Dear Richard and John, With your help I solved my initial problem and I think I have a rough understanding what happens. However, I don't know really why it happens... You write: "An int is signed only if it's smaller than INT_MAX in magnitude. Otherwise it's unsigned (hence #defining INT_MIN as (-INT_MAX - 1)" Is this defined somewhere in the C++ standard (where?) or is this gcc specific? I was told that e. g. in MS Visual C++ 0xffffffff is interpreted as -1. And what exactly happens, when the numbers in an enum are too large to fit into an signed int? The compiler tries an unsigned int, then an long int then unsigned long and so on? Is this defined in the C++ standard or gcc specific? And is there a possibility to find out whether enum elements have signed or unsigned type? Thank you very much again, Hajo Am Donnerstag, den 03.12.2009, 15:25 +0000 schrieb Richard Earnshaw: > On Thu, 2009-12-03 at 14:40 +0000, Richard Earnshaw wrote: > > > > 0xFFFFFFFF (which is an unsigned int) > > > > wrong. That's an int with a bit pattern that's all 1's. On a two's > > complement machine with 32-bit ints that's equivalent to -1. > > Forget that. I'm talking nonsense. An int is signed only if it's > smaller than INT_MAX in magnitude. Otherwise it's unsigned (hence > #defining INT_MIN as (-INT_MAX - 1) > > R.
