[Bug c++/8057] [3.3/3.4 regression] Templates/non-templates and warnings about statements without effects

["gdr at gcc dot gnu dot org" <gcc-bugzilla at gcc dot gnu dot org>]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=8057


gdr at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mark at codesourcery dot com
         AssignedTo|gdr at gcc dot gnu dot org  |unassigned at gcc dot gnu
                   |                            |dot org
             Status|NEW                         |ASSIGNED


------- Additional Comments From gdr at gcc dot gnu dot org  2003-07-07 16:04 -------
This PR has nothing to do with diagnostics per se.  The diagnostic thing is just
a symptom of a much fuhndamental desease: It has to do with function cloning and
inlining. More specifically, the double emission of diagnostic is due to the
fact that the constructor is cloned, hence its body is expanded twice.
The absence of diagnostic in Z::Z() is due to the fact that it is an inline
function and starting from some GCC-3.x, an inline function is defined (hence
its body expanded) only if used. To see what I mean, make the definition of
Z::Z() a non-inline function and you'll see the warning.
If you add the following

    struct A {
       void f();
    }; 

    void
    A::f() { Y::i; }

you'll see that the diagnostic is emitted only once.

So, this issue has nothing to do with templates per se.  It all has to do with
cloning and inlining.  So I'll hand it off.