On Wednesday, October 9, 2002, at 05:15  PM, Andrew Henrick wrote:

>
> Q3.  How can I detect an underflow/overflow in the system.  This may  
> seem like an odd question, but it comes in reference to some code that  
> I found in "Object-Oriented Implementation of Numerical Methods" by  
> Besset.  He was working in Visual Age, so his exception handling is  
> different, but the idea is to cause an overflow by repetitive  
> multipilication:

Mmm I recall but only vaguely that in VisualWorks there was a changeset  
to introduce a 'really large' number, and a 'really small' number.  
(Can't remember if +/- infinity came with that too).

This was handy for those calculations were you wanted to work with  
those concepts.

Take a look on the web for 'double dispatch' and smalltalk and remember  
it's object (message send +) object
which leads to all sorts of interesting decisions (at a cost) you can  
do for the + message.

Mmm I wonder if someone has a changeset that considers degree of error  
in calculations handy.

--
======================================================================== 
===
John M. McIntosh <johnmci at smalltalkconsulting.com> 1-800-477-2659
Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
======================================================================== 
===