🚀 go-pugleaf

In article <1992Sep12.160053.28778@isy.liu.se> svan@isy.liu.se writes:
>> What is wrong with the
>> following constructions and how can I do it in g++?
>>
>> ----------
>>
>> class a {
>> public: 
>>	int x();
>> };
>>
>> int a::x() { return 1; }
>>
>> main() {
>>	printf("%x\n", (int(*))a::x);
>> }
>
>You're trying to cast a pointer to member to a pointer to object
>type. That can't be done according to the standards
>documentation I've read over C++ (Annotated C++ Ref. Man. by
>Ellis & Stroustrup). 

To a member function, not just a member.  It can be done because I
encountered it while trying to port some code from AT&T C++ to g++.
It just doesn't seem to be possible in g++.

>Pointer to members may even not be explicitly converted to void*!
>
>The reason for this is that pointers to members may be
>a structure containing both addresses and offsets. There is
>no point in allowing a cast to any ordinary pointer.
>

There is if you want to choose different functions at a certain point
in a program depending on current circumstances.  Sorting elements by
different criteria, for example.


The workaround I've discovered is as follows :

typedef int (a::*afp)( void );   // can not do it without typedef and
				 // using intermediate variables
class a {
public:
  int a(void);
  int b(void);
  int c(void);
};

int a::a(void) { return 1; }
int a::b(void) { return 2; }
int a::c(void) { return 3; }

main() {
  afp y = a::x;

  printf("%x\n", y);  //  this gives me the address of the member
		      //  function 
}


But still I can't execute the function directly like its seems I
should be able to:

inline int a::do_something( afp x ) { return x(); }   // no way


The only way I can get it to go is to have another member function
like: 

int a::executor(cfp x) {
  if (x == a)
    return a();
  if (x == b)
    return b();
  if (x == c)
    return c();
}


Where to get the value from a passed member function pointer the usage
is: 

inline int a::do_something(afp x ) { return executor(x); }   


I know this is ugly and stupid and I hope there is a better way.


>O----------------------------O----------------------------------O
>| Jonas Svanberg             | Email: svan@isy.liu.se           |
>| Div. of Information Theory O----------------------------------O
>| Department of E.E. (ISY)   | "And Noah said to the animals:   | 
>| Linkoping University       |  - Go out and multiply!          |  
>| S-581 83 Linkoping         |  The snake said:                 |
>| SWEDEN                     |  - But how can I? I'm an adder!" |
>O----------------------------O----------------------------------O
>


JE
--
 RM 2104, HSc Building,  Div. Neuroscience
 U. Calgary School of Medicine,  3330 Hospital Drive NW
 Calgary, Alberta       T2N 3Y4
 (403) 220 4493  (wk)