Xref: utzoo comp.sys.att:11683 comp.unix.sysv386:4522
Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!tut.cis.ohio-state.edu!pacific.mps.ohio-state.edu!linac!att!att!fang!atssc!so9see!akgua!atl2!rwc
From: rwc@atl2 (Richard W. Cook)
Newsgroups: comp.sys.att,comp.unix.sysv386
Subject: Re: COFF & ELF
Message-ID: <199@atl2>
Date: 30 Jan 91 16:51:21 GMT
References: <1991Jan19.162846.19735@mccc.edu>
Followup-To: comp.sys.att
Organization: AT&T Computer Services Group, Duluth GA
Lines: 30

in article <1991Jan19.162846.19735@mccc.edu>, pjh@mccc.edu (Pete Holsberg) says:
> 
> Running SV/386 R3.2.2 on an AT&T 6386 with the V5.0 ANSI C compiler.  I
> have a C++ compiler that has COFF libraries.  Can I change them to ELF
> with the coff2elf thing in /usr/ccs/bin?  (Docs are at the office.)
> 
Yes:  If you only need to relink those libraries into an executable,
      you can use coff2elf to convert them.

      AT&T's C++ 2.1 Language System supports the ANSI C V5.0 compiler,
      so you're o.k. if planning on using it in conjunction with the 
      V5.0 compiler.

No:   I can't speak for other vendor's C++ products (not that I speak in
      any offical capacity for AT&T, either) but AT&T's C++ Language
      System depends on ar() to provide it with symbol table data.
      The V5.0 compiler version of ar() produces output that is incompatible
      with those earlier releases of C++.

      This incompatibility could be easily corrected in early distributions
      of C++ 2.1 with a shell script change.  I'm not sure how the "official"
      distribution handles it....

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Richard Cook                           
AT&T Computer Systems                  {attmail,att}!cpsc53!rwc
Customer Application Engineering       (404) 564-6605
Atlanta, GA.
           "I don't speak for AT&T and she don't speak for me."
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-