Simple Macro Function Pointer Trick...

[Chris Thomasson]

I was wondering if the following technique will produce undefined behavior:
_______________
#include <cstdio>

#define CALL_MACRO_FUNCTION(func_ptr)func_ptr()

#define MY_MESSAGE() "Press <ENTER> to exit."

int main(void) {
puts(CALL_MACRO_FUNCTION(MY_MESSAGE));
getchar();
return 0;
}

_______________

I define 'MY_MESSAGE' as a macro function and only pass the name of it to
'CALL_MACRO_FUNCTION' which in turn uses the name to actually invoke it. The
name of a macro function is analogous to a function pointer. This is useful
when you need to have precise control over when a macro will actually
expand.

--
Chris M. Thomasson
http://appcore.home.comcast.net

[James Kanze]

On Sep 14, 4:16 pm, "Chris Thomasson" <cris...@comcast.net> wrote:
> I was wondering if the following technique will produce undefined behavior:
> _______________
> #include <cstdio>

> #define CALL_MACRO_FUNCTION(func_ptr)func_ptr()

> #define MY_MESSAGE() "Press <ENTER> to exit."

> int main(void) {
> puts(CALL_MACRO_FUNCTION(MY_MESSAGE));
> getchar();
> return 0;
> }
> _______________

Not related to your actual question, but the above shouldn't
compile with a conformant implementation of <cstdio>. puts and
getchar should be in std, and no where else. (I don't think any
of the implementations are conformant in this regard. But
what's the point of using <cstdio> instead of <stdio.h>, unless
you expect the functions to be in std::.)

> I define 'MY_MESSAGE' as a macro function and only pass the
> name of it to 'CALL_MACRO_FUNCTION' which in turn uses the
> name to actually invoke it. The name of a macro function is
> analogous to a function pointer. This is useful when you need
> to have precise control over when a macro will actually
> expand.

It should work. When expanding a macro, the preprocessor
rescans the substituted text for further expansions, and it only
recognizes a function style macro if the name is immediately
followed by an opening parentheses. (Thus, for example, in C,
getchar could be a function style macro. But in expressions
like:
int (* pf)() = &getchar ;
or
(getchar)() ;
the macro is not recognized, and the actual function which it
hides is used.)

--
James Kanze (GABI Software) email:james...@gmail.com
Conseils en informatique orientée objet/
Beratung in objektorientierter Datenverarbeitung
9 place Sémard, 78210 St.-Cyr-l'École, France, +33 (0)1 30 23 00 34

[Chris Thomasson]

"James Kanze" <james...@gmail.com> wrote in message
news:1189845690.6...@n39g2000hsh.googlegroups.com...

On Sep 14, 4:16 pm, "Chris Thomasson" <cris...@comcast.net> wrote:
>> I was wondering if the following technique will produce undefined
>> behavior:
>> _______________
>> #include <cstdio>

[...]
>> _______________

> Not related to your actual question, but the above shouldn't
> compile with a conformant implementation of <cstdio>.

[...]

DOH! Ah crap. Of course your correct. Here is some better code:

________________
#include <cstdio>

#define CALL_MACRO_FUNCTION(func_ptr)func_ptr()

#define MY_MESSAGE() "Press <ENTER> to exit."

int main(void) {
{
using namespace std;
puts(CALL_MACRO_FUNCTION(MY_MESSAGE));
getchar();
}

return 0;
}

________________

>> I define 'MY_MESSAGE' as a macro function and only pass the
>> name of it to 'CALL_MACRO_FUNCTION' which in turn uses the
>> name to actually invoke it. The name of a macro function is
>> analogous to a function pointer. This is useful when you need
>> to have precise control over when a macro will actually
>> expand.

> It should work.

Yeah. I agree. BTW, thank you for you response.