[Fwd: null pointer dereferencing]

[apamecha at houston dot geoquest dot slb dot com (Abhisek Pamecha)]

hi,
PFA the mail wherein I have added a code sample which seems to lead to 
incorrect compiled code using

>g++ -v


Reading specs from /usr/bin/../lib/gcc-lib/i686-pc-linux-gnu/3.0.2/specs
Configured with: ./configure
Thread model: single
gcc version 3.0.2

Please have a look and if possible do let me know too if you think it is 
a bug in the compiler trying to optimize.

thanks
Abhishek

--- Begin Message ---

• From: Abhishek Pamecha <apamecha at houston dot geoquest dot slb dot com>
• To: Ganeshmurthy Patil <gpatil at houston dot geoquest dot slb dot com>,Arnab Nandi <anandi at houston dot geoquest dot slb dot com>,Suresh Govindaraju <sgovindaraju at slb dot com>
• Date: Thu, 28 Aug 2003 18:01:03 +0200
• Subject: null pointer dereferencing
• Organization: Schlumberger

class A{
public:
A() {}
~A() {}
void do_something(void){}

int b;

};

int main(){

A *a = 0;
//null pointer dereferencing!!!!!
a->do_something();

// a->b = 10; fails though
return 0;
}

this code works! Ganesh, this is similar to the strat log code that had same problem (CreateComp_dialog.cc) (suresh told me about it).

Only reason I can attribute to it is that the method is not changing the state of the object and a method belongs 'more' to a class than to an object. The compiler might have generated such a nasty side effect by replacing the statement by a direct call to the func instead of first de-referencing it in runtime. From a language point of view, it seems incorrect to me.

Can you provide different reasons why it should work the way it is?

thanks
Abhishek

--- End Message ---