Thread: Any suggestions for finding the largest possible value of INT?

> Where is the ~0UL coming from, is it part ot stdbool or something else.

It has nothing to do with <stdbool.h>. The UL is a suffix that makes the constant into an unsigned long. An integral constant without a suffix such as 57 is a (signed) int by default, while 57U is an unsigned int, and 57UL is an unsigned long. A floating-point constant such as 1.0 with a decimal point but no suffix is double by default, while 1.0F is a float, and 1.0L is a long double.

Edit: This is not a complete list, and you can also use the lower-case version of these letters, though the upper-case version is usually easier to read properly, for example 57L is a long int, and so is 57l, but the latter looks like 571, so you should use L.

I have not seen the ~0 or ~value before, where is that coming from? I have seen ~1 used to set boolean values before. I see how it works and it seems straigh forward but I would like to know more about what it is. What would I call setting unsigned int x = ~0u; ?

I looked though my Kernighan and Ritchie 2nd edition and I don't see it.

Originally Posted by robatino

See section 2.9 (Bitwise Operators) on page 48.

Oh, so it's one's complement? I'll have to re-read that section. I really need to get a copy of "The C Answer Book"

Thanks again, I really appreciate the help.

This is such a great book, I wish every language had something as good.

It is as simple as this:
0 always has all bits as zero
~ always negates all bits
therefore with ~0, all bits are ones.

Once you've got a number where all bits are ones, how the processor interprets that when you do things with it is a different story, and that's where number representations come into play.

If the 0 is a signed type, then in two's complement, all bits are zero, but in ones' complement and sign-magnitude, it has two representations, plus zero and minus zero, so this is no longer true. So to be portable, one has to make the 0 unsigned, either with an unsigned suffix, or an explicit cast, before applying ~.

Edit: Actually, I'm not 100&#37; sure about this - I don't know if the standard says anything about whether the bit representation of plain signed int 0 is all zeroes or not. Anyone know?

Originally Posted by robatino

plus zero and minus zero

oh god, don't tell me 0 can be signed as well? I miss those days in primary school when all the number of the world could be counted off your fingers...

QuantumPete

Is there any machines currently in existance (aside from obscure Cray's or IBM and Data General machines from 1970's - and, ok, some graphics processors may not use this either, but I'm thinking of machines where you can run C or C++-code) where -1 is NOT all ones?

--
Mats