Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!linus!philabs!seismo!harpo!floyd!clyde!akgua!msdc!win
From: win@msdc.UUCP
Newsgroups: net.bugs.4bsd
Subject: Ranlib problem in 4.1BSD
Message-ID: <1049@msdc.UUCP>
Date: Mon, 7-Nov-83 16:31:07 EST
Article-I.D.: msdc.1049
Posted: Mon Nov  7 16:31:07 1983
Date-Received: Wed, 9-Nov-83 00:59:43 EST
Lines: 42

On our VAX 750 running 4.1BSD if you are trying to ranlib an object
archive and the currently existing __.SYMDEF happens to have the
correct lucky number in it, ranlib will calmly give up and tell you:

    ranlib: archive.a (__.SYMDEF): old format .o file

This happens when the first octal number in the currently existing
__.SYMDEF file is a magic number.  Rather than skip this file in the
loop, ranlib continues on causing the above mentioned error to
happen.

Below is a simple patch to fix this problem when the first octal
number in __.SYMDEF happens to be a "magic" number.

Diffs follow:

***************  ( old ranlib.c )
*** 64,69

     fread((char *)&exp, 1, sizeof(struct exec), fi);
     if (N_BADMAG(exp))
            continue;
     if (exp.a_syms == 0) {
            fprintf(stderr, "ranlib: warning: %s(%s): no symbol table\n", *argv, archdr.ar_name);

--- 65,74 ----- ( new ranlib.c )

     fread((char *)&exp, 1, sizeof(struct exec), fi);
+ #ifdef ORIGINAL
     if (N_BADMAG(exp))
+ #else
+    if (N_BADMAG(exp) ||
+       strcmp(archdr.ar_name,"__.SYMDEF") == 0)
+ #endif
            continue;
     if (exp.a_syms == 0) {
            fprintf(stderr, "ranlib: warning: %s(%s): no symbol table\n", *argv, archdr.ar_name);


    Win Strickland, Jr.
    Medical Systems Development Corporation
    ...!allegra!akgua!msdc!win