This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: egcs 1.03a & maps (bug?)
- To: markus at sfb504 dot uni-mannheim dot de (Markus Klink)
- Subject: Re: egcs 1.03a & maps (bug?)
- From: Joe Buck <jbuck at gr05 dot synopsys dot com>
- Date: Thu, 2 Jul 1998 13:41:48 +0200 (MET DST)
- Cc: oliva at dcc dot unicamp dot br (Alexandre Oliva), egcs at cygnus dot com
> Indeed the suggestion from Alexandre to use the optimzer showed a positive
> result.
Yes, but that fix won't work in general. It worked in this case because
the symbol with the name that was too long was inlined.
> The error did not show on my Linux box as I verified later on. Can I
> expect to get rid of these errors when I use optimization. I really would
> like if I had not to install the GNU assembler.
The error is in Sun's assembler; it breaks badly if it is given a symbol
that is too long, and gives error messages that don't make sense. The
name is so long because of mangling. There is a new mangling scheme
under test, but since it breaks link compatibility to use a new mangling
scheme it won't be out for a while.
An alternative is to derive from string and use the derived class.