RE: SIGSEGV in jc1

["Boehm, Hans" <hans_boehm at hp dot com>]

I believe I had it all along.  I updated and rebuilt with the same result.
It would still be good to see whether someone else gets the same result.

I tried building the calculator directly from source, and (for the first
time to my knowledge!) that worked correctly.  There were still two other
less serious issues:

1) The source files contain copyright (C in a circle, code 251) symbols in
comments.  I have no idea by what rules that's legal/illegal.  But javac
seems to accept it, but gcj doesn't.  (This stuff was written by real
lawyers, not language lawyers.)

2) I saw strange (IO related ?) failures when I tried to link statically.

Hans


> -----Original Message-----
> From: Alexandre Petit-Bianco [mailto:apbianco@cygnus.com]
> Sent: Friday, April 20, 2001 4:40 PM
> To: Boehm, Hans
> Cc: 'Per Bothner'; tromey@redhat.com; java@gcc.gnu.org
> Subject: Re: SIGSEGV in jc1
> 
> 
> 
> Boehm, Hans writes:
> 
> > I built the current 3.0 tree on X86 with limited success.  If I try
> > to build my constructive reals calculator, it dies in jc1.  The
> > original command line mentions one .java file and multiple .class
> > files.
> 
> I think that this patch from Per was fixing the problem. Does your
> compiler have this patch?
> 
>   http://gcc.gnu.org/ml/gcc-patches/2001-04/msg00421.html
> 
> ./A
>