🚀 go-pugleaf

> Date: Mon, 6 Apr 92 18:37:48 EDT
> From: purtill%idacrd@uunet.uu.net (Mark Purtill)
> Message-Id: <9204062237.AA05389@>

This is a bug with libg++, it sounds like.  The behavior you site, is
correct.  Another possibility, is that you are just using libg++
incorrectly.

> On a Sparcstation 2 (which is not connected to the outside;  I'm
> typing this in from a printed transcript, so I might err).
> % gcc -v
> Reading specs from /usr/local/lib/gcc-lib/sparc-sun-sunos4.1/2.1/specs
> gcc version 2.1
> % more /etc/motd
> SunOS Release 4.1.1 (CARS_2) #3: Thu Aug 15 10:35:49 EDT 1991
> % cat >A.h
> #pragma interface
> struct A {
>     A() ;
>     };
> inline A::A() {}
> ^D
> % cat > B.C
> #include "A.h"

> struct B {
>     A* _copy(A* t);
>     };

> A* B::_copy(A* t) {
>     A* x = new A();
>   return 0 ;
>     }
> ^D
> % gcc -c B.C	
> 	And now if you examine B.o with nm, you should find A::A() as
> an undefined global.  Removing the "#pragma interface" causes the
> problem to go away.
> 	Now, "#pragma interface" does not seem to be documented, so
> maybe this is correct behaviour for gcc.  If so, then this is a libg++
> bug, which occurs in (at least) int SplayBag s.