Path: utzoo!attcan!uunet!know!sdd.hp.com!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!cica!iuvax!att!att!cbnewsk!ellson
From: ellson@cbnewsk.att.com (john.ellson)
Newsgroups: comp.object
Subject: Re: Re: How would yo model this situation?
Message-ID: <1990Oct29.151116.17943@cbnewsk.att.com>
Date: 29 Oct 90 15:11:16 GMT
References: <17486@rouge.usl.edu>
Organization: AT&T Bell Laboratories
Lines: 55

From article <17486@rouge.usl.edu>, by pcb@cacs.usl.edu (Peter C. Bahrs):
>> I wrote... Responses...
> 
> Well I have seen a lot of interesting responses on the net and through
> email.  I wish news allowed graphical postings so I could send a
> picture of the solutions.
> 
> But,
>   what happens to the rope object?  It breaks in two, but each
> piece has most of the characteristics of the original piece (except
> things like someone on the other end, length, ...).
>   The two people can still access their respective rope objects.
>   Is this still one rope object that now mangages 2 pieces instead
>   of one?  (this is simple to do)
>   Or should the rope object mutate?  i.e. break into two distinct
>    objects, each with a unique identity and local state.
>   (Is there an OO term for this?)
>   If this is the case, the old rope object must tell each person...
>     "...here, this is the new name of your rope object..." and then
>   the rope object must destroy itself.  Then the person object cannot
>   hard code a declartion to a rope object but keep an association or
>    delegation list somewhere.
> 
>              So,      | rope |   mutates, then disappears
>                         /  \
>                     rope2a  rope2b
>   
>   Any comments?  I suspect that all of this can be modeled in the
>   behaviors of the objects in languages like C++ and Smalltalk.  But
>   are there systems that provide mutation?
> 
> /*----------- Thanks in advance... --------------------------------------+
> | Peter C. Bahrs                                                         |
> | The USL-NASA Project                                                   |
> | Center For Advanced Computer Studies      INET: pcb@swamp.cacs.usl.edu |
> | 2 Rex Street                                                           |
> | University of Southwestern Louisiana      ...!uunet!dalsqnt!gator!pcb  | 
> | Lafayette, LA 70504                                                    |
> +-----------------------------------------------------------------------*/

Perhaps, because you know that the rope is breakable, you should model
rope-end as the objects instead of rope.  Then each person has a relationship 
with (is holding) a different rope-end instance where those rope-ends in turn
have a connected relationship with each other.

Then, when the rope breaks, only the connected relationship is
changed. The person-holding-rope-end structures are
unmodified.   The behavior of the connection relationship would be to
immediately instantiate two new rope-ends.

You could also model splicing by deleting two rope-end instances and
connecting the remaing rope-ends directly.


John Ellson   (j.ellson@att.com)