I understand it\'s possible to use bitmasks in enum values, but I don\'t know how to create it.
I have a simple enum :
enum State
{
minimizing = 0,
Use a different bit for every value in your enumeration, such as:
enum State
{
minimizing = 0x01, // 00000001
maximizing = 0x02, // 00000010
minimized = 0x04, // 00000100
maximized = 0x08 // 00001000
}:
Then, you can combine multiple values with bitwise or (minimizing | maximized
) and test for values with bitwise and (bool is_minimized = (flags & minimized);
).
I just tried this in VS2012, the optimizer seems to correctly combine bits without any need for assistance if you're using bitfields.
struct BITS { int x: 1; int y:1; };
then
BITS b;
b.x = b.y = 1;
Sets both bits with one instruction.
I am going to assume that myState
has the type of your enum State
.
The traditional use of enum
is to create the constant values that a variable of this type can take. You wish to set you variable myState
to a combination of the values defined in the enum
.
The enum
defines 1, 2, 4, and 8 as valid values, yet you want to be able to set the variable to 4 | 2 = 6. While C uses your implementation-defined int
type for all enum
, it is not the case in C++. myState = 6
is not valid in C++. Actually, myState = 4
is not either valid in C++, you need to cast explicitly or use one of the constant names of the enum
.
Although possible in C, it is not good practice to set myState
to a value that is not defined by its type (for example to 6).
In your case, a solution that seems consequent would be:
typedef enum {
OTHER,
MINIMIZED,
MAXIMIZED
} win_size_t;
typedef struct {
win_size_t current;
win_size_t next;
} state_t;
state_t myState;
That way, you can write to the fields current
and next
undependently.
If you still want to have bit fields, you can set the size of the elements of your struct in bits. It is kind of dangerous though, the implementation of bit fields depend on your compiler. I am not even sure if compilers would accept to have an enum type in a bit field (should be ok in C, since enum
s are int
).
typedef struct {
win_size_t current : 2; // not tested
win_size_t next : 2;
} state_t;
The previous given solutions are valid of course. My point is that if your variable myState
has your enum State
as type, it should only use the members of the enum
for its values, not a combination.
Maybe myState
has another type, what do I know.
If myState
is not of the enum State
type, then you may use the constants defined in your enum
in combination.
enum State {
MINIMIZING = (1u << 0),
MAXIMIZING = (1u << 1),
MINIMIZED = (1u << 2),
MAXIMIZED = (1u << 3),
};
unsigned int myState = 0;
myState |= MAXIMIZED; // sets that bit
myState &= ~MAXIMIZED; // resets that bit
This allows you to do two things in one assignment:
myState = MAXIMIZED | MINIMIZING;
But also things you are not likely to want:
myState = MAXIMIZED | MINIMIZED; // does that make sense?
You can get this effect by specifying all fields in the enum and increased in powers of two to get the bitmask effect. For example, in your case:
enum State
{
minimizing = 1,
maximizing = 2,
minimized = 4,
maximized = 8
};
So you can then have your combinations of (State.maximized | State.minimizing)
. However this won't apply the restriction of only being State.maximized
or State.minimized
. If you want to do that you could convert them to a single bit, but I imagine in this example you would want to be able to have a case where it is neither maximized nor minimized.