Date: Mon, 28 Sep 1998 12:38:44 +0300 (IDT)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: Test User <user AT ts001d22 DOT cin-oh DOT concentric DOT net>
cc: djgpp AT delorie DOT com
Subject: Re: Make works on some machines, not others
In-Reply-To: <slrn70rt4c.5bh.user@ts001d22.cin-oh.concentric.net>
Message-ID: <Pine.SUN.3.91.980928123822.29939D-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Precedence: bulk


On 27 Sep 1998, Test User wrote:

> >Make works just fine on several computers.  However, on a couple of
> >important machines, make will crash on every invocation with the
> >following error message:
> 
> >  General protection exception
> >  Fault location 0277:0044
> >  Interrupts in service: 1
> 
> Sounds to me like you're running Windows. That's what's going on. 

I don't know why are you so sure this is Windows' fault.  The DJGPP
port of GNU Make works for me on both DOS and Windows (all Windows
versions) without any problems, but there could always be some bug
lurking in the ported Make that only happens on some systems.

To the person who posted the original problem report: please post the
details, like the version of Make you are using, which Makefiles cause
it to crash (does "make --help" crash also?), and all the relevant
info about your system setup on those machines where it crashes.  For
example, do all the machines where it crashes have something in common
in the way they are configured?

> Windows and DOS are not designed to be totally compatible. That's
> what you get for using Wind'oh!s.

I don't know where did you get that idea.  To the best of my
experience, your assertion is profoundly untrue.  It is actually the
other way around: Windows 9X has a lot of baggage for reasons of
compatibility with DOS applications that prevents it from being a
better OS.

> If your DOS box is dying, then it's more than just GNU make crashing.
> The error message is appearing in a separate window, which means that
> Windows is causing the error. This isn't a regular crash that can
> even be _detected_ by gdb. Your DOS box would crash before GDB would
> know what's going on.

This is also not true.  Unless the problem happens in the very
beginning of the startup code, a debugger will at least allow to see
where exactly in the code does the problem happen, because you can run
Make in single step or in small increments to zero in on the locus of
the problem.

> Run make from DOS (without Windows running)
> and see what happens.

It would indeed help to know whether Make works on the same machine in
DOS mode, but it still should work under Windows as well.