Prior to the 1999 version of C, the result of division was considered compiler-dependent. This left the option to the compiler vendor to decide whether its customers would prefer fast division or consistent division. Division by a compile-time-known value of 2 could be done with an arithmetic right shift of one bit. On a 2s-complement machine, this would correspond to floor behavior; on a 1s complement machine it would correspond to truncation toward 0. Such a bit shift was much faster than actual division. If the divisor was not known to be 2 at compile time, the compiler would generate code to perform an actual division; if the divisor happened to be 2 for a particular division, the result would be truncated to 0. Thus, on a 2s-complement machine, one compiler would be allowed for dividing −7 by 2 to return −4 in some cases and −3 in other cases within one program.
Starting with 1999, C required compilers to perform integer division as truncation toward 0 in order to achieve more predictable behavior.
Another issue is that if one writes code with −7/2, it is treated as −(7/2) which would nearly always be evaluated as −3 regardless of compiler and hardware behavior. That is because in C, − is always a unary operator; there is no such thing as negative integer literals in C. People often forget this detail and are surprised.
Comment
in c language the division of -3/2 is -2 but compiler is giving -1 why?
Replies
Prior to the 1999 version of C, the result of division was considered compiler-dependent. This left the option to the compiler vendor to decide whether its customers would prefer fast division or consistent division. Division by a compile-time-known value of 2 could be done with an arithmetic right shift of one bit. On a 2s-complement machine, this would correspond to floor behavior; on a 1s complement machine it would correspond to truncation toward 0. Such a bit shift was much faster than actual division. If the divisor was not known to be 2 at compile time, the compiler would generate code to perform an actual division; if the divisor happened to be 2 for a particular division, the result would be truncated to 0. Thus, on a 2s-complement machine, one compiler would be allowed for dividing −7 by 2 to return −4 in some cases and −3 in other cases within one program.
Starting with 1999, C required compilers to perform integer division as truncation toward 0 in order to achieve more predictable behavior.
Another issue is that if one writes code with −7/2, it is treated as −(7/2) which would nearly always be evaluated as −3 regardless of compiler and hardware behavior. That is because in C, − is always a unary operator; there is no such thing as negative integer literals in C. People often forget this detail and are surprised.