Standard way to determine max range of size_t?

[Michael Ekstrand]

I'm looking for a standard (or standard-ish) way to determine the maximum
value representable by a size_t. I can't seem to find anything
officially standard - cstddef doesn't seem to define such a thing, nor
does climits.

Applying grep to my /usr/include reveals an stdint.h header which defines
a SIZE_MAX, some further research indicates that this is a C99 standard
header but not standard for either C89 or standard C++. My primary target
compiler is G++, so that would be acceptable, but I would rather use a
standard means if one exists.

So, my question is: Is there a mechanism in standard C++ to determine the
maximum value of a size_t? If there is not a define or some other
declaration of this limit, is it reliable to assume that ((size_t) -1)
(or some more appropriate style of cast) is the maximum value which can be
stored in a size_t?

Thank you,
- Michael

[mlimber]

On Mar 24, 12:18 pm, Michael Ekstrand <use...@elehack .netwrote:

I'm looking for a standard (or standard-ish) way to determine the maximum
value representable by a size_t. I can't seem to find anything
officially standard - cstddef doesn't seem to define such a thing, nor
does climits.

Applying grep to my /usr/include reveals an stdint.h header which defines
a SIZE_MAX, some further research indicates that this is a C99 standard
header but not standard for either C89 or standard C++. My primary target
compiler is G++, so that would be acceptable, but I would rather use a
standard means if one exists.

So, my question is: Is there a mechanism in standard C++ to determine the
maximum value of a size_t? If there is not a define or some other
declaration of this limit, is it reliable to assume that ((size_t) -1)
(or some more appropriate style of cast) is the maximum value which can be
stored in a size_t?

Thank you,
- Michael

Yes. Include <limits>, then do:

std::numeric_li mits<std::size_ t>::max()

Cheers! --M

[Michael Ekstrand]

On Sat, 24 Mar 2007 09:23:56 -0700, mlimber wrote:

>So, my question is: Is there a mechanism in standard C++ to determine the
maximum value of a size_t? If there is not a define or some other
declaration of this limit, is it reliable to assume that ((size_t) -1)
(or some more appropriate style of cast) is the maximum value which can be
stored in a size_t?

Yes. Include <limits>, then do:

std::numeric_li mits<std::size_ t>::max()

Cheers! --M

Thank you. This is exactly what I needed.

- Michael

[Andrew Koenig]

"Michael Ekstrand" <us****@elehack .netwrote in message
news:eu******** **@aioe.org...

I'm looking for a standard (or standard-ish) way to determine the maximum
value representable by a size_t.

(size_t)-1 should do it. The point is that size_t is guaranteed to be an
unsigned type, and a signed integer is converted to unsigned by taking it
modulo 2^n, where n is the number of bits in the unsigned type.