On Dec 2, 2005, at 10:28 AM, James Newtons Massmind wrote:

> Any yet there is not a SINGLE computer language that tracks the actual
> precision of calculations! E.g. When you multiply 1 byte times another 
> 1
> byte the result should be stored in 2 bytes

Um, a fair number of languages and compilers track assorted arithmetic
exceptions, don't they?  Even if only to the extent of printing an
error message and aborting...

> http://www.sxlist.com/techref/expeval2.asp is my poor attempt to do
> such a thing. Play with it and tell me why programmers don't demand
> that ability in their compilers?
>
Because most programmers treat numeric precision the same way that
network protocol designers treat bandwidth: just provide much more
than is actually needed, and you won't have to worry about it.  Thus
"integers" are usually 32 bits these days, and 80 bit floats, although
I can remember days where 32 to 36 bit floats were the norm...

And we can all think of plenty of examples of just how badly this scheme
has worked for strings input from the 'real world.'  Sigh.

BillW
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist