Re: Problems with running yarn on faster computers

From: rex (rex@ptw.com)
Date: Wed, 22 Sep 1999 12:02:52 -0700

On Wed, Sep 22, 1999 at 11:42:32AM -0700, rex wrote:
> On Wed, Sep 22, 1999 at 03:36:53PM +0100, Andrew Hodgson wrote:
> > >
> >>I was rather hoping I could keep my older setup (yarn, yes, boxer
> >>etc), but I get various runtime errors from the programs in the yes
> >>suite (yes.exe, yesx.exe etc).
>
> If you're getting divide by zero errors, yes is probably written in
> Borland's Pascal. There is a bug in the compiler that causes problems
> with fast processors. I don't know if there is a fix w/o the source
> code.

Update:
You might contact Joe if you're comfortable searching through Yes with
a debugger and patching the code.

==============================================================================
I ran into this problem in June '97 when I got a new machine, a 266
PII. It later happened at work on a 233 HP Vectra.

Using Turbo Debugger, I found and patched the offending code,
and Procomm now works fine.

The following files required patching (they all use the same
code fragment, so it was easy after the first time):

PCPLUS EXE
PCSETUP EXE
PCEDIT EXE
DT_PATCH EXE
PCKEYMAP EXE
PCINSTAL EXE

Basically, the patch bypasses the DIV instruction, and sets the
result to the maximum value that can fit in 16 bits (FFFFH).
It works fine on both machines now.

Joe
jdhagen@itis.com

===========================================================================