variadic macros and and empty replacement for __VA_ARGS__

[Urs Thuermann]

I want to write a macro that produces debug output and has a variable
number of arguments, so that I can use like this:

int i;
char *s;

DBG("simple message\n");
DBG("message with an int argument, value is %d\n", i);
DBG("two arguments: int %d, strings %s\n", i, s);

The output should be a fixed prefix, say "DEBUG ", followed by the
function name the macros is used in, and then by the string passed to
the macro, including further arguments.

The best solution I have found so far, is

#define DBG(fmt, ...) printf("DEBUG %s:" fmt, __func__, __VA_ARGS__)

This works as intended as long as there is at least one argument
following the format strings, but it expands to
printf("DEBUG %s:" fmt, __func__,) when there is no argument and that
causes a syntax error.

The draft of C99 I have here, disallows to give no arguments for the
ellipsis ... as stated in 6.10.3:

[#4] If the identifier-list in the macro definition does not end
with an ellipsis, the number of arguments, including those
arguments consisting of no preprocessing tokens, in an invocation
of a function-like macro shall agree with the number of parameters
in the macro definition. Otherwise, there shall be more arguments
in the invocation than there are parameters in the macro
definition (excluding the ...). There shall exist a )
preprocessing token that terminates the invocation.
Is there a better way to write the DBG() macro to avoid this?

urs

[Harald van Dijk]

Michael Mair wrote:

raxip schrieb:

This works as intended as long as there is at least one argument
following the format strings, but it expands to
printf("DEB UG %s:" fmt, __func__,) when there is no argument and that
causes a syntax error.

#define macroFunc(a,b,. ..) ( Func(a, b, ##__VA_ARGS__) )

This is "gcc-C".
It is not valid C99, though.

I believe it can be valid C99 if the result of macroFunc(...) is
stringized, and __VA_ARGS__ is empty. (Of course, in that case, it's
useless and doesn't answer the question.)

As for the original question, I had taken a look at a non-GCC way of
writing ,##__VA_ARGS__ earlier. I don't recommend anyone actually use
this for hopefully obvious reasons, but in case anyone cares, here it
is. Please let me know if there are bugs in it (there may well be).

#include <stdio.h>

#define CONCAT(x, y) CONCAT_(x, y)
#define CONCAT_(x, y) x##y

#define ID(...) __VA_ARGS__

#define IFMULTIARG(if,t hen,else) \
CONCAT(IFMULTIA RG_, IFMULTIARG_(if, \
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, \
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, \
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, \
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, \
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, \
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, \
1, 1, 0, ))(then,else)
#define IFMULTIARG_(_0, _1, _2, _3, _4, _5, _6, _7, _8, _9, \
_10, _11, _12, _13, _14, _15, _16, _17, _18, _19, \
_20, _21, _22, _23, _24, _25, _26, _27, _28, _29, \
_30, _31, _32, _33, _34, _35, _36, _37, _38, _39, \
_40, _41, _42, _43, _44, _45, _46, _47, _48, _49, \
_50, _51, _52, _53, _54, _55, _56, _57, _58, _59, \
_60, _61, _62, _63, ...) _63
#define IFMULTIARG_0(th en, else) else
#define IFMULTIARG_1(th en, else) then

#define PROVIDE_SECOND_ ARGUMENT(x, ...) \
CONCAT(IFMULTIA RG(ID(__VA_ARGS __), INSERT_, ADD_), \
SECOND_ARGUMENT ) (x, __VA_ARGS__)
#define ADD_SECOND_ARGU MENT(x, y) y, x
#define INSERT_SECOND_A RGUMENT(x, y, ...) y, x, __VA_ARGS__

#define DEBUG(...) printf("DEBUG %s: " \
PROVIDE_SECOND_ ARGUMENT(__func __, __VA_ARGS__))

int main() {
DEBUG("1\n");
DEBUG("%d\n", 2);
DEBUG("%d %d\n", 1, 2);
}

[Peter Shaggy Haywood]

Groovy hepcat Urs Thuermann was jivin' on 27 Mar 2006 15:34:46 +0200
in comp.lang.c.
variadic macros and and empty replacement for __VA_ARGS__'s a cool
scene! Dig it!

I want to write a macro that produces debug output and has a variable
number of arguments, so that I can use like this:

int i;
char *s;

DBG("simple message\n");
DBG("message with an int argument, value is %d\n", i);
DBG("two arguments: int %d, strings %s\n", i, s);

The output should be a fixed prefix, say "DEBUG ", followed by the
function name the macros is used in, and then by the string passed to
the macro, including further arguments.

I think this should work:

#define DBG(...) printf("DEBUG $s: ", __func__), printf(__VA_ARG S__)

--

Dig the even newer still, yet more improved, sig!

http://alphalink.com.au/~phaywood/
"Ain't I'm a dog?" - Ronny Self, Ain't I'm a Dog, written by G. Sherry & W. Walker.
I know it's not "technicall y correct" English; but since when was rock & roll "technicall y correct"?

[Urs Thuermann]

> I think this should work:

#define DBG(...) printf("DEBUG $s: ", __func__), printf(__VA_ARG S__)

This is the first answer in this thread, which seems to understand my
post and solves the problem of inserting the __func__ between the fmt
and __VA_ARGS__ without having the ,) when __VA_ARGS__ is empty.

I wanted to avoid calling printf() twice, because I need this in the
linux kernel (with printk instead of printf) and I'd prefer it to be
atomic. Also, I can't make a function debug(char *name, char *fmt, ...)
and call printk() from there, because the linux kernel doesn't have
vprintk().

But I will follow your suggesstion and will do it like follows:

#ifdef DEBUG
int debug = 0;
#define DBG(...) (debug & 1 ? (printk(KERN_DE BUG "%s: ", __func__), \
printk(__VA_ARG S__)) : 0)
#else
#define DBG(...)
#endif

where KERN_DEBUG is a string constant defined in the linux kernel
sources.

urs

[pemo]

Couldn't you get rid of the int debug = 0 using this?:

#define DEBUG 1

#ifdef DEBUG
#define DBG(...) (DEBUG ? (printk(KERN_DE BUG "%s: ", __func__), \
printk(__VA_ARG S__)) : 0)
#else
#define DBG(...)
#endif

[pemo]

pemo wrote:

Urs Thuermann wrote:

I think this should work:

#define DBG(...) printf("DEBUG $s: ", __func__), printf(__VA_ARG S__)

This is the first answer in this thread, which seems to understand my
post and solves the problem of inserting the __func__ between the fmt
and __VA_ARGS__ without having the ,) when __VA_ARGS__ is empty.

I wanted to avoid calling printf() twice, because I need this in the
linux kernel (with printk instead of printf) and I'd prefer it to be
atomic. Also, I can't make a function debug(char *name, char *fmt, ...)
and call printk() from there, because the linux kernel doesn't have
vprintk().

But I will follow your suggesstion and will do it like follows:

#ifdef DEBUG
int debug = 0;
#define DBG(...) (debug & 1 ? (printk(KERN_DE BUG "%s: ", __func__), \
printk(__VA_ARG S__)) : 0)
#else
#define DBG(...)
#endif

where KERN_DEBUG is a string constant defined in the linux kernel
sources.

Couldn't you get rid of the int debug = 0 using this?:

#define DEBUG 1

#ifdef DEBUG
#define DBG(...) (DEBUG ? (printk(KERN_DE BUG "%s: ", __func__), \
printk(__VA_ARG S__)) : 0)
#else
#define DBG(...)
#endif

[Urs Thuermann]

"pemo" <us***********@ gmail.com> writes:

Couldn't you get rid of the int debug = 0 using this?:

#define DEBUG 1

#ifdef DEBUG
#define DBG(...) (DEBUG ? (printk(KERN_DE BUG "%s: ", __func__), \
printk(__VA_ARG S__)) : 0)
#else
#define DBG(...)
#endif

No, your code doesn't make sense.

In my code, debug is initialized to zero (i.e. no debug output) but
can be set at module load time. There are other debug macros, that
test for debug&2 and debug&4.

urs