Re: c++/2009

[Randall Hopper <aa8vb at yahoo dot com>]

The following reply was made to PR c++/2009; it has been noted by GNATS.

From: Randall Hopper <aa8vb@yahoo.com>
To: mmitchel@gcc.gnu.org, Mark Mitchell <mark@codesourcery.com>
Cc: aa8vb@nc.rr.com, gcc-gnats@gcc.gnu.org, nobody@gcc.gnu.org
Subject: Re: c++/2009
Date: Thu, 12 Apr 2001 13:20:54 -0400

 mmitchel@gcc.gnu.org:
  |Synopsis: g++ (-lGLU)
  |
  |State-Changed-From-To: feedback->closed
  |State-Changed-By: mmitchel
  |State-Changed-When: Thu Apr 12 09:06:41 2001
  |State-Changed-Why:
  |    Cannot be fixed.
 
 Mark Mitchell:
  |
  |Thanks for the additional information.
  |
  |There's no way to make that work; you can't mix code from different
  |C++ compilers in general.
  |
  |With g++ 3.0, the situation will improve because there is a
  |cross-vendor ABI that several vendors are implementing.
 
 Ok, thanks for the reply.  However my reason for filing the bug was not to
 imply that gcc should generate a valid binary under these conditions, but
 rather...
 
 ...that GCC should at the very least fail the link and give the developer a
 clue that something is amiss, rather than just compiling a broken binary
 that core dumps deep in the bowels of iostream with a seg fault leaving the
 developer absolutely no clue what is wrong.
 
 I spent hours researching this failure on the net to find only one other
 who had hit this and knew the reason for the failure.  However, I found a
 number of others that had hit it and just gave up without a clue as to what
 was really happening.
 
 Can libstdc++ be built so that it will fail to link (or fail to dynamically
 link) when linked with libC (both define cin/cout/cerr -- why doesn't link
 fail -- I'm not a compiler expert)?  Alternatively, what about a post-ld
 link-check for libC.so on IRIX.
 
 Thanks,
 
 Randall