Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!bloom-beacon!mcgill-vision!clyde.concordia.ca!ccu.umanitoba.ca!rpjday
From: rpjday@ccu.umanitoba.ca
Newsgroups: comp.lang.c++
Subject: multiple inheritance problem?
Message-ID: <1990Aug24.104618.17291@ccu.umanitoba.ca>
Date: 24 Aug 90 10:46:18 GMT
Organization: University of Manitoba, Winnipeg, Canada
Lines: 56


  G'day eh?  I am experiencing a bizarre compile time problem using
the AT&T 2.0 cfront with the generic Sun C compiler.  To wit:

class base {
 .
 .
public:
    objptr() ....

}

  From this, I derive two more classes:

class derived1: virtual public base {
...
}

class derived2: virtual public base {
...
}

  The reason for the "virtual" is that I use multiple inheritance
to derive a third class from the above two derived classes.

class derived3: public derived1, public derived2 {
...
}

  All of the above is based on the explanation given in Eckel's book
"Using C++" (as opposed to, I suppose, admiring it or taking it
for a walk).  The reason for the "virtual" is allegedly to prevent
ambiguity in class "derived3" when it realizes it has inherited
two identical "objptr" methods.  This sounds perfectly reasonable
until one gets around to compiling.  The error msg (which I reproduce
from memory, so be tolerant) is

AT&T 2.00 C++ ... error: bus error <or something nasty like that>

(The little addendum to the above error msg is not my editorial,
it's part of the error msg.)
  The msg gives the line number for a line in the application program
for which

derived3  derf ;
.
.
.
.	derf -> objptr() ;     <-- offending line


Is the multiple inheritance really set up properly?  Is this a 
known bug?  Have I done something totally stupid? 

R. Day
U of M