> As  I  said before, I don't understand you. For
 > ordinary functions and
 > variables, 'static' has to have the same effect in C++
 > as in C.
   I'm talking about methods. I tried this on some compilers long time ago, 
and it failed.
class x {
private:
void    xx(void);
};
static void x::xx(void)
{
}
   BC++ 3.1 complained that multiple storage classes specified. OK, first one 
is static. But what is second? 'Method that has to be visible to other 
modules even if it is private and is used only in this module, with the 
compiler being responsible for preventing others to access it'? Impressive.
 > Stroustrup discussed this in "The Design and
 > Evolution  of  C++"  (p.
 > 266/7).  As  a rule of thumb, he has always been very
 > reluctant adding
 > new features to C++  if  their  effect  could  be
 > reached  using  the
 > existing features (a virtual accessor method in this
 > case).
   Correct, but it's two-in-one, virtual accessor to the static variable. And 
he is contradicting with himself, 'long float' = 'double', 'for' = 'do 
something, while(), { ... do something }', wtf is 'signed' or 'auto'? Anyone 
there who is seriously using it? And where are 'bit' enums that have values 
0x0001, 0x0002, 0x0004, and so on instead of 0, 1, 2, 3...? I don't think i'm 
the only one who would like to see them.
 > Yes,  RTTI  provides  this  functionality.  Using
 > RTTI  can  be  very
 > convenient sometimes, but it's good to always think
 > twice before using
 > it because it tempts you to design badly.
   Dunno, never used one (hopefully never will).
 > PS:  I  agree  with  the  general message of your
 > message: overloading
 > keywords (static, virtual) doesn't seem to be very
 > fortunate.
   Not the overloading itself, but the context-dependant usage that is 
different for ordinary functions and functions that have implied first 
parameter as a pointer to something (and even the function itself is called 
method, accessor or maybe something else).
---
---------------