omega/omega/omega.bug

Message-Id: <32D0F245.5510@atlcom.net>
Date: Mon, 06 Jan 1997 07:38:29 -0500
From: Sheldon Simms <sheldon@atlcom.net>
Mime-Version: 1.0
To: wtanksle@mailhost2.csusm.edu
Subject: More Omega
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Status:

Since I was looking through old mail... This is a message I sent
to Erik Max Francis in late November

I was trying to find out why I sometimes encoutered tangible (!m_statusp( m,
INTANGIBLE )) monsters sitting on unrevealed secret passages. The way this
looks to the player is that he sees a tangible monster, like a goblin,
sitting there but when he tries to attack, Omega prints "Ouch!".

The function m_unblocked() in util.c determines whether a monster can move
onto a space. It explicitly disallows tangible monsters from moving onto a
secret space unless they are smart, in which case they can "open" the secret
door or passage, which then becomes non-secret.

So I thought that tangible monsters might be getting put on secret walls
when the level is populated. After tracking down the code, I found that the
function findspace() (also in util.c) finds a space to put a new monster
when the level is being populated. findspace() didn't care whether a space
was secret, only whether the locchar was FLOOR, which it is for secret
passages. So I figured (and still think) that modifying findspace() to
disallow secret spaces has solved that problem.

However, as I read findspace() I immediately thought about Laurence's
comment about monsters having a party because the findspace was written,
this was very likely to happen. Basically it picked a random space on the
level and if that space wasn't ok, it started searching across and down for
the first acceptable empty space. So if the uppermost room or corridor on a
level was on (for example) row 16, with every row above that being filled
with WALLs, 25% (16/64) of all the monsters on the level will be in the
upper-left-most room or corridor - they're having a party!

The reason I'm telling you all this is that I rewrote findspace() to
eliminate the party effect. In my new findspace(), if the randomly chosen
space is not ok, it searches either across or down (alternating each time a
space is selected at random). If it finds no acceptable space in the row or
column it searched, it tries another random space. Obviously it could try
one random space after another, but I was wary of the possibility that such
a loop might run for quite a while.

Anyway, the code follows for inclusion in the official source, if you choose
to use it:

/* finds floor space on level with buildaux not equal to baux,
sets x,y there. There must *be* floor space somewhere on level.... */

int spaceok( int i, int j, int baux )
{
    return (( Level->site[ i ][ j ].locchar == FLOOR ) &&
            ( Level->site[ i ][ j ].creature == NULL ) &&
            ( !loc_statusp( i, j, SECRET )) &&
            ( Level->site[ i ][ j ].buildaux != baux ));
}

void findspace( int *x, int *y, int baux )
{
    int i, j, tog = TRUE, done = FALSE;

    do {
        i = random_range( WIDTH );
        j = random_range( LENGTH );
        if ( spaceok( i, j, baux ))
        {
            done = TRUE;
        }
        else
        {
            if ( tog )
            {
                tog = !tog;
                while( 1 )
                {
                    i++;
                    if ( i >= WIDTH )
                        break;
                    else if ( spaceok( i, j, baux ))
                    {
                        done = TRUE;
                        break;
                    }
                }
            }
            else
            {
                tog = !tog;
                while( 1 )
                {
                    j++;
                    if ( j >= LENGTH )
                        break;
                    else if ( spaceok( i, j, baux ))
                    {
                        done = TRUE;
                        break;
                    }
                }
            }
        }
    } while ( !done );
    *x = i;
    *y = j;
}


--
W. Sheldon Simms III
sheldon@atlcom.net


--
W. Sheldon Simms III
sheldon@atlcom.net
Received: from future.atlcom.net by owl.csusm.edu (AIX 4.1/UCB 5.64/4.03)
          id AA07766; Mon, 6 Jan 1997 04:28:38 -0800
Received: from lpm2-10.atlcom.net by future.atlcom.net with SMTP (5.65/1.2-eef)
	id AA20687; Mon, 6 Jan 97 07:37:38 -0500
Return-Path: <sheldon@atlcom.net>
Message-Id: <32D0F32E.74F@atlcom.net>
Date: Mon, 06 Jan 1997 07:42:22 -0500
From: Sheldon Simms <sheldon@atlcom.net>
X-Mailer: Mozilla 2.0 (Macintosh; I; 68K)
Mime-Version: 1.0
To: wtanksle@mailhost2.csusm.edu
Subject: Still More Omega
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Status:
X-Status: R

here's another mail I sent to Max in November:


    I just want to tell you about a few of the more dangerous bugs I have
found in Omega.

1) In inv.c, the function givemonster(), when grain is given to a
   horse, the grain object is freed:

       else
           nprint1( "...and now seems satiated." );
       morewait();
       free( (char *)o );

   This is a bug because when givemonster() returns to give() in
   command2.c, give() tries to access the object:

       givemonster( m, obj );
       print2( "Given: " );
       nprint2( itemid( obj ));

   This sometimes caused "bus error" crashes on my Mac. I have
   simply commented out the free() statement for now. I'll probably
   leave it that way and simply accept the memory leak.

2) In guild1.c in the function l_arena(), after you kill your arena
   opponent, the monster is freed:

       print1( "Let the battle begin...." );

       time_clock( TRUE );
       while ( Current_Environment == E_ARENA )
           time_clock( FALSE );

       free(name);        /* <- bad */
       free(corpse);      /* <- bad */
       if (melee)         /* <- bad */
           free(melee);   /* <- bad */

    This is incorrect because the dead monster can be picked up and
    carried out of the arena. The four lines marked "bad" should be
    removed.

3) The star-gem bug. This happens in save.c because the possessions of
   hiscore_npc monsters are not saved. It is easily fixed by moving one
   line in each of the functions save_monsters() and restore_monsters().
   In save_monsters():

       if ( tml->m->id != HISCORE_NPC )
       {
           /* several lines snipped */

           /* ok &= save_itemlist( fd, tml->m->possessions ); <- move this */
       } /* else it'll be reloaded from the hiscore file on restore */
       ok &= save_itemlist( fd, tml->m->possessions );     /* <- to here */

   This change causes possessions to be saved for all monsters. The change
   to restore_monsters() complements the change to save_monsters():

       if ( ml->m->id == HISCORE_NPC )
       {
           /* stuff deleted */
       }
       else
       {
           /* more stuff deleted */

           /* ml->m->possessions = restore_itemlist( fd ); <- move this */
           ml->m->meleestr = Monsters[ ml->m->id ].meleestr;
       }
       ml->m->possessions = restore_itemlist( fd );     /* <- to here */

4) There is also another bug that I haven't fixed yet. This involves
   the time when you bash a statue and it glides into the floor leaving a
   stairway down. When this happens on levels where there is not supposed
   to be stairway down, taking the created stairway can cause problems. I
   just encountered this problem on the 5th level of the astral plane (the
   high astral plane). Trying to take the stairway down caused a crash.

--
W. Sheldon Simms III
sheldon@atlcom.net
Received: from mail.Virginia.EDU by owl.csusm.edu (AIX 4.1/UCB 5.64/4.03)
          id AA16322; Mon, 20 Jan 1997 06:20:52 -0800
Received: from uva.pcmail.virginia.edu by mail.virginia.edu id aa25972;
          20 Jan 97 9:31 EST
Received: (from root@localhost) by uva.pcmail.virginia.edu (8.7.6/8.6.6) id JAA04707 for wtanksle@owl.csusm.edu; Mon, 20 Jan 1997 09:31:03 -0500 (EST)
Message-Id: <199701201431.JAA04707@uva.pcmail.virginia.edu>
From: Kent Peterson <kmp2x@uva.pcmail.virginia.edu>
Date: Mon, 20 Jan 97 09:30:48 EST
X-Mailer: UVa PCMail 1.9.0
To: wtanksle@mailhost2.csusm.edu
Subject: Omega bugs
Status:
X-Status: R

l_safe - siren goes off, need morewait afterwards

send_to_jail - two of the release options have the x/y coords
        reversed (subtle!)

l_tavern - sleeping needs to change the date, drinking needs to
        pass time

l_library - cost of studying should be moved so as to apply only
        to option e, not any of the others

1) m_talk_druid - non-druid characters always have their alignment
        *decreased* by neutralization?  that's what seems to be
        happening here ...

2) m_sp_mb - manaburst explosion doesn't kill it, anything that
        explodes itself should die by my lights.

3) m_sp_dragonlord - dragonlord should be capitalized everywhere

4) boots/spells of heroism should give immunity to fear

5) l_void - need a morewait after "Death peers quizzically ..." etc

6) l_escalator - need a morewait after stairs start moving

7) m_simple_move - if blocked on straight line and both diagonals,
        try moving randomly once

8) casino closed for investigation - need to make this last the
        entire day

9) what is the difference between m_smart_move and m_normal_move?
        A planned one?

10) transcribe_monster_actions - when figuring attacks, when deciding
        to attack center, it blocks instead

11) void i_symbol - for loop cutting hp and for loop cutting items
        should be 2 separate loops (?)
        also, if you use the symbol twice on the same hour on
        succeeding days, your god gets angry; there should be a
        SymbolDay variable as well
        (same with crystal ball) (and in i_enchantment)
        (and i_helm)

12) holy hand grenade (i_antioch) - count>3 / count<3 are the wrong
        way around; should be count>3 for monster to have time to
        throw it back

13) Sceptre of High Magic - we need 2 HiMagic variables here, one for
        the sceptre and one for the throneroom (same for the
        throne's routine)

14) berserk attacks while shadowformed - indefinitely

15) i_perm_breathing fun - check p_drown to make sure it gives you
        the normal chance to drop/bash/etc to survive

16) surrendering to an undead guardsman

17) saving in countryside is broken

18) area effect (snowball) targeted on edge square (so that part of
        the area is off the edge of the map)

19) pressing escape while targetting staff (of disruption) is broken

20) "Really.  Well the police are being notified" - more always
        appears, shouldn't

21) killing a monster with riposte does not leave the normal treasure
        pile that monsters carry; the * character gets printed,
        but there's nothing there

22) restoring from a save when near a monster often will make the
        monster "sleep" for a dozen rounds while you whack away
        safely at it

23) Slowness - should be temporary (a dozen rounds/seconds/minutes).  Right
        now it's so crippling it's better to save/restore than go
        through it.
        (correction, it *is* temporary.  tested it.  it's just
        way too long-lasting.)

24) Shadow Form - should lose immunities at end (plain bug)
        when the spell aborts, the various immunities should be
        set to 0, rather than being decremented
        (same thing with slowness?)

25) breathing stat (powered armor etc) - when this
        gets cursed or reversed, it ought to call the drown
        function rather than simply killing you

26) dispel - should also remove slowness and/or shadowform

27) names for several temples should be reconciled between strategic
        teleports and *roomname

28) hint something about how azoth needs to be blessed
        "They say blessings can really change an item's attitude."

29) Encounters - fewer high-powered magical beasties on roads and plains;
        wolves and lions are bad enough.  Too easy to die while
        traveling.

30) Sacrificing - lower curve of required values; it's too hard to get
        something worth giving as it stands (wands/staves should be
        possible), especially at middle levels.  Reasonable values would
        be 200 gp at level 10, current values are more like 2-3 thousand

31) Fire resistance - should be partial (depending on pluses), rather than
        total (too easy to kill dragons & Elemental Lord of Fire)
        (same with deflection - missile attacks become laughable,
        should depend on plusses to shield, or total value of
        spell cast - the longer the spell has to last, the more
        effective it is?)
        resistance to electricity/cold damage probably also
        applies here; fire really stands out though

32) DragonLord - should have a chance of waking up every round someone's in
        the inner cave; it's too easy to steal repeatedly
        Also, when games are saved and restored within the cave, the
        portcullis should restore as down

33) Prime Sorceror - should always be created with the Star Gem when in the
        Circle plane, even if game saved/restored while in Astral

34) Law/Chaos worship stones - these should depend a lot more on alignment;
        as it stands it seems like even if you're very lawful the Star
        View stone has pretty good chances of feeding you something
        nasty; on the other hand, the reward shouldn't be as big (make
        the stat increase random, 1-6?)

35) Gym training - far too expensive; I *always* save/restore to make sure
        the training pays off because I can't afford to throw away that
        kind of money.  A cost of 300-800 is more like it, or
        maybe 1000 with half to 3/4 of your money back if your
        ability doesn't improve.  (or maybe the refund gets added
        to your credit)

36) 7 league boots - each time used, add 1 to fragility, when reaches
        terminal, "your boots dissolve into dust with a soft sigh"
        (right now they break the game - the countryside becomes
        irrelevant) (esp. when combined with robbing the DragonLord)

37) Gambling - should take time (~10 mins per game), and should be more
        detailed (what the heck does it *do*, besides randomly rolling
        electronic dice to see if you gain or lose money?  what are the
        rules of the game?)

38) Tavern - have drinking take time (1 drink = 2-10 minutes, a round
        = 10-30 minutes) - we need to be able to while away the
        night hours

39) Explorer's Club - needs to sell more stuff - maybe stimtabs/nutritabs/
        powtabs/immunotabs, for appropriate prices? (100/200/1000/50)
        Also, the "listen for rumors" option could be expanded to
        something more like a daytime tavern; you could have a
        drink/snack and hear the rumors meanwhile (and have it take ~5
        minutes)

40) The mercenary legion should sell crossbows/longbows/arrows/bolts,
        at 1/10 price to legion members - then they WOULD be the
        "best equipped" (arrows and bolts especially, those are hard
        to find)

41) boots of leather/cloaks of wool ought to add 1 to defense, to
        have a reason to use them aside from defense from
        disintegration


Kent Peterson                 It is easier to ask for forgiveness
kmp2x@virginia.edu                 than to ask for permission.
kent@dirac.physics.jmu.edu
          Abortions should be safe, legal, early, and rare.
Received: from Tree.EGR.UH.EDU by owl.csusm.edu (AIX 4.1/UCB 5.64/4.03)
          id AA20516; Thu, 13 Feb 1997 16:12:26 -0800
Received: from lonestar by tree.egr.uh.edu (NX5.67f2/NX3.0M)
	id AA11355; Thu, 13 Feb 97 18:23:29 -0600
From: David J Robertson <djr98665@tree.egr.uh.edu>
Message-Id: <9702140023.AA11355@tree.egr.uh.edu>
Received: by lonestar.egr.uh.edu (NX5.67e/NX3.0X)
	id AA08345; Thu, 13 Feb 97 18:23:28 -0600
Subject: Re: Omega: Bug reports
To: wtanksle@mailhost2.csusm.edu
Date: Thu, 13 Feb 1997 18:23:27 -0600 (CST)
In-Reply-To: <WQ7/ygUEFd7V089yn@owl.csusm.edu> from "William Tanksley" at Feb 10, 97 06:49:42 pm
X-Mailer: ELM [version 2.4 PL23]
Content-Type: text
Content-Length: 6580
Status: RO
X-Status: R

>
> David, thank you!  That was a perfect bug report for me -- even gave the way
> to fix it :).

  No problem.
>
> I've got some good news of my own -- it just MIGHT take a little less time
> to do this than I'd thought.  Someone else had already started it :).
> That's real good news; not only does it mean less work from me, but it also
> means everyone gets the game sooner.

   Cool!
>
> As for your suggestion -- I don't know.  The first one (have "know all
> spells" effects give MORE knowledge of already-known spells) makes sense,

  Hmm, the only "know all spells" effect I know of is using the
key that was lost.  For this item, I agree that reducing the cost
in addition might be too good.  I was really talking about the
extra spells you get from the guilds.  They don't help the spell
cost if you already know the spell they teach you.

  Well, here is another bug/suggestion.  Have fun. :)


I was trying to track down a bug with ghost items/no items being
left on a monsters death, and found it.  The following is pseudo
code for m_pulse in mon.c.

m_pulse(m){
  gain a hit point every so often
  if asleep, wake up if player is in range
  if awake
   if wandering
    move around, then reset wandering if in range
   else
    do the monster mstrike 1/2 the time if more than 1 space away
    (some) monsters move toward the player if more than 1 space away
(1) hostile monsters attack the player if 1 space away
(2)greedy monsters pick up stuff
   do monsters special function (if in range)
 }

  Now the problem is in lines (1) and (2).  During the attack the
 player phase, the monster can be killed by a riposete.  If this
 happens, then the monster drops its stuff, and its hp are set to
 -1.  Then the code returns to m_pulse.  The (dead) monster picks
 up everything it dropped, including its corpse.  Then the (dead)
 monster does its special attack.  Then, back in time_clock in
 time.c, the monster is deallocated.  However the stuff it picked
 up is lost, since the code in time_clock assumes that the monster
 dropped everything on death, leading to a memory leak.

  When I was looking at the m_pulse, I also notice that the monsters
 can do an awful lot in each melee round, with the worst situation
 occuring when it is two spaces away.  Then it can:

   (1) do a ranged attack (1/2 the time)
   (2) move next to you
   (3) do a melee attack
   (4) do its special attack (even if dead)

  I think it would be good idea to change the flow code so that
 if the monster does a ranged attack, it won't move (or maybe it
 should only have a chance at moving).  Then if the monster does
 move, it shouldn't get a chance to do a melee attack.  There
 should be a random chance for the special attack to occur.
 There should also be a hitpoint check placed on all monster
 actions after the tacmonster function has a chance of being
 called, to prevent the bugs I mentioned above.

  I've enclosed a revised m_pulse function below (the original
m_pulse follows it).

/*               Revised function                   */
/* consider one monster's action */
void m_pulse(m)
struct monster *m;
{
  int range = distance(m->x, m->y, Player.x,Player.y)
  int STRIKE=FALSE;
  pol prev;

  if (Time % 10 == 0)
    if (m->hp < Monsters[m->id].hp)
      m->hp++;

  if ((! m_statusp(m,AWAKE)) && (range <= m->wakeup)) {
    m_status_set(m,AWAKE);
    resetgamestatus(FAST_MOVE);
  }

  if (m_statusp(m,AWAKE)) {
    if (m_statusp(m,WANDERING)) {
      if (m_statusp(m,MOBILE)) m_random_move(m);
      if (range <= m->sense && (m_statusp(m, HOSTILE) ||
          m_statusp(m, NEEDY)))
        m_status_reset(m,WANDERING);
    }
    else /* not wandering */ {
      if (m_statusp(m,HOSTILE))
        if ((range > 2) && (range < m->sense) && (random_range(2) == 1)
            (los_p(m->x,m->y,Player.x,Player.y) &&
            (Player.status[INVISIBLE] == 0)) {
           STRIKE=TRUE;
           monster_strike(m);
        }

      if ((m_statusp(m,HOSTILE) || m_statusp(m,NEEDY))
          && (range > 1) && m_statusp(m,MOBILE)) &&
           (!STRIKE || (random_range(2) == 1))
        monster_move(m);
      else
        if (m_statusp(m,HOSTILE) && (range ==1)) {
          resetgamestatus(FAST_MOVE);
          tacmonster(m);
        }
    }
    /* if monster is greedy, picks up treasure it finds */
    if (m_statusp(m,GREEDY) && (m->hp >0) )
      while (Level->site[m->x][m->y].things != NULL) {
        m_pickup(m,Level->site[m->x][m->y].things->thing);
        prev = Level->site[m->x][m->y].things;
        Level->site[m->x][m->y].things =
          Level->site[m->x][m->y].things->next;
        free((char *) prev);
      }
    /* prevents monsters from casting spells from other side of dungeon */
    if ((range < max(5,m->level)) && (m->hp > 0) &&
        (random_range(2) == 1))
      monster_special(m);
  }
}


/*                 Original Function                     */
/* consider one monster's action */
void m_pulse(m)
struct monster *m;
{
  int range = distance(m->x, m->y, Player.x,Player.y);
  pol prev;

  if (Time % 10 == 0)
    if (m->hp < Monsters[m->id].hp)
      m->hp++;

  if ((! m_statusp(m,AWAKE)) && (range <= m->wakeup)) {
    m_status_set(m,AWAKE);
    resetgamestatus(FAST_MOVE);
  }

  if (m_statusp(m,AWAKE)) {
    if (m_statusp(m,WANDERING)) {
      if (m_statusp(m,MOBILE)) m_random_move(m);
      if (range <= m->sense && (m_statusp(m, HOSTILE) || m_statusp(m, NEEDY)))
        m_status_reset(m,WANDERING);
    }
    else /* not wandering */ {
      if (m_statusp(m,HOSTILE)) {
        if ((range > 2) && (range < m->sense) && (random_range(2) == 1))
          if (los_p(m->x,m->y,Player.x,Player.y) &&
              (Player.status[INVISIBLE] == 0)) monster_strike(m);
     }
      if ((m_statusp(m,HOSTILE) || m_statusp(m,NEEDY))
          && (range > 1)
          && m_statusp(m,MOBILE)) {
        monster_move(m);

      }
      if (m_statusp(m,HOSTILE) && (range ==1)) {
        resetgamestatus(FAST_MOVE);
        tacmonster(m);
      }
    }
    /* if monster is greedy, picks up treasure it finds */
    if (m_statusp(m,GREEDY))
      while (Level->site[m->x][m->y].things != NULL) {
        m_pickup(m,Level->site[m->x][m->y].things->thing);
        prev = Level->site[m->x][m->y].things;
        Level->site[m->x][m->y].things =
          Level->site[m->x][m->y].things->next;
        free((char *) prev);
      }
    /* prevents monsters from casting spells from other side of dungeon */
    if (range < max(5,m->level))
      monster_special(m);
  }
}

David Robertson