RE: Main Proc for Subproc not compiling

> From: Barbara Morris
>
> So I wouldn't use a pointer, I would just pass the DS by reference.  The
> caller can declare as many DS's as he wants.

Ugh.  What was I thinking?  This is of course the right answer.


> And instead of giving the caller a pointer to the DS itself in the first
> place, I'd just give an index into an array (or an occurrence).

This is an interesting design pattern.  Your locator method returns a key,
which you then use with each of your getter/setter methods.  With this
technique, the subprocedure is entirely responsible for data access,
including caching.  What's really interesting about this approach is that
you can conceivably support any manner of collision detection, including
dirty reads and simultaneous updates.

The only thing I don't like is writing having 42 different setters and
getters for any reasonably complex data entity.  There's also the additional
overhead of the procedure call and the data positioning for each call.  But
if you only use the methods to get data immediately after a locate and set
data immediately prior to an update, then the overhead compared to I/O is
probably negligible (though that depends on your caching, but now we're
wandering pretty far afield, and in general the overhead should be
acceptable).  And I can still use my "versioned" data structure method for
highly complex entities.

Thanks, Barbara.  As usual, one of your posts gives me a lot of food for
thought.

Joe