Re: Internal error #19991113

[Rod Stewart <stewart at lab43 dot org>]

Bangs head on wall.  Opps, sorry.  I misread your report.  Discard what I
said earlier.  For some reason I thought you had a snapshot from 1999, my
mistake.

You might want to try changing some of your compile options as a temporary
work around.  Sometimes compiling with '-O', or '-O1' will work.

There were a few problems with the 20000911 snapshot, but I think for the
most part they were related to bootstrapping, so if you built your
compiler it may be OK.  (But using the head of the tree never hurts. :))

You might also want to use the g++ binary as opposed to gcc to compil c++
code.  My understanding is they do not produce the same code, with c++
being the better one.

Sorry about the confusion.
-Rms

On Wed, 13 Sep 2000, David Seekins wrote:

> hmmm... I guess I must not understand the way this software is
> snap-shotted - I thought I grabbed the very latest code (which I ftp'd
> yesterday) and it was labeled 20000911 - I suppose "pretty old" is a
> relative term - but I would say this is pretty new (2 days old) code - I
> guess I must be wrong.
> 
> I will grab the code from the CVS repository this time, rebuild it and try
> again...
> 
> Thanks again...
> 
> Dave Seekins - dseekins@lightsurf.com
> 
> 
> ----- Original Message -----
> From: "Rod Stewart" <stewart@lab43.org>
> To: "David Seekins" <dseekins@lightsurf.com>
> Cc: <gcc-bugs@gcc.gnu.org>
> Sent: Wednesday, September 13, 2000 12:06 PM
> Subject: Re: Internal error #19991113
> 
> 
> > On Wed, 13 Sep 2000, David Seekins wrote:
> >
> > > The output of the run is listed below (and is included in the
> > > output.txt file attached). The source file and the ii file and all the
> > > includes (many more than are actually needed - sorry about the size)
> > > are in the attached tar ball. If there is anything I forgot, or any
> > > additional information that you need, or if there is anything I can do
> > > to help - please don't hesitate to contact me.
> >
> > > Compiler version:
> > >
> > > egcs-20000911 snapshot
> >
> > That's pretty old.  I would recommend you try updating to the current
> > mainline.  From the GCC main page:
> >   May 1, 2000
> >   Richard Earnshaw of ARM Ltd, and Nick Clifton of Cygnus, a Red Hat
> >   company, have contributed a new backend for the Arm and Thumb
> >   processors.
> >
> >   The new backend combines code generation for the Arm, the Thumb and the
> >   StrongArm into one compiler, with the target processor and instruction
> >   sets being selectable via command line switches.
> >
> > And there have been a number of other 'fixes'.  I can build your program
> > with my native armv4l-linux compiler
> > gcc version 2.96 20000909 (experimental)
> > although minus the '-mthumb-interwork-Wall' option.
> >
> > [root@zip gcc_bug_dir]# g++ -c  -Iincludes  -Wpointer-arith
> > -Wstrict-prototypes -Winline -Wundef -Woverloaded-virtual -g -O2
> > -ffunction-sections -fdata-sections -fno-rtti -fno-exceptions -fvtable-gc
> > -finit-priority  -o imbroken.o memfixed.cxx
> > [root@zip gcc_bug_dir]# ls -l imbroken.o
> > -rw-r--r--   1 root     root        41736 Sep 13 12:05 imbroken.o
> >
> > Good luck,
> > -Rms
> >
>