Thread: question

The unsigned means that the type has a non-negative range but same size and alignment requirements as the corresponding signed type (i.e., signed char and int in this case). For DWORD, read MSDN.

Originally Posted by lolguy

whats the unsinging char and int ?

The unsinging character is mute.

Just kidding! "unsigned" means it has no positive or negative sign (which means it is always a positive number). Without the sign, you have an extra bit to use as part of the stored value. Everytime you add a bit you double the possible range of values, so while a normal char can be from 0 to 127 (7 bits) with the sign (- or +), an unsigned char is a full 8-bit number, 0-255, but it cannot be set negative.

Originally Posted by MK27

so while a normal char can be from 0 to 127 (7 bits) with the sign (- or +)

That would be the case for sign and magnitude and one's complement, but with two's complement the range of an 8-bit signed char would be [-128,127]. Also, it is implementation defined whether char is signed or unsigned.

Originally Posted by lolguy

what about unsinged char ? how will it non nagative and its just a char? ?

You might want to rephrase that using proper English. I am afraid that I do not understand your question considering what MK27 and I have stated.

Originally Posted by lolguy

what about unsinged char ? how will it non nagative and its just a char? ?

The unsinged character avoids open flame. Ha ha ha again! Don't even ask about the unhinged char.

A char is a value. These values translate to alphanumeric symbols when you use printf, etc. Have a look at the ASCII table. So for example:

Code:

char X='X';
char X=88;

are exactly the same thing. The first form is for convience.

Originally Posted by laserlight

with two's complement the range of an 8-bit signed char would be [-128,127]. Also, it is implementation defined whether char is signed or unsigned.

I did not know this -- I suppose that's because we don't need a "-0"?

Originally Posted by MK27

I did not know this -- I suppose that's because we don't need a "-0"?

I am not sure if we can consider the lack of a negative zero as the reason (perhaps the fact that the sign bit has a -(2^N) value instead of -(2^N - 1) value is the reason), but either way it is the way the mathematics works out.

Originally Posted by laserlight

it is the way the mathematics works out.

Hey look my 666th post!

That the sign is the first bit and that if set the remaining bits are flipped I can understand. If this is the math:

Originally Posted by Wikipedia, the free encyclopedia

The two's complement of a binary number is defined as the value obtained by subtracting the number from a large power of two and adding 1(specifically, from 2N for an N-bit two's complement).

then I don't quite get it: wouldn't that mean if the number is 1000001, the number should be 128-1+1=128, when actually 10000001 is -127. And 10000000 (-128) would be 128-0+1=129?

Originally Posted by MK27

then I don't quite get it: wouldn't that mean if the number is 1000001, the number should be 128-1+1=128, when actually 10000001 is -127.

Since the most significant bit is negative, it should be: -128 + 0 + ... + 0 + 1 = -127.

Originally Posted by MK27

And 10000000 (-128) would be 128-0+1=129?

Why do you add the 1 when all the remaining bits are unset?

Originally Posted by laserlight

Why do you add the 1 when all the remaining bits are unset?

Is this definition wrong then?

subtracting the number from a large power of two and adding 1

Again, flipping the bits makes sense (I imagine this is what really happens) but this math formula is what looks unsinged to me. Altho either way I can't find -128 unless it is just treated as a special case, 10000000, which would have been 128 unsigned.

The only way I can see to describe this is to simply say 2N-value*sign, which will invariably work. I assume that the computer does not really use these silly formulas anyway -- Math people are ridiculous

Originally Posted by MK27

Is this definition wrong then?

Probably: take a look at the "Calculating two's complement" of the same article where it gives a correct formula without the addition of 1.

EDIT:

Originally Posted by MK27

The only way I can see to describe this is to simply say 2N-value*sign, which will invariably work.

There is no need to multiply by a sign value because the operation is performed with a fixed number of bits.

Originally Posted by laserlight

Probably: take a look at the "Calculating two's complement" of the same article where it gives a correct formula without the addition of 1.

Okay, cool, I thought that was my math ignorance at work.

Originally Posted by wikipedia

The value of a two's-complement binary number can be calculated by adding up the power-of-two weights of the "one" bits, but with a negative weight for the most significant (sign) bit

Makes sense and it is different than 2N-value+1.