Path: utzoo!mnetor!tmsoft!torsqnt!jarvis.csri.toronto.edu!rutgers!ucsd!tut.cis.ohio-state.edu!UUNET.UU.NET!convex!csmith
From: convex!csmith@UUNET.UU.NET (Chris Smith)
Newsgroups: gnu.g++.bug
Subject: VINDEX_MAX on machines with negative addresses
Message-ID: <8911181252.AA12383@mozart>
Date: 18 Nov 89 12:52:10 GMT
Sender: daemon@tut.cis.ohio-state.edu
Distribution: gnu
Organization: GNUs Not Usenet
Lines: 16

If you follow this advice from g++.texinfo

                            In GNU C++, there are two ways of doing this: if
   the assumption is made that the largest offset into a virtual function
   table will always be smaller than the first text address available to the
   user, then define the symbol `VTABLE_USES_MASK', and set
   `VINDEX_MAX' to the largest power of two less than or equal to that
   size. 

you get a ton of  VINDEX_MAX redefined  messages because of this line in 
cplus-tree.h:

   #define VINDEX_MAX ((unsigned)128)

Since the right value for VINDEX_MAX is machine-dependent, shouldn't
it be defined (when needed) in tm.h ?