objc-parse.y created in egcs/gcc/

objc-parse.y created in egcs/gcc/

[Geoff Keating]

I got this from my automated tester.  It means that
some Makefile is creating objc-parse.[yc] in the source directory,
and there is no .cvsignore entry for it there.  One of those two
things should not be happening; IMHO, for development versions it
should be created in the object directory, and for release tarballs it
should not be created at all (unless --enable-maintainer-mode?).

------- Start of forwarded message -------
Date: Sat, 29 Jul 2000 17:38:06 -0700 (PDT)
From: Geoffrey Keating <geoffk@mailhost.cygnus.com>
To: geoffk@sloth.cygnus.com
Subject: Output from "cron" command
Content-Type: text
X-UIDL: 1ded47a1279a0d069d990f3ba468d414

Your "cron" job on sloth.cygnus.com
/home/geoffk/tbox/tbox.sh

produced the following output:

? egcs/gcc/objc-parse.y
? egcs/gcc/objc-parse.c
------- End of forwarded message -------


-- 
- Geoffrey Keating <geoffk@cygnus.com>
>From caught@prodigy.net Sat Jul 29 19:09:00 2000
From: kenneth topp <caught@prodigy.net>
To: gcc-bugs@gcc.gnu.org
Subject: BUG:  (0 && 1) == 1  (shows up in xmms with gcc 2.96)
Date: Sat, 29 Jul 2000 19:09:00 -0000
Message-id: <Pine.LNX.4.21.0007291851001.14356-100000@dynasty>
X-SW-Source: 2000-07/msg00756.html
Content-length: 1785

Hello, 

Software: xmms v1.2.2 (current release)
Platform: redhat rawhide
Glibc: redhat's glibc-2.1.91-16

in xmms-1.2.2/xmms/input.c (line 326), there this code:
  
 if (ip->is_our_file(filename) && !g_list_find(disabled_iplugins,ip))
         break;

the ip->is_out_file(filename) evaluates to 0, but it evaluates
positively (ie: run's that break)

If I add a printf to that if block the code executes properly (although
slower).  if I use egcs-2.91.66, it also works properly.

Let me know if there are more details I can assist with (or if this is
known issue ;)

Kenneth Topp

btw - the result of this bug, is that xmms passes off to the wrong input
plugin to try to determine song name, and length.

-----------related function-----
void input_get_song_info(gchar * filename, gchar ** title, gint * length)
{
    GList *node;
    InputPlugin *ip = NULL;

    node = get_input_list();
    while (node)
    {
        ip = (InputPlugin *) node->data;
        if (ip->is_our_file(filename) && !g_list_find(disabled_iplugins, ip))
            break;
        node = node->next;
    }
    if (ip && node && ip->get_song_info)
        ip->get_song_info(filename, title, length);
    else
    {
        (*title) = NULL;
        (*length) = -1;
    }
}
--------------
compiled like:

gcc -DHAVE_CONFIG_H -I. -I. -I.. -I/usr/lib/glib/include -D_REENTRANT
-I/usr/X11R6/include -O2 -march=i386 -mcpu=i686 -fomit-frame-pointer
-funroll-all-loops -finline-functions -ffast-math -m486
-DDATA_DIR=\"/usr/share/xmms\" -DPLUGIN_DIR=\"/usr/lib/xmms\"
-DPLUGINSUBS=\"Output\",\"Input\",\"Effect\",\"General\",\"Visualization\"
-DLOCALEDIR=\"/usr/share/locale\" -DI386_ASSEM -I../intl -O2 -march=i386
-mcpu=i686 -fomit-frame-pointer -funroll-all-loops -finline-functions
-ffast-math -m486 -c input.c

Re: objc-parse.y created in egcs/gcc/

[Mark Mitchell]

>>>>> "Geoff" == Geoff Keating <geoffk@cygnus.com> writes:

    Geoff> I got this from my automated tester.  It means that some
    Geoff> Makefile is creating objc-parse.[yc] in the source
    Geoff> directory, and there is no .cvsignore entry for it there.

I think this is a temporary hiccup from the patch from Marc Espie that
I installed, but then immediately backed out after the parallel make
problems.

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com
>From sharmasa@home.net Sun Jul 30 10:55:00 2000
From: sharmasa@home.net
To: conftest.i@quaker.sam.com, conftest.s@quaker.sam.com, gcc-bugs@gcc.gnu.org
Subject: Bug report 
Date: Sun, 30 Jul 2000 10:55:00 -0000
Message-id: <20000730140257.A14678@quaker.sam.com>
X-SW-Source: 2000-07/msg00763.html
Content-length: 519

This works on the same machine with gcc 2.7.2.3 and on another machine with gcc 2.91.66.
This was actually out of an autoconf script.
 don't know if this will be helpful, the gdb backtrace

 on the core file
(gdb) bt
#0  0x40076262 in strrchr ()
#1  0xbffff594 in ?? ()
#2  0x40030cb3 in __libc_start_main (main=0x80483f0 <main>, argc=1, argv=0xbffff594, 
    init=0x804829c <_init>, fini=0x8048444 <_fini>, rtld_fini=0x4000a350 <_dl_fini>, 
    stack_end=0xbffff58c) at ../sysdeps/generic/libc-start.c:78
(gdb) quit