What is a BitMask

[miteshpandey]

Can anybody explain what a Bit Mask is. I have seen its usage in various structure fields.

What are its advantages?

Any links relating to it will be helpful.

[gstercken]

Can anybody explain what a Bit Mask is. I have seen its usage in various structure fields.

Basically, a bit mask is a value which can be used to mask out certain bits of another value by applying bitwise AND, OR and XOR.

For example, let's assume that a 16 bit value is used to store the R,G,B components of a color like this:

The leftmost 5 bits for the red component, the next 6 bits for the green component, and the rightmost 5 bits for the blue component:

RRRR RGGG GGGB BBBB

In this example, the color value for yellow would be

1111 1111 1110 0000 (binary), or FFE0 (hex), or 65504 (decimal).

Now let's assume that you need a way to filter out the individual R,G,B components from a color: This can be done by AND-ing the following bitmasks with the color value:


Red: 1111 1000 0000 0000 (binary) = F800 (hex) = 63488 (dec)
Green: 0000 0111 1110 0000 (binary) = 07E0 (hex) = 2016 (dec)
Blue: 0000 0000 0001 1111 (binary) = 001F (hex) = 31 (dec)

As an example, consider the following code:

  const unsigned short redMask   = 0xF800;
  const unsigned short greenMask = 0x07E0;
  const unsigned short blueMask  = 0x001F;

  unsigned short lightGray = 0x7BEF;

  unsigned short redComponent   = (lightGray & redMask) >> 11;
  unsigned short greenComponent = (lightGray & greenMask) >> 5;
  unsigned short blueComponent =  (lightGray & blueMask);

(Note the right shift required to move the resulting bits to the rightmost position).

What are its advantages?

Well, the above is just one example - there are many others. Just think of things like window styles, which are also coded as single bits in a 32 bit value. Bit masks can be used to set or clear individual styles.

[miteshpandey]

Thank you very much gstercken.

I have another question for you.

How is the presence of bit value tested?

I have seen that inorder to test for a certain bit like selection it's used like this

if(dwStyle & BS_SELECTION == BS_SELECTION)

Please explain the mechanism. I have somewhere heard that this is done for compatiblity reason.

Any help will be appreciated much.

[Scrambler]

In expression if(A&B==B) there are 2 operations.
First is bitwise-and - &. Let for example A=01010101 (binary) and B=00111100. Bitwise-and result is a byte/word/dword (depending on operands) in which bit is 1 only if corresponding bits in both operands are 1. Otherwise bit is 0. So, A&B=00010100.
Second is equality operator. You need it only if both A and B contain more than 1 nonzero bit. In example above result is not equals to any of operands. But you need to know when all nonzero bits in B are also nonzero in A.
But if you know that there is only 1 nonzero bit in B you can write if(A&B) instead.

[miteshpandey]

Thank you scrambler. Now I understand much more.

But I am confused about one thing.

I have read somewhere that the left variable (dwStyle) could be omitted. But for future compatibility it is kept. But if we omit dwStyle everything becomes nonsense.

Perhaps I am mixing thing here.

[miteshpandey]

Thank you scrambler. Now I understand much more.

But I am confused about one thing.

I have read somewhere that the left variable (dwStyle) could be omitted. But for future compatibility it is kept. But if we omit dwStyle everything becomes nonsense.

Perhaps I am mixing thing here.

[gstercken]

[QUOTE=miteshpandey]I have read somewhere that the left variable (dwStyle) could be omitted. But for future compatibility it is kept. But if we omit dwStyle everything becomes nonsense.

Perhaps I am mixing thing here.

You must indeed be mixing things... What you have in mind is probably the CWnd member function ModifyStyle(): It doesn't require you to retrieve the current style first, you simply pass the combination of styles to be set or cleared, respectively.