addressof overloaded static member function bug

[Nathan Sidwell <nathan at cs dot bris dot ac dot uk>]

Hi,
here's a bug present in the latest snapshot (1998/08/03) (I don't know
if it was in earlier snapshots too).
--begin overload.ii
class A
{
public:
  static int F1(void *);
  static int F1(void *, int);
  static void fn();
};

void A::fn()
{
  int (*f)(void *);
  
  f = &F1;
  f = &A::F1;
}
--end overload.ii

--begin session log
nathan@laie:12>uname -a
SunOS laie 5.5.1 Generic sun4u sparc SUNW,Ultra-1

nathan@laie:13>egcs-0803-g++ -v -c overload.ii 
Reading specs from
/home/staff/nathan/solaris/local/SunOS_5/lib/gcc-lib/sparc-sun-solaris2.5.1/egcs-2.91.53/specs
gcc version egcs-2.91.53 19980803 (gcc2 ss-980609 experimental)

/home/staff/nathan/solaris/local/SunOS_5/lib/gcc-lib/sparc-sun-solaris2.5.1/egcs-2.91.53/cc1plus
overload.ii -quiet -version -o /var/tmp/ccmivULs.s
GNU C++ version egcs-2.91.53 19980803 (gcc2 ss-980609 experimental)
(sparc-sun-solaris2.5.1) compiled by GNU C version 2.8.1.
overload.ii: In function `static void A::fn()':
overload.ii:13: assignment to `int (*)(void *)' from `int (*)(void *,
int)'
--end session log

If the declarations of the two A::F1 functions are swapped round, the
diagnostic does not occur!

The behavour I expect is that both A::F1(void *, int) and A::F1(void *)
are candidates, but the context of the &operator requires an int
(*)(void *) and therefore the latter function should be selected.

Notice that the explit selection of A::F1, rather than plain F1 does the
right thing, even though we're in a member of A and therefore there
should be no difference.

If my logic is flawed on this point, then it's still a bug, because the
declaration order of the two F1's should not affect the diagnostic.

nathan
-- 
Dr Nathan Sidwell :: Computer Science Department :: Bristol University
      You can up the bandwidth, but you can't up the speed of light      
nathan@acm.org  http://www.cs.bris.ac.uk/~nathan/  nathan@cs.bris.ac.uk