Hi- Having recently discovered #defines, I'm wondering why anyone would use constant variables instead???? What do you use?? Am I missing something???? Is there any difference- speed wise etc? I would imagine #defines are better because they become part of the compiled program memory versus constant variables that have to be accessed separately?? Thanx Edited by - simulator on August 2, 2001 6:26:01 PM Edited by - simulator on August 2, 2001 6:30:07 PM

defines get put in before your program compiles.

You can even things like this:

#define printg printf 

Although I don''t know why you''d want to!

Hi & thanx,

Yes I know about how defines work...
Perhaps constant variables would be those defined at run time eg:


Const long lPitch;

(ie: where lPitch must be determined only once the program is running) (not sure this is even permitted.... must try it)


thanx


--- yes it works- so that must be why???

But still, even for plain, known-ahead-of-time constants- what do most people use: #define?????

Edited by - simulator on August 2, 2001 6:42:59 PM

Edited by - simulator on August 2, 2001 6:49:10 PM

Constant values are placed inline in the same way that #defines are, so there''s no speed difference. I use const for constants, and #define for other things, but both are quite acceptable. Do what looks nice to you.

Hmm- OK good points, thank you

Normie - you''re actually mistaken there - #defines are, as you said, substitutions. So:

#define NUMBER "AKLSJFD"
array[NUMBER] = 113;

would become:

array["AKLSJFD"] = 113;

...which raises a compiler error. However:

#define INCREMENT(a, b) a += b

If you pass a char* and an float to that function, you''ll get an error which doesn''t seem to make sense (Illegal use of floating point). And that''s the only problem I''ve ever seen coming from lack of type safety. It''s not actually as bad as some people make out.

#define''s are also nasty in that if you do something like

#define IncCheck(a,b) (a++ < b++) ? a++ : b++

and then
IncCheck(i,j)
both i & j will be incremented at least once,
because the direct substitution expands to
(i++ < j++) ? i++ : j++

The const keyword was introduced to allow the compiler to type check arguments (can''t do that with macros) and the inline keyword was used to fix the expansion problem that macros suffer. The only time I''d recommend using macros is conditional code compilation, ie
#if defined _DEBUG
printf("This is a debug version");
#endif

Brad

brad_beveridge: If you wrote that as an inline function, using the same code, you''d get the exact same logic error. I think what you meant was more like this:

#define Check(a, b) ((a > b) ? a : b)
Check(a++, b++);

Which would of course cause an error. But show me someone who would write something like that and not see the problem within a few minutes, and I''ll show you someone who needs to learn the basics of c.

There are still advantages that macros have over inline functions. In some cases, it''s just that it requires less code than the inline function, and in some cases, nothing can go wrong:

#define CREATEWINDOW(Title) CreateWindowEx(0, "randomclassname", Title, WS_VISIBLE, 0, 0, 640, 480, NULL, NULL, GetModuleHandle(NULL), NULL)

And although the following isn''t the best example, there are some things that macros can do that inline functions cannot:

// Declare and clear a DirectX structure:
#define DECLARE_STRUCT(type, var) type var; ZeroMemory(&var, sizeof(var)); var.dwSize = sizeof(var)


In summary, use macros where appropriate, and use inline functions where they''re appropriate. A macro should use each parameter exactly once (to avoid the Check(a++, b++) problem), and should be commented to describe the parameters if the parameters are not obvious.