🚀 go-pugleaf

In the future please report all g++ bugs to the
        bug-g++@prep.ai.mit.edu
list, thanks.  If they are back-end bugs, they can be reported here,
but please let us know if you think the problem is in the back-end.

You don't need to resubmit this one.

> From: d9hh@dtek.chalmers.se
> Message-Id: <9207192148.AA09547@hacke0.dtek.chalmers.se>
> To: bug-gcc@prep.ai.mit.edu
> Date: Sun, 19 Jul 92 23:48:19 MET DST

> Hi, I think I've found a bug, even though I'm not a 100% sure, this is
> probably something you can figure out in a split second.

> Here we go...

> Version:
> GCC 2.1 and 2.2.2 (these are the versions I've seen the error (?) in).

> Running on machine:
> Some kind of Unix machine (GCC2.2.2), but I've also seen it on my Amiga and on
> a Sun (version 2.1)

> This is what I think is wrong:
> When compiling C++ programs: When assigning a function pointer to
> the adress of another function something in the type checking fails.

> Ok to see it clearly here are some sourcecodes:

> file one (C++)
> --------------

> // Yeah I know I should use iostream.h instead, but I'm trying to
> // make a point here that's all...
> #include "stdio.h"

> void foo(int&);

> void (*bar)(char);

> main() {
> 	int e = 34;
> 	bar = &foo;     // *** Compiler should complain here
> 	bar(e);         // ...booom!!
> }


> void foo(int& a) {
> 	printf("output: %d\n",a);
> }

> EOF


> file two (still C++)
> --------------------

> #include "stdio.h"

> void foo(int*);

> void (*bar)(char);

> main() {
> 	int e = 34;
> 	bar = &foo;    // **** Compiler should complain here
> 	bar(e);
> }


> void foo(int* a) {
> 	printf("output: %d\n",*a);
> }

> EOF

> Ok. when compiling these two files doing 'gcc file1.cc' or 
> 'gcc file2.cc' the compiler says nothing at all about the difference in
> type for bar and foo. See? bar takes a char and foo takes an int reference
> or in file two, an int pointer, still, the compiler says nothing at all.
> When running the programs, all I get is 'bus error' or 'segmentation 
> fault'... (Or a nice crash on my Amiga.. :-).

> Trying to compile with 'gcc -ansi -pedantic -Wall' makes no difference.

> When trying to compile the same thing on CC instead, the compiler
> says this is clearly an error. The most interesting thing, though
> that kind of convinces me is that when compiling file two as plain C
> (which it is) GCC gives a warning about 'assignment from incompatible
> pointer type'.

> Hope this helps. And thanks for providing the world with the best
> compiler ever!


> -- 
> Henrik Harmsen     Internet:  d9hh@dtek.chalmers.se
>                Chalmers University of Technology, Sweden. 
>       "I haven't lost my mind -- it's backed up on tape somewhere."