SIZE_MAX under c89

[aegis]

I've looked through the c89 standard and I do not see SIZE_MAX
anywhere. So this was a new addition to c99? Isn't it legal, since c89
does not define SIZE_MAX, to just use #define SIZE_MAX (size_t)-1 ?

 

[goose]

I've looked through the c89 standard and I do not see SIZE_MAX
anywhere. So this was a new addition to c99? Isn't it legal, since c89
does not define SIZE_MAX, to just use #define SIZE_MAX (size_t)-1 ?

do you mean
#define SIZE_MAX (sizeof (size_t) -1)

?

 

[Coos Haak]

Op 5 Jun 2005 15:05:25 -0700 schreef aegis:

I've looked through the c89 standard and I do not see SIZE_MAX
anywhere. So this was a new addition to c99? Isn't it legal, since c89
does not define SIZE_MAX, to just use #define SIZE_MAX (size_t)-1 ?

I don't know the answer to your first question, I don't have c89.
The #define seems wrong, SIZE_MAX is a constant, size_t is a type.
Perhaps #define SIZE_MAX UINT_MAX will do.

 

[Keith Thompson]

do you mean
#define SIZE_MAX (sizeof (size_t) -1)

Um, no. If size_t is 4 bytes, that defines SIZE_MAX as 3.

 

[Keith Thompson]

Op 5 Jun 2005 15:05:25 -0700 schreef aegis:


I don't know the answer to your first question, I don't have c89.
The #define seems wrong, SIZE_MAX is a constant, size_t is a type.
Perhaps #define SIZE_MAX UINT_MAX will do.

No, #define SIZE_MAX UINT_MAX won't do, unless size_t happens to have
the same range as unsigned int.

I believe SIZE_MAX is new in C99.

The suggested definition
#define SIZE_MAX (size_t)-1
defines SIZE_MAX as the value -1, converted to type size_t. I would
put parentheses around the definition:
#define SIZE_MAX ((size_t)-1)
It should give you the correct value on most or all systems. (There
may be some issues if size_t has padding bits; I'm too lazy to track
down the details.)

 

[goose]

Um, no. If size_t is 4 bytes, that defines SIZE_MAX as 3.

Gah!!!

Disregard my post, posting at an unseemly time of night!!!

 

[pete]

I would
put parentheses around the definition:
#define SIZE_MAX ((size_t)-1)

In case of maybe C99 compiler:

#ifndef SIZE_MAX
#define SIZE_MAX ((size_t)-1)
#endif

It should give you the correct value on most or all systems. (There
may be some issues if size_t has padding bits; I'm too lazy to track
down the details.)

I don't know what issues you mean.
Do you know the rules for converting negative integer values
to unsigned type values?

 

[bjrnove]

In case of maybe C99 compiler:

#ifndef SIZE_MAX
#define SIZE_MAX ((size_t)-1)
#endif


I don't know what issues you mean.
Do you know the rules for converting negative integer values
to unsigned type values?

This asumes that size_t is unsigned. Will size_t always be unsigned? If
not the size will actually be -1 and not the capacity of size_t.

 

[Martin Ambuhl]

This asumes that size_t is unsigned. Will size_t always be unsigned? If
not the size will actually be -1 and not the capacity of size_t.

Yes, size_t is unsigned by definition. From the standard:

7.17 Common definitions <stddef.h>
[...]
size_t

which is the unsigned integer type of the result of the
sizeof operator;

 

[CBFalconer]

.... snip ...

This asumes that size_t is unsigned. Will size_t always be unsigned?
If not the size will actually be -1 and not the capacity of size_t.

Yes, because the standard says it is unsigned.

 

[Keith Thompson]

In case of maybe C99 compiler:

#ifndef SIZE_MAX
#define SIZE_MAX ((size_t)-1)
#endif

That's good.

I don't know what issues you mean.
Do you know the rules for converting negative integer values
to unsigned type values?

Usually I do, but they had slipped my mind when I wrote the above, and
I didn't have my copy of the standard handy. You're right,
((size_t)-1) is guaranteed to be the maximum value that can be
represented in type size_t; padding bits and other representation
considerations are irrelevant. (The same thing works for any unsigned
type.)

 

[Eric Sosman]

I've looked through the c89 standard and I do not see SIZE_MAX
anywhere. So this was a new addition to c99? Isn't it legal, since c89
does not define SIZE_MAX, to just use #define SIZE_MAX (size_t)-1 ?

`SIZE_MAX' and `(size_t)-1' produce the same value, but
in different forms. `SIZE_MAX' is in a form suitable for
use in #if directives, but the preprocessor cannot evaluate
`(size_t)-1'. (The preprocessor cannot evaluate casts,
because the preprocessor operates at a stage of compilation
before types themselves have come into existence.)