Timon Gehr: > Comparing signed/unsigned is perfectly reasonable. Right, but only if the numbers don't get implicit reinterpretations to other intervals, as C/C++/D do. Bye, bearophile