From: Gary Hubbard <ghubbar AT sandia DOT gov>
Newsgroups: comp.os.msdos.djgpp
Subject: g77 problems
Date: Thu, 20 Nov 1997 09:00:28 -0800
Organization: Sandia National Laboratories
Lines: 27
Message-ID: <34746CAC.427@sandia.gov>
Reply-To: ghubbar AT sandia DOT gov
NNTP-Posting-Host: hubbard.ran.sandia.gov
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Precedence: bulk

I have been using the DJGPP port of g77, and have had a few
problems.  These may all be in the GNU stuff itself, or in the
DJGPP port, I have no way of knowing.

    First, I get page faults in a number of different ways.  I
compiled a program from the net originally run on a SUN.  It had
a STOP 'stop message' as its last executable line.  When run that
way it crashed at random at exit.  Running it again, with the same
inputs, might or might not reproduce the same effect.  Removing
the message after the word stop eliminated the problem.  I also
get page faults if I don't delete a specific output file before
running the program.  Shouldn't I just get a clean "Cannot create"
type message.

    Second, and more important is the executable size.  As with
many fortran programs, I need to compile with a -fno-automatic
option.  Since there is no recursion, this is no real limitation,
except that the executable size grows proportionally with the
runtime size.  Shouldn't the storage go in a .bss section, which
is created and zeroed at load time, or does DPMI prevent this
somehow.  Loading a 30MB file off the disk is just too slow.  I
seem to remember seeing a gzip based loader which would probably
help, but the link step would still be very slow.  Is there a way
to avoid storing all the zero bytes on the disk

Thanks,
	Gary