Re: Should this compile and link??

[Ross Boylan <RossBoylan@stanfordalumni.org>]

In your example, you have created an anonymous structure, and then
created a typedef for it.

Since it is anonymous, there is no way to refer to it, and thus your
definition of f1 doesn't get associated properly. I'm a little
suprised the compiler doesn't give you an error in c4e3_func.cc when
you refer to the class or structure TEST_TYPE.

Consider this to clarify things:

struct SILLY {
	void f1(void);
};
typedef SILLY TEST_TYPE;

If you want to define f1, you must say
SILLY::f1
not TEST_TYPE::f1.

However, in your example there is no name SILLY, so no way to refer to
the function.

Alternately, you could 
typedef struct {
	void f1(void) (puts("something");};
} TEST_TYPE;

and then you'd probably be OK.

I haven't communed with the standards statement, but I think this is
what's going on.


On Wed, Jan 16, 2002 at 02:27:29PM -0600, Bill Priest wrote:
> c4e3.cc
> ---------------
> #include "c4e3.h"
> 
> int main(void)
> {
> 	TEST_TYPE test1;
> 
> 	test1.f1();
> }
> ----------------
> c4e3_func.cc
> ----------------
> #include <cstdio>
> #include "c4e3.h"
> 
> void TEST_TYPE::f1(void)
> {
> 	puts("c4e3");
> }
> ------------------
> 
> c4e3.h
> ------------------
> #ifndef C4E3_H
> #define C4E3_H
> 
> typedef struct // TEST_TYPE
> {
> 	void f1(void);
> } TEST_TYPE;
> 
> #endif
> --------------------
> 
> w/ stock g++ 3.0.3 the following commands
> 
> g++    -c -o c4e3.o c4e3.cc
> g++    -c -o c4e3_func.o c4e3_func.cc
> g++ c4e3.o c4e3_func.o -o c4e3
> c4e3.o: In function `main':
> c4e3.o(.text+0x11): undefined reference to `TEST_TYPE::f1()'
> 
> If I uncomment TEST_TYPE in c4e3 everything compiles and links
> as it should.
> 
> If I change c4e3.cc to include "c4e3_func.cc" and change the command to
> g++ -o c4e3 c4e3.cc
> then it compiles and links even if the c4e3.h file is unchanged (ie TEST_TYPE is commented out).
> 
> This doesn't make a whole lot of sense to me; but I'm not a C++ expert or even novice.  
> Neither of my books (Thinking in C++ and C++ for C programmers) seems to cover this explicitly.
> 
> Is there something I'm missing here.  Scoping or namespace resolution was the only thing I could
> come up with but didn't see anything in either of the books.
> 
> Please set me straight,
> 
> Bill
> 
> _________________________________________________________
> Do You Yahoo!?
> Get your free @yahoo.com address at http://mail.yahoo.com
>