Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!mcsun!ukc!tcdcs!swift.cs.tcd.ie!wai
From: wai@swift.cs.tcd.ie
Newsgroups: comp.lang.c++
Subject: Re: Pointers to functions anyone?
Message-ID: <7120.2708a9e8@swift.cs.tcd.ie>
Date: 2 Oct 90 14:53:28 GMT
References: <6472@wolfen.cc.uow.oz> <1069@exodus.Eng.Sun.COM>
Distribution: comp
Organization: Computer Science Department, Trinity College Dublin
Lines: 54

In article <1069@exodus.Eng.Sun.COM>, landauer@morocco.Eng.Sun.COM (Doug Landauer) writes:
>> I'm trying to set up an array of pointers to functions within a class,
>> much like the example here (for simplicity no array here):
>> 
>> #include <stream.h>
>> 
>> class test {
>>   void f1(int i)			// function to be called
>>     { cout<<"ok"<<i<<"\n"; }
>> 
>>   void (test::*ptf)(int);		// pointer-to-function
> 
> 		Whoops, your comment shows what's missing here.
> 		ptf is *not* a "pointer-to-function".  Instead,
> 		it is a "pointer-to-MEMBER-function", which is
> 		a different beast altogether.
>> 
>> public:
>>   test() { ptf=test::f1; }		// constructor - assigns &f1 to ptf

   Is there any difference between 
    { ptf = &test::f1; }
   Both of them work okay.

> Read the E&S ARM section about pointers to members, and your compiler
> should be much happier about compiling your code.  In order to make
> your little test acceptable to cfront, all I had to do was change the
> call to give the ptr-to-member-fn an explicit "this" for which to call
> the function that it pointed at.  Like so:
> 
> Change the line
> 
>>     { (*ptf)(i); }			// try to call f1()
> 
> to a line like
> 
>>     { (this->*ptf)(i); }                    // try to call f1()
> 
>> What am I doing wrong??
> 
> All pointers-to-members must be given an object for which to call
> their pointed-at functions.  Even if the call is from within another
> member function.   Ptrs-to-members are usually implemented as
> OFFSETs within the class, rather than as true pointers.
> 
> I hope this helps....
> -- 
> Doug Landauer - Sun Microsystems, Inc. - Languages - landauer@eng.sun.com
>     Matt Groening on C++:  "Our language is one great salad."

Is there any way to creat a generic pointer to member function class and yet
be able to avoid all the signature type checking stuff?

Wai.