[MUD-Dev] Virtual Chemistry

[clawrenc at cup.hp.com]

In <Pine.LNX.3.96.970801224918.136Q-100000 at mpc.dyn.ml.org>, on
08/01/97 
   at 03:09 PM, Matt Chatterley <root at mpc.dyn.ml.org> said:

>On Wed, 30 Jul 1997 clawrenc at cup.hp.com wrote:

>> First thoughts on a representational model:
>> 
>>   Create a coordinate system.  
>>   The axis of the coordinate system are the principles of your
>>     alchemy system (yes, lotsa dimensions).
>>   The purpose of the coordinate system is to define the 
>>     behaviour of a substance.
>>   Given a substance, a coordinate location can be computed from 
>>     its components.
>>   Given a coordinate location, the effect of the substabce can 
>>     be derived.

>I was initially unsure about this, but the more I thought - the more
>I liked it, and, for what I want to create, it makes perfect sense,
>and seems to work nicely (at least theoretically).

FWIW I'd strongly recommend using a sparse array and interpolating the
missing chunks at runtime.

>> Thus the leaves of the herb WaggaWagga might have a coodinate weight
>> of (-5, 8, 3) presuming the simplistic case of only 3 principles.  The
>> root of the UmbaUmba plant might have a value of (2, 9, -10).  Mix
>> then together with GooGoo mud (0, 0, 2), and you get a substance (-3,
>> 17, -5).  That location can be plotted in the coordinate system and a
>> result computed:
>> 
>>   A minor weakening agent (-3) that also a *really* effective sleeping
>> draught (17), and reduces physical dexterity (-5).

>Yup. Of course, you'd need many more properties tracked.. but still,
>we have the basic spirit captured here. First things which came to
>mind were properties to track reactivity of substances, acidity, and
>magical capacitance. More general things with relevant to the mixing
>of chemicals. A question rose subsequently here - what about by
>products? How do we tell if a by product should be produced? I have
>yet to answer that one.

Yeah, the above example is deliberately simplistic and trivial.  

By products are really a side effect of standard combinations.  Some
sort of indexed reaction is needed.  If you further extend the array
as follows, it can be done:

  Each substance has a unique ID value in its array.

  Each substance array also encludes a member which is a list of
arrays (maps really) which represents what quantities of what products
are produced when the current substance is mixed with the named (by
ID) substances.  

>Incidentally, bringing in the concept of tracking state produces a
>flaw in the design of many stock (LP) bases - you have a drink
>inheritable, which all drinkable substances must include, defining
>alchoholic properties etc (this is pretty standard accross LP,
>excepting very different games). Of course, this is not going to
>really work with liquids - in theory you can drink any liquid, but
>you don't want to inherit drink into all of them!

The problem is a bad object model.  They should have a state base
class from which a liquid class is descended (or is a base class).