  1. Tile Size & Screen Size.

In a perfect world, all players would have limitless hard-drive space, screen-resolutions that stretch into the ten-thousands and download the several hundred meg your game takes up in a second. Problem is, it doesn't work like that. Once again, file-size is affected by the choice you make over tile sizes. Deciding to expand your tiles from 16x16 pixels to 32x32 pixels will increase the file space each tiles takes up by 4. If the programming has decided they want tiles for each of the five-hundred different flavours of gnome they've implemented, we're talking of a big increase.

Can you do everything you want with 16x16? Can you get all the detail you need and is the detail really necessary? In some cases, a good artist can 'imply' detail in a very small space. Do you really need to see the texture on the knight's chain mail or would a light grey mesh give the same impression?

Another point to take into consideration is the screen size. A small screen size and a large tile size could cause trouble and restrict the amount of 'dungeon' you can see at one time. Small tiles on a huge screen could mean the player has trouble making out the tiles and misses out important details. Again - balance.

See what the programmer wants. How many tiles do they want onscreen. Make sure that they understand how big that would allow you to make the tiles. A sample screen knocked up in five minutes is usually a good visual aid for both you and the programmer. Does it look right?

If there is a lot of other information needed onscreen, the amount of screen left over to the level window may be cut down. Make sure you take all of that into consideration.


  1. Resizing or Reducing Colours.

Okay. You've created all your tiles. Five hundred gnomes of all shapes, sizes and colours. All in 64x64 pixel, 24bit colour tiles. The programmer gets in touch and tells you that they can't use them. They need 16x16, 16 colour. Bugger...

Well, you've got three choices - call it a day, move to Tibet and take up the serene life of a monk or restart from scratch or start reducing tile size and colour depth. In the extreme case stated above, the reduction is going to pretty much murder your graphics and perhaps restarting is the only way to go. In other situations, when you only need to reduce the colour depth or just resize, you might just get away with it.

One thing to remember when reducing colours, it always helps to choose a good utility/art package. The best way to test this is to use it with other graphics first. Choose a picture (maybe a scan of some sort) with a fair spread of colours and detail. Reduce the colours. If you're coming down from 24bit to 16bit, chances are you won't notice the reduction but down to anything less and the package has to create it's own palette of used colours.

Most often, it's a best-fit and most-used kind of technique. Some packages offer different methods for reduction and some use a dithering algorithms to fool the eye. Sometimes it even works. Remember though, you may be reducing the colours of some fairly small tiles and when they begin to repeat in the game, the pattern these dithered tiles create may become very noticeable. The key to reduction is getting a good palette. Again, this depends on the tool you use.

One thing I have noticed is once your package has reduce the colours, it tends to organise the colours in the palette in a manner which I find a little 'random'. Personally, I like my colours divided up into nice hue sets but after reduction, most packages generate light-to-dark range or that bizarre mac-style organisation that I dislike so much :) It's a little gripe but remember that if you're drawing more tiles after the reduction, finding the exact colour you're after may become a pain in the ass.

Resizing is another matter. Most resizing does cause your tiles to lose all their detail and essentially makes touching-up afterwards a must. Some packages anti-alias or blur the tiles to cover the loss of detail. This is all very nice within the tile but you've got tiles with areas that are supposed to be transparent (for overlaying onto floor tiles, etc.) then the program tends to blur the edges of the shapes and destroy the clean edge you need. A good package will allow you turn this off.


  1. 1001 Monsters.

Not just monsters but potions, swords, shields, helmets, scrolls, spell-books, traps, walls, floor, stairs, food... well, you get the idea. When you think of all the tiles you may be called on to draw, you start begin longing for those simple ascii-characters. When drawing your tiles, it's always a good start to have some sort of idea of how many monsters, etc. will be required. If you need several different types of mold (green mold, yellow mold, red mold, blah-blah- blah), you would probably be able to get away with the same graphic but with changes to the colour. It saves time and allows you concentrate on making the basic mold design stand out from your other creatures.

It is also useful to find a theme within creatures. If you need several ranks of orc, maybe changing their size and armour might help. If you have a set of monsters that have a specific trait (nether beasts, lava beasts), it'll help player recognition if you style those creatures similarly.


  1. 3D packages.

All styles can enjoy the benefits that using a 3D package can give. For the 2D bitmap tiles, building your objects and monsters with a 3D package means that size or colour depth isn't really an issue any more. The object you create can be rendered from any angle, at any size and with varying amounts of colour (if the package you're using is any good in the first place). And because the output is from the 3D package itself, outputting a smaller render will generally give a better result than trying to reduce the size of the tile with another graphics utility.

Ofcourse, the amount of work required initially is massively increased as you need to model all the objects you wish to show. The benefit comes from the flexibility afterwards. For the user wishing to create modeled creatures for a Quake-style roguelike, a 3D package is essential. I'm sure there are people out there who can quite happily create some very good models with a pencil and graph paper but with so many cheap and shareware modelers on the market, save yourself the bother.


  1. Size matters?

Okay, first you're asked to draw a mouse and then you're asked for a poison breathing dragon. Do your work in scale? Draw a mouse that takes up a tile then draw a dragon that fills half the screen? 'Course not. If you're drawing a dragon that takes up a tile but a mouse in scale would be hard pushed to fill a pixel. It takes some getting used to be it's usually a good idea to throw scale out of the window. Just pretend that mouse is really, really, really close. :)


  1. Conclusion.

As you've probably gathered by now, how you go about creating your graphics depends largely on the games needs. If you have been set limits either by the programmer or the hardware you're aiming for, always experiment to get the best you can from the tools you have available. If you're given an open-ended project with no definite requirements, planning will save you a lot of work in the long run. Decide which colour depth and tile size is best for the game and stick with it.

Whatever style you choose, its all a question of whether your roguelike needs these graphics. Do they honestly add something to the game? Ignore those who automatically scream 'no!' or have built up some kind of barrier to considering the option. Graphics are just another feature to add to your game. Like everything you add, whether it works or not is up to you.

Object Representation - Sean Middleditch [sean.middleditch@iname.com].

Representing objects in a roguelike. A difficult thing to do, in fact. When I started War of the Runes, I wasn't sure exactly how to go about doing it. Would everything be hard coded into the game? Would objects be malleable? What kinds of different objects need to be represented?

Well, for starts, I knew I didn't want anything to be hard coded. One thing about War of the Runes is that EVERYTHING would need to be able to be changed. So objects need to be stored in external files, with all descriptions of the object need to be stored; from description, to behavior, to properties. And this also means all objects needed to be malleable. the state of an object can change at any time through any number of ways. And what kinds of objects were there? There's weapons and armor, food and drink, doors, chests, bags, rings, books, anything that can exist in a real world.

So what is the best way to represent objects in a roguelike? Well, first we need to start with an object class:

class Object { public:

What kind of data do we need for objects? For starts, we need to know what the object is. We can do this with a name and a description;

char *Name, *Desc;

That's fairly self-explanatory. Information about where the object is would be useful, too:

int X, Y;

There are lots of different objects in the roguelike universe. Armor, weapons, books, rings, potions, lawn-chairs, etc. Each object can only be of one type.

int ObjType;

This field will be set to a value we define with #define's.

#OBJ_WEAPON 1 #OBJ_ARMOR 2 #OBJ_POTION 3

You get the idea. Every different type of object in the game would need a separate number. This will help in a LOT of areas. For example, characters can only equip items of type 2 (armor).

Since I'm sure some of you may be a bit confused (I know I'm not the best of teachers) why don't we start defining an example item before we continue? A longsword.

We'd set the Name field to point to "long sword". Simple enough. The Desc field would be "a long, sharp, metal stick". Well, you may want to use something other than that.

The X,Y coordinates can be set to whatever... let's say 2,10.

What about the object type? 1 (OBJ_WEAPON) of course!

Now what? What does a longsword do? The game knows it's a weapon, but the game needs more information than that.

Let's make a class to define what an object does.

class ObjArg { public:

Each ObjArg will define one aspect of an object. We'll need a way to store what each ObjArg defines.

int ArgType;

We'll give this meaning through the use of more defines.

#define OARG_ATTACK 1 #define OARG_DEFEND 2 #define OARG_HEAL 3

Now we know what an ObjArg defines. Now we need to store the actual definition. We'll do this by storing an array of 5 integers.

int Args[5];

And that's all there is to the ObjArg class. The meaning of the Args array's field will depend on the value of ArgType.

Now, we want every object to be pretty versatile, so we'll make it so every object has 5 of the ObjArg classes.

} Defines[5];

And that's all we need for a basic Object class

};

So how does the ObjArg class work? Well, for our longsword, we'll need only one. It's ArgType will be set OARG_WEAPON. Now what about the Args array? Let's say for ArgType 1 (weapon), the first field of Args is the number of damage dice, the second field is sides per die, the third field is damage modifier, and the fourth field is the modifier to the character's skill. The fifth field will be unused.

So, if we wanted long sword's to cause 1 to 8 damage, with no additional modifiers to damage or attack skill. So...

Args[0] = 1 Args[1] = 8 Args[2] = 0 Args[3] = 0

Now if the character equips the long sword and attacks, we'll first check to see what kind of object is in the character's hand. We see the ObjType field is set to 1 (weapon). OK, so he/she can attack with that object. Now we'll look through the Defines array until we find an ObjArg entry whose ArgType is set to 1 (attack). The game see that Defines[0].ArgType = 2, so we'll use that ObjArg to look up the weapon's statistics. The game checks Defines[0].Args[3] = 0, so there's no skill modifier. The game does whatever combat system and determines the character hit. It check the damage (Args fields 0, 1, 2) and sees that the long sword does 1d8+0 damage. The game rolls the damage, hurts the target, etc.

Well, that's all you need for simple objects. Although, you could do a LOT more, using the sample code I've given here as a base. For example, some ObjArgs are in effect whenever an item is equipped (the long sword's attack field, or armor's defend field). Some objects, like potions, don't effect unless a character uses the object (or in the case of objects with ObjType = 3, the character quaffs the potion). Only then will their ObjArg fields (like healing, in the case of a potion of healing) take effect. So you might want to store how many times an object can be used, and how many times it has been used.

The complete code for the Object class is:

  1. define OBJ_WEAPON 1
  2. define OBJ_ARMOR 2
  3. define OBJ_POTION 3
  1. define OARG_ATTACK 1
  2. define OARG_DEFEND 2
  3. define OARG_HEAL 3

class Object { public:

char *Name, *Desc;

int X, Y;

int ObjType;

class ObjArg { public:

int ArgType;

int Args[5]; } Defines[5]; };

If you have any questions, feel free to e-mail me at sean.middleditch@iname.com

The End

Inventory Abstraction - Kenneth Power [uncle_wiggly@bigfoot.com].

An Inventory tracks Items in Storage. Items are anything you deem trackable: gloves, clubs, food, coins, flowers. Storage is anything defined as able to hold items: chests, sacks, hands, buildings.

A basic Inventory does not care about extraneous fluff, such as container limitations. Keeping inventory implementation basic enables code reuse as discussed later.

Throughout this paper, psuedo code is used to model examples. The examples use the following sample item definition:

  define Item:
     variable Name


Tracking items There are two basic ways to track items. First is with a simple list, rather like writing a list of groceries. Lists usually are FIFO (First in First out) or LIFO (Last in First out). Other ways may exist. Indeed in some languages, there are very exotic forms of lists. A trouble with lists is the retrieval of information from the middle of the list. The list is examined from either end until the information is located.

Our example simple list:

  list define Inventory

Second is through a more complex scheme wherein more than the item is tracked. Also tracked is information about the item. This is so items may be grouped. Grouping has its pro's and con's like anything else. Use of grouping, for example, allows easier display of items (in my own opinion of course). In a way, grouping is a more esoteric list form, using such things as dictionaries, maps and other terms.

In this example, the items are tracked by their name. Example:

  list define Inventory:
     list define itemNames
     keyedList define qtys


Add items What use is an Inventory if you are not able to add items to it? Thus, the first function we define is the add() function.

In basic form, add() only wants to place the passed item to the list:

  List define Inventory:
     define add(item):
        insert item

With the complex form, add() is more detailed. Questions are raised: is the item already in inventory - this might affect or tracking feature-? Did I/you make an adjustment to the tracking feature if necessary? Note how this is done in the example:

list define Inventory:

  list define itemNames
     keyedList define qtys
     define add(item):
        if item is in me.itemNames       #do I exist?
           then me.qtys.key(item.name) = +1
        if item is not in me.itemNames   #I don't exist
           then me.qtys.key.add(item.name)
           me.itemNames.add(item.name)
           me.qtys.key(item.name) = +1 


Remove items The remove() function is really the opposite of add(). Find the item passed and remove it from the list. Let's examin it with our two pseudo codes.

Of course, this example doesn't take into consideration language dependent behavior (C - did you fix your pointers?). Thus the abstraction is the same for add():

  List define Inventory:
     define remove(item):
        delete item

Here, as in the complex add() function, more work needs accomplished. We not only find the item, but we make certain our tracking feature is adjusted accordingly.

  list define Inventory:
     list define itemNames
     keyedList define qtys
     define remove(item):
     if item is in me.itemNames
        then me.qtys.key(item.name) = -1
        if me.qtys.key(item.name) == 0
           then me.qtys.delete(item.name)
     if item is not in me.itemNames
        then error

At the completion, our example would show the proper quantity for the item. Plus, if we have no more item in inventory, no quantity or listing will exist.


Display items (opt.) This is listed as optional, because you may not code Inventory display as part of your Inventory. It may be in your UI code. However, during testing, having a simple Inventory Display feature is very useful.

Like always, the simplest way is the list method. Merely iterate the list, printing each item:

  List define Inventory:
     define display():
        For each item in Inventory:
           print item.name

Because of our tracking feature, we need print item and qty. Otherwise uncertainty will exist. The only added feature of the complex over simple is the header printed "Item Qty".

  List define Inventory:
     define display():
        print "Item     Qty"
        for each item in me.itemNames:
           print item, me.qtys.key(item.name)


Possible benefits For developers using OO style languages (C++, SmallTalk, Java, etc), having a simple Inventory Object lets you easily include it in other areas of the game. Containers (combinations of Items and Inventory), Buildings (Structure and Inventory), and more can be made. Of course non-OO languages(C, Pascal, Basic) can use Inventory as functions in other parts of the game. The point is: once you define what a basic inventory will be in your game, find how to use it in more areas.

An Example Here is an example inventory, using the Python language. I find Python to be a grat language for prototyping. It is easy to spot errors, fix them, etc. Then you may easily recode in any other language.

The Item Object - class Item:

  def __init__(self, name):	#object constructor
     self.name = name

The Inventory Object - class Inventory:

  def __init__(self):
     self. NameList = []	#a list of item names
     self.qtys = {}		#a dictionary of item quantities
     self.contents = []	#a list of actual items
  def add(self, itemList = []):
     for item in itemList:
        #Check item is self
        if item is self:
           print 'Cannot store', item,' in itself!'
        if item in self.contents:
           print 'You already have this item!'
        else:
           #Check if item is in nameList
           if item.name in self.NameList:
              #merely insert
              self.qtys[item.name] = self.qtys[item.name] + 1
           #otherwise add to nameList
           else:
              self.qtys[item.name] = 1
              self.nameList.append(item.name)
           self.contents.append(item)
  def remove(self, item):
     #does item exist?
     If item not in self.contents:
        print item.name, 'is not in your inventory'
     else:
        self.qtys[item.name] = self.qtys[item.name] - 1
        if self.qtys[item.name] == 0:
           del self.qtys[item.name]
           self.NameList.remove(item.name)
        self.contents.remove(item)
  def display(self):
     #let's show everything!
     Print "Item\tQty"
     for item in self.NameList:
        print item,"\t", self.qtys[item]


More Info For information about Python, visit: http://www.python.org Please send all comments, queries, and error notifications to the author. Written by: Ken Power email: uncle_wiggly@bigfoot.com Version: 1.0 Date: Oct. 25, 1999