Date: Sun, 2 Jun 1996 10:17:08 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: Trash <trash AT dimensional DOT com>
Cc: John Brothers <jbxt AT netcom DOT com>, djgpp AT delorie DOT com
Subject: Re: djgpp and Win95 comments
In-Reply-To: <31AE2250.3D9A@dimensional.com>
Message-Id: <Pine.SUN.3.91.960602100838.19090Q-100000@is>
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII


On Thu, 30 May 1996, Trash wrote:

> > I had used v1 w/DOS, and a week ago, I upgraded to Win95.  My program
> > compiled fine using v1 gcc.  However, gdb no longer worked - it complained
> > about dpmi, and all the various incarnations I tried all failed in one way
> > or another.

In v1.x, you should use gdb-dpm under Windows.

> >   So, I would like to make the following suggestion:  for the readme.1st file
> > in the v2 directory, I think that mak373b.zip should also be a mandatory
> > file to download, if you've recently upgraded to Win95.

Was the previous Make the one from v1.x distribution?  If so, your 
problems are due to a well-documented limitation: v1.x programs cannot 
spawn v2 programs.  The `readme.1st' file clearly tells you to remove the 
old v1.x installation before installing v2.

> I wouldn't even try to run djgpp in a dos box.

Why not?  It runs just fine for me in that configuration.

>  Also djgpp runs fastest on my system with a large read
> write cached smartdrive.  Windows 95 rems out your smartdrive when it
> loads windows.

Windows 95 rems out SmartDrv because it has its own protected-mode disk 
cache that is about twice as fast as SmartDrv.

>  With a large smartdrv I've found the ramdrive useless.

Do you have benchmarks to support this?  My testing shows that even with 
a large SmartDrv, a RAM disk makes GCC faster.