[Tutor] 2s complement binary for negative

[Dave Angel]

Ian D <duxbuz at hotmail.com> Wrote in message:
> Can anyone clarify please?
> 
> 
> Just reading this:
> 
> https://wiki.python.org/moin/BitwiseOperators
> 
> 
> The section on 2's complement binary for negative integers.
> 
> 
> It states:
> 
> 
> "Thus the number -5 is treated by bitwise operators as if it were written "...1111111111111111111011". "
> 
> 
> I am wondering why would this not be -4?
> 
> 
> I though -5 would be written ..1111111111111010
> 
> 
> 5 being 101b
>

That would be the case for a machine/language using ones-complement. 


 http://en.m.wikipedia.org/wiki/Ones'_complement

Twos-complement is usually explained by saying that for negative
 integers you take the ones-complement value and add one. But it
 might be more correct to say that for ones-complement machines
 you take the obvious result when subtracting numbers,  and if
 it's negative you subtract one from it.

Forty plus years ago I used a ones-complement machine,  the CDC
 6400, as well as a twos-complement one, the IBM 360. Since then,
 every architecture I used, and every one I implemented,  were
 twos-complement, including all the Intel and AMD and Motorola
 ones.

Twos-complement has some advantage in the hardware needed for most
 arithmetic,  but that doesn't matter to us as users. What does
 matter is when we mix arithmetic and logic, on the same value.
 And when we support more than one size of integer.

Probably the most visible difference is the fact that there are
 two values for zero in ones-complement,  all zero bits,  and all
 one bits. When doing arithmetic,  it's normal to consider them
 equal,  but when doing bitwise operations they're clearly
 different.  In fact on the CDC, False and True are usually +0 and
 -0. So on that architecture python would need two equality
 operators.  According to the page you referenced,  Python chose
 to hide the underlying architecture when running on an occasional
 ones-complement machine.  Good for them. 








-- 
DaveA