From: Linley Henzell <zel AT olis DOT net DOT au>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: DJGPP vs Borland C++
Date: Fri, 31 Jan 1997 09:44:43 -0800
Organization: Microtronics Information Systems
Lines: 46
Message-ID: <32F22F8B.12AE@olis.net.au>
References: <Pine DOT SUN DOT 3 DOT 91 DOT 970128115313 DOT 10987H-100000 AT is>
Reply-To: zel AT olis DOT net DOT au
NNTP-Posting-Host: ppp00.olis.net.au
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

Eli Zaretskii wrote:
> 
> On Mon, 27 Jan 1997, Linley Henzell wrote:
> 
> > help, but I finally managed to isolate my problem to a call to atoi. For
> > some reason, djgpp just doesn't seem to like my programs calling this
> > function. And I am using v2; is this one of its documented bugs?
> 
> What's the bug?  Can you post a short program that calls `atoi' and
> exhibits the bug?

I tried, but I couldn't isolate this problem, so I assume it is a result
of my program's overall structure. Someone suggested that null pointers
might be causing the GPFs, but as the variable I was passing to atoi was
an element of a local array, I don't think this was the problem.

> 
> > few minor irritations, like the immense volume of warnings which djgpp
> > generates (even without -Wall),
> 
> Errr, as opposed to what?  Some other compiler?  Then maybe some of those
> warnings tell you the cause for the bug?  In my experience, any warning
> that appears without -Wall is real trouble, so you should never disregard
> them.

As opposed to Borland C++ 3.1. I get two types of warnings:
warning: passing unsigned char [or whatever] as arg x of "foo(..., ...
etc)" changes signedness,
and another warning about inlining the "read" & "write" file I/O
functions. As BC never complained about these, and as they don't seem to
affect my program in any noticeable way, I'm ignoring them.

> > and the doubling of my .exe file from
> > ~150K to over 300K, but I can cope with that.
> 
> Add -s switch to the gcc link command line, and you will get back to 150K
> or so.

I tried adding -s to the linker options under RHide 1.1, but this didn't
work. I'm already using fstrength-reduce, and I think that without
either of these, my .exe output file ballooned out to over 500K.

Oh well.

Thanks for all the help,
Linley