On Wednesday 07 August 2002 05:24 pm, David Laight wrote:
> I wrote code for one system that
> did arithemtic on decimal numbers (up to 10 digits) with
> a sign 'overpunch'.

Sounds like we're both former IBM mainframe programmers. :-)  The same 
systems had (and still have) hexadecimal floating point - the values are 
M*(16^S), not M*(10^S) as in IEEE.  

Of course, the latest machines in that family also have IEEE floats (thanks 
to Java and the scientific Fortran community) and instructions that covert 
between the two formats.  Such things make a hypothetical htonf() routine 
"fun".

Ross
-- 
Ross A. Patterson
CatchFIRE Systems, Inc.
5885 Trinity Parkway, Suite 220
Centreville, VA  20120
(703) 563-4164