difference between ~0U and ~0

Hello All,

I am confused with the behaviour of the following program
#include<stdio.h>

int main(void)
{
printf("\n~0U = %d",~0U);
printf("\n~0 = %d",~0);
printf("\n~0U>>1 = %d",~0U>>1);
printf("\n~0>>1 = %d\n",~0>>1);
return 0;
}

==========
Output
=========

~0U = -1
~0 = -1
~0U>>1 = 2147483647
~0>>1 = -1
===========
According to my understanding ~0 is -1.So I will get all bit 1 as a
result of ~0 in 2's complement system.

So the first two line of output saying ~0 is same as ~0U. Then why
there is difference in the output of ~0U>>1 and ~0>>1 ?
please provide some input.

On 03-Mar-12 22:03, somenath wrote:
> Hello All,
>
> I am confused with the behaviour of the following program
> #include<stdio.h>
>
> int main(void)
> {
> printf("\n~0U = %d",~0U);

~0U has type (unsigned int); the appropriate format specifier is %u, not %d.

> printf("\n~0 = %d",~0);

~0 has type (int), so %d is correct.

> printf("\n~0U>>1 = %d",~0U>>1);

~0U>>1 has type (unsigned int); the appropriate format specifier is %u,
not %d.

> printf("\n~0>>1 = %d\n",~0>>1);

~0>>1 has type (int), so %d is correct.

> return 0;
> }
>
> ==========
> Output
> =========
>
> ~0U = -1
> ~0 = -1
> ~0U>>1 = 2147483647
> ~0>>1 = -1
> ===========
> According to my understanding ~0 is -1.

That is true for if the zero's type is signed. However, "0U" is
unsigned, so the result of the ~ operator cannot be negative.

> So I will get all bit 1 as a result of ~0 in 2's complement system.

That is an implementation detail and will only confuse you at this point.

> So the first two line of output saying ~0 is same as ~0U.

Perhaps on your system, but your code invokes undefined behavior by
using the wrong format specifier, so anything is possible.

> Then why there is difference in the output of ~0U>>1 and ~0>>1 ?

Fix the bugs in your code and you'll see that you're asking the wrong
questions.

S


--
Stephen Sprunk "God does not play dice." --Albert Einstein
CCIE #3723 "God is an inveterate gambler, and He throws the
K5SSS dice at every possible opportunity." --Stephen Hawking

On 3/3/2012 9:03 PM, somenath wrote:
> Hello All,
>
> I am confused with the behaviour of the following program
> #include<stdio.h>
>
> int main(void)
> {
> printf("\n~0U = %d",~0U);
> printf("\n~0 = %d",~0);
> printf("\n~0U>>1 = %d",~0U>>1);
> printf("\n~0>>1 = %d\n",~0>>1);
> return 0;
> }
>
> ==========
> Output
> =========
>
> ~0U = -1
> ~0 = -1
> ~0U>>1 = 2147483647
> ~0>>1 = -1
> ===========
> According to my understanding ~0 is -1.So I will get all bit 1 as a
> result of ~0 in 2's complement system.
>
> So the first two line of output saying ~0 is same as ~0U. Then why
> there is difference in the output of ~0U>>1 and ~0>>1 ?
> please provide some input.

because ~0U is unsigned, but ~0 is signed.
'>>' does different things for signed and unsigned values.

somenath <pal_somenath@emc.com> writes:

> I am confused with the behaviour of the following program
> #include<stdio.h>
>
> int main(void)
> {
> printf("\n~0U = %d",~0U);
> printf("\n~0 = %d",~0);
> printf("\n~0U>>1 = %d",~0U>>1);
> printf("\n~0>>1 = %d\n",~0>>1);
> return 0;
> }
>
> ==========
> Output
> =========
>
> ~0U = -1
> ~0 = -1
> ~0U>>1 = 2147483647
> ~0>>1 = -1
> ===========
> According to my understanding ~0 is -1.So I will get all bit 1 as a
> result of ~0 in 2's complement system.
>
> So the first two line of output saying ~0 is same as ~0U. Then why
> there is difference in the output of ~0U>>1 and ~0>>1 ?
> please provide some input.

I don't think anyone has said yet that the result of right shifting a
negative value is, explicitly, implementation-defined. Some machines
will do one things and some another. The C standard leaves it up to the
implementation to say what it does. The result of ~0U >> 1 is defined
by the language -- it must be UINT_MAX/2 (that's C's division
there, of course, not mathematical division) but ~0 >> 1 can be anything
the implementation chooses. The two most likely possibilities are
INT_MAX and -1.

--
Ben.

somenath wrote:

> According to my understanding ~0 is -1.
> So I will get all bit 1 as a
> result of ~0 in 2's complement system.

Close, but backwards.

~0 is all bit 1.

So you will get -1 as a
result of ~0 in 2's complement system.

--
pete

On Thursday, March 8, 2012 8:55:20 AM UTC-5, pete wrote:
> somenath wrote:
>
> > According to my understanding ~0 is -1.
> > So I will get all bit 1 as a
> > result of ~0 in 2's complement system.
>
> Close, but backwards.
>
> ~0 is all bit 1.
>
> So you will get -1 as a
> result of ~0 in 2's complement system.
>
> --
> pete

That's true.

1 in 8-bit format is: 0000 0001
To get neg- 1:
1. Negate bits. 1111 1110
2. Add 1: 1111 11111

Same results regardless of the bit size.

James Piper wrote:
>
> On Thursday, March 8, 2012 8:55:20 AM UTC-5, pete wrote:
> > somenath wrote:
> >
> > > According to my understanding ~0 is -1.
> > > So I will get all bit 1 as a
> > > result of ~0 in 2's complement system.
> >
> > Close, but backwards.
> >
> > ~0 is all bit 1.
> >
> > So you will get -1 as a
> > result of ~0 in 2's complement system.
> That's true.
>
> 1 in 8-bit format is: 0000 0001
> To get neg- 1:
> 1. Negate bits. 1111 1110
> 2. Add 1: 1111 11111
>
> Same results regardless of the bit size.

Another way for negatising in two's complement is:
1. subtract 1. 0000 0000
2. Negate bits: 1111 11111

I've worked on assembly code
that had it done either way in different parts of the code.
I don't know if it was done by the same person.

--
pete

pete <pfiland@mindspring.com> writes:

> James Piper wrote:
>>
>> On Thursday, March 8, 2012 8:55:20 AM UTC-5, pete wrote:
>> > somenath wrote:
>> >
>> > > According to my understanding ~0 is -1.
>> > > So I will get all bit 1 as a
>> > > result of ~0 in 2's complement system.
>> >
>> > Close, but backwards.
>> >
>> > ~0 is all bit 1.
>> >
>> > So you will get -1 as a
>> > result of ~0 in 2's complement system.
>> That's true.
>>
>> 1 in 8-bit format is: 0000 0001
>> To get neg- 1:
>> 1. Negate bits. 1111 1110
>> 2. Add 1: 1111 11111
>>
>> Same results regardless of the bit size.
>
> Another way for negatising in two's complement is:
> 1. subtract 1. 0000 0000
> 2. Negate bits: 1111 11111
>
> I've worked on assembly code
> that had it done either way in different parts of the code.
> I don't know if it was done by the same person.

The first method turns x into -x; the second turns -x into x.

Remember, only 22 more shopping days before March 32.