[MUD-Dev] Re: Reputations, More Mazes

Eli Stevens {KiZurich} c718157 at showme.missouri.edu
Sun Jan 17 03:21:11 New Zealand Daylight Time 1999


J. C. Lawrence:
>???  You questioned the signal to noise ratio of the list, or of your
>ability to contribute to the list?  Ouch.


Sorry, I was unclear.  In the past, I have not posted trivial ideas that
have been sparked by the list because I did not think that they would
significantly add to the conversation.  I am slowly getting a better
idea of where the line between brainstorming and spam is drawn here.  :)

The S:N of Mud-Dev is great.  Lurking here gives me a lot to ponder,
some of the views I agree with, some I do not, but they all make me=20
think.

Actually, I think I may have come across wrong a few times in my post,
(I will try and refrain from posting late at night in the future :) so
bear with me while I clarify (and apologize if I offended).

Me:
>> ...for a sophomore in college with a scant four years (mostly self
>> taught) of programming experience. =20

No judgement was intended on the value of self teaching here, I was=20
just giving a bit of my background.  :)

Me:
>> Quzah's thread on the "maze of the mind" system got me thinking on a
>> better way to model vast, twisting, untamed wilderness without
>> having to store thousands of rarely visited rooms (much less code
>> them).

"Better" refered to manually coding large mazes, not Quzah's method
of maze generation.

---

>In principle this is good stuff (and well workable).  The problem is
>translating it to a workable, interesting and believable environment
>within the game world.  Nobody here has yet done that, tho we have all
>nodded out heads wisely and said, "Yes, that's a good system that
>would work well in principle."  Implementation is a whole other bitch:
>
>  Okay, you have a random value.  Exactly how do you translate that
>into a generated room which is fairly consistent with the rooms
>surrounding it (encluding those NE/NW/SE/SW), or is at least
>believable without being a very simple variation of combinations of a
>few components in a room:
>
>    Tree?  Yes/No
>    Rock?  Yes/No
>    Pond?  Yes/No
>    etc.
>
>  Simple random combinations of the above makes for a mighty boring
>area.


Even with a decent paragraph splicer and fancy ways of saying the same
thing, I am afraid you are right.

Perhaps if the group carries three or four (or 17, whatever) bytes of=20
data along with the seed, and used them as restrictions to the possible=20
outcomes it might work better?

/*---*/

char *ground_desc[] =3D {
   "craggy and mountainous",
   "rough and broken",
   "gently hilled",
   "mostly flat"};

char *tree_desc[] =3D {
   "blue spruce"
   "pine and oak"
   "oak and silver maple"
   "maple and birch"};

char *underbrush_desc[] =3D {
   "carpets of needles"
   "leaves and thick ivy"
   "shrubs and berry bushes"
   "thick brambles"

byte ground_val =3D 2;
byte tree_val =3D 2;
byte underbrush_val =3D 0;

print( "Around you tower " + tree_desc[tree_val + rand(2)] +=20
       " trees, dappling the sun on the " +
       underbrush_desc[underbrush_val + rand(2)] +
       " that cover the " + ground_desc[ground_val + rand(2)] +
       " ground." );

/*---*/

A run of this might result in:

Around you tower oak and silver maple trees, dappling the sun on the
leaves and thick ivy that cover the gently hilled ground.

As long as the _val s stay the same, the land described will stay very
similar.  The values only change when the players move in a direction
specified to increase a value.

Ex:  Use the above simple example, but replace the arrays with sparsly
populated arrays of size 256 (sparse to allow easy additions and to=20
allow some items to happen more often), the rand(2) with rand(17) and
limit the byte _val to 256 - 17.  Now, when the players step n, ne, or
nw, add one to the tree_val.  Going s, sw or se subtracts one, and so
on for all the _val s (possibly with different directions).  Then add
more templates, make each paragraph a template, etc.  Of course, there
will always be some funky rooms, but...  over all I think it would
work, provided enough word choices were given. =20

I suppose the best way to find out would be to code it.  Ick.  :)

Silence is random
Eli - c718157 at showme.missouri.edu






More information about the MUD-Dev mailing list