Proper way to turn off debug macros at compile time.

[chris.fairles]

So I took a look at assert.h because I knew it uses -DNDEBUG to turn
off its assert statements but it does some funky stuff that I don't
quite understand. I made my own def's something like:

#ifdef DEBUG_LEVEL_1
#define iprint(expr) printf(#expr " = %d\n", expr)
#else
#define iprint(expr)
#endif

#ifdef DEBUG_LEVEL_2
#define
#define pprint(expr) printf(#expr " = %p\n", (void *)(expr))
#else
#define pprint(expr)
#endif

But it was suggested to me that I use some do { } while(0) method or
((void) 0) instead of just leaving the define's blank. Whats the
general opinion on this? (I looked in c-faq without much luck, maybe a
good q to put in less this is implementation specific).

 

[Eric Sosman]

So I took a look at assert.h because I knew it uses -DNDEBUG to turn
off its assert statements but it does some funky stuff that I don't
quite understand. I made my own def's something like:

#ifdef DEBUG_LEVEL_1
#define iprint(expr) printf(#expr " = %d\n", expr)
#else
#define iprint(expr)
#endif

#ifdef DEBUG_LEVEL_2
#define
#define pprint(expr) printf(#expr " = %p\n", (void *)(expr))
#else
#define pprint(expr)
#endif

But it was suggested to me that I use some do { } while(0) method or
((void) 0) instead of just leaving the define's blank. Whats the
general opinion on this? (I looked in c-faq without much luck, maybe a
good q to put in less this is implementation specific).

(The line immediately after the second #ifdef is
kinda funky ...)

The only reason I can think of to insert "code that
does nothing" instead of "nothing" is that the former might
silence a compiler warning in some situations. For example,
with DEBUG_LEVEL_1 undefined in a construct like

if (whatever > 0)
do_something();
else
iprint(whatever);

.... at least one widely-used compiler will warn you that
the `else' is empty. (Compilers can warn about whatever
they want, even perfectly legal C.) If the iprint() macro
expanded to `(void)0' or some other kind of no-op, such a
warning might not appear.

On the other hand, some compilers that would otherwise
have been silent might then start whining about "statement
with no effect" or something of the kind. You cannot win
every time!

My own style -- not "better" or "worse" than yours --
is to communicate my intent to the person who will someday
get a flurry of compiler warnings and come browsing through
the code to see what all the fuss is about:

#ifdef DEBUG_LEVEL_1
#define iprint(x) printf(...etc...)
#else
#define iprint(x) /* nil */
#endif

 

[Guest]

So I took a look at assert.h because I knew it uses -DNDEBUG to turn
off its assert statements but it does some funky stuff that I don't
quite understand. I made my own def's something like:

#ifdef DEBUG_LEVEL_1
#define iprint(expr) printf(#expr " = %d\n", expr)
#else
#define iprint(expr)
#endif

#ifdef DEBUG_LEVEL_2
#define
#define pprint(expr) printf(#expr " = %p\n", (void *)(expr))
#else
#define pprint(expr)
#endif

But it was suggested to me that I use some do { } while(0) method or
((void) 0) instead of just leaving the define's blank. Whats the
general opinion on this? (I looked in c-faq without much luck, maybe a
good q to put in less this is implementation specific).

One thing you should consider in addition to what is already said, is
that ((void) 0) allows you to write

pprint(p1), pprint(p2), pprint(p3);

It is up to you to decide whether you want to allow that.