C和C ++中的宏定义

Question

I am a new programmer in C/C++ having programmed in Java for quite a while. I am currently understanding some C code. Here I am seeing some macro definitions like:

/* Flags for ds_flags */ 
#define DSF_OVER        (1<<0)  
#define DSF_DISPLAY     (1<<1)    
#define DSF_CALLFLOW    (1<<2) 

I am not able to understand why do we have to define these macros in such a manner. What is the advantage gained in these rather than in defining like:

#define DSF_OVER        0
#define DSF_DISPLAY     1    
#define DSF_CALLFLOW    2

Answer 1

When seeing something like

#define DSF_FOO 0x800
#define DSF_BAR 2048

not many people are able to quickly see that only one bit is set, and which it is.

#define DSF_FOO (1<<11)

however makes this very clear.

Answer 2

The only potential advantage is that it's easier to see that the code correctly defines each constant with one distinct bit set.

Most people will see that at a glance anyway if you just write 1, 2, 4 instead of 1<<0 , 1<<1 , 1<<2 , so perhaps it's difficult to see that advantage in this example. Once you get to 1<<15 , some people would miss a typo like 32748.

Answer 3

These are bit values, eg 1, 2, 4. 8.

Bit 0 = (1 << 0) = 1
Bit 1 = (1 << 1) = 2
Bit 2 = (1 << 2) = 4
Bit 3 = (1 << 3) = 8
...etc...

It's a more convenient and robust way of defining them than using explicit values, especially as the bit indices get larger (eg (1<<15) is much easier to understand and more intuitive than 32768 or 0x8000 , in that it obviously means "bit 15" rather than some possibly arbitrary number).

Answer 4

Using (1<<x) in a define makes it clear that the value is a single bit and not a number.

For the compiler it makes no difference because (1<<2) is computed at compile time and not at runtime. Showing clearly that they are single-bit values is instead useful for whoever reads the code because for example they could be multiple values that can be combined or that a single variable can be used to store multiple flags:

// Multiple options are combined with bitwise-or
show_message(DSF_CAPTION|DSF_ALERT, "Hey...");
...

// Checking is made using bitwise-and, not equality
if (status & DSF_RUNNING)
  ...

Also requiring specific bits is sometimes needed when dealing with hardware (eg on a certain I/O port may be you need to specify the fifth bit because that's how the hardware is wired, and (1<<4) is more readable than 16 for that).

Answer 5

It's for clarity really - it shows that the different definitions are bit masks that will typically be used together to filter values from an input. I don't know if you already know what a bit mask is - here's a link: http://www.vipan.com/htdocs/bitwisehelp.html

Answer 6

Regardless of whether you prefer (1 << n) or not, everyone should be aware of what hexadecimal numbers signify. Whenever you encounter hex in source code it always means that som bit- or byte manipulations are taking place. You don't use hex notation for any other purpose. Therefore it is wise to write masks, numeric bit manipulation literals etc as hex.

Answer 7

Some times the position of the bits represent some bit operations like in your case:

#define DSF_OVER        (1<<0)  is 1
#define DSF_DISPLAY     (1<<1)  is 2  
#define DSF_CALLFLOW    (1<<2)  is 4 (This is not 3)

If you were to add a new item later, you will do it as

#define DSF_OVER       1
#define DSF_DISPLAY    2  
#define DSF_CALLFLOW   4
#define DSF_CALLNEW    5.

You cant have a value 5 here since it is two bits enabled (101 in binary). To avoid such potential errors it is always safe to use macros using >> . It can't produce a value of 5 (101) by error in when the next bit should be 1000 in binary.

It is all about programming convenience to produce error free code.