question on #define

h/w · 12-03-2003, 12:49 PM

can we do a #define at any point in the code?

say i perform some task, and get a result xyz.

can i do a '#define RESULT xyx" at some point in the code?

thanks.

MartinN · 12-03-2003, 12:58 PM

A define is expanded by the C pre-processor. What it does is that it replaces, textually, the first argument of the define with the second before passing the code on to the compiler.

Because of that, your question doesn't make sense! If the "answer" is not known at compile time, then (obviously) the pre-processor can't make the replacing.

Regards
Martin

h/w · 12-03-2003, 01:06 PM

right! thanks!

jim mcnamara · 12-03-2003, 02:04 PM

#define will expand macros which can do things:

Code:

#define zout(z) memset(&z,0x00,sizeof(z))

.........
zout(mystring);
will expand to:
memset(&mystring,0x00,sizeof(mystring));

You also #define multiline macros which are really more like a way to inline a function:

Code:

// this function does a complex square calc 
#define CMPLXsqr_old(out)	\     
   (out).y = (old.x+old.x) * old.y;\  
   (out).x = tempsqrx - tempsqry

MartinN · 12-03-2003, 03:12 PM

Now I must raise a finger of warning

about the use of pre-processor macros.

Look at this seemingly innocent macro:

Code:

#define MAX(x, y)   ((x)>(y)?(x): (y))

Now consider what happens if we call this with

Code:

MAX(++j, ++k);

That looks pretty straight forward and correct, right? Wrong! Well what happens is that whichever variable is bigger gets increased twice! Why? Because the macro has been textually expanded to

Code:

++j > ++k ? ++j : ++k;

This is why inline is a good keyword in C++.

Regards
Martin

h/w · 12-03-2003, 04:40 PM

right. thanks.

jim mcnamara · 12-03-2003, 04:42 PM

But is not available for a lot of plain C implementations.

h/w · 12-03-2003, 05:14 PM

u mean, inline isnt ?