[om-list] Re: System Design

[Mark Butler]

Luke Call wrote:

> You mentioned the many mathematical rules one would want to store and
> utilize in our system, and how a pure  OO Model would do it, and I
> replied something about constraints and rules inside objects. But that
> was an inadequate answer. It does seem like we need a body of such rules
> somewhere--how would objects in a simulation know which one to use at a
> given time? 

Any given simulation needs to be run from the perspective of a given person as
stored in a particular schema. Each schema may trust whole bodies of knowledge
generated by other parties, but may override that trust on an exception basis,
replacing a distrusted statement or theory with an alternative or nothing at
all.

> Or, how would a system where knowledge is represented
> strictly by logical statements do a simulation, how would it relate one
> piece of information to another (I guess that's where you were headed
> with namespace?), and do such simulations in a way where load time and
> object traversal / query performance are reasonable?

Logical statements can be stored in an indexed database just like anything
else.  It is just an artifact of present technology that there are no common
persistent storage systems oriented toward storing large collections of
complex logical expressions.

Storing large collections of restricted logical expressions, e.g. sequences of
numeric tuples from the output of a simulation, is identical to what
traditional databases do now.

> Would the system load all these logical statments into objects & run it
> from there? It sounds cumbersome but maybe I'm not visualizing it the
> way you are.

Well of course we need a real database eventually - I just think we could
start on a long list of necessary modules with an in-memory database that
loads and stores everything at once to get started.

Real databases are notoriously difficult to develop - Oracle is making money
hand over fist while their competitors struggle along with very similar
products solely because their database handles heavy workloads and
sophisticated operations in a much smoother manner.  Just the database work
could easily take a couple of decades to mature with a large body of
experienced contributors.  Look at where Linux is at as it approaches the ten
year mark - in about ten more years it will truly be mature.


-- 
Mark Butler	       ( butlerm at middle.net )
Software Engineer  
Epic Systems              
(801)-451-4583