Help with bit-field semantics in C and C++
Gabriel Dos Reis
gdr@integrable-solutions.net
Fri Aug 27 03:56:00 GMT 2004
Joe Buck <Joe.Buck@synopsys.COM> writes:
| On Wed, Aug 25, 2004 at 12:26:24PM +0100, Dave Korn wrote:
| > It seems to me that "unspecified" and "undefined" are repeatedly being
| > confused in this discussion. They are both clearly defined technical terms
| > in the standard; they have different meanings; everyone participating in
| > this thread should make very sure they know which one they're referring to
| > at any time.
| >
| > "undefined" is the one that covers rm -rf'ing your hard drive.
| >
| > "unspecified" is what the standard says that the result of the truncating
| > cast we've been discussing is. That means (IIUIC) that the result has to be
| > one of the enumeration values, but that the standard does not specify which
| > _particular_ one.
|
| If you are correct, then it appears that every C++ compiler I know of gets
| this wrong. That is, it seems that every C++ compiler happily allows
| out-of-range integers to be assigned to enums, and does so in a way that
| the effect is observable.
|
| If your interpretation is correct, then I can write the following test for
| a conforming compiler; g++ fails this test. I'm deliberately putting
| is_zero and is_one in separate functions, to catch a compiler that might
| allow storing the 3, but that would later assume that global must be zero
| or one and optimize the "if" away.
There is a very simple principle : if you initialize an integer
variable, or bit-field of such type, its value is always in range --
even if the value you used to initialize the variable was out-of-range
(therefore generated an implementation-defined value of the
destination type). Enumerations are no exception.
Do you expect any of the following assertions to fail?
#include <assert.h>
#include <limits.h>
// assume 'signed char' has limits -128 and 127
int main()
{
signed char i = 200;
assert(i >= -128);
assert(i <= 127);
assert(i != 200);
}
-- Gaby
More information about the Gcc
mailing list