🚀 go-pugleaf

In article <890139w.702661419@aucs> 890139w@aucs.acadiau.ca (John Weatherbie) writes:

> [Yes, again, the great Is Windows an OS debate, rearing its head again...]

>OS/2 is a great operating system, Windows is a program that runs under OS/2 and DOS. Get it right!!! Buy OS/2 and forget the term UAE.

Just a question: If I were to write an OS/2 app with the following code in
it, what would happen (let's assume 16-bit PM api)

xor	ax, ax		<- wait--is this code in a code segment?
mov	ds, ax		<- selector 0 is typically invalid
mov	bx, 0ffffh	<- and what if the segment is < 64K
mov	[bx], ax	<- and who says we can write to it, past the hard end

; If I'm here w/o a 'fault', I just zeroed out a random chunk of memory...!

Okay, maybe it's not called a UAE, but it's some kind of processor fault.
Or OS/2 is lying to you (there are three potential faults in this code).
Or the memory protection isn't turned on.

Reality check: most Win apps which UAE under Windows will do something
similar under OS/2 2.0.  The difference may be that OS/2 2.0 recovers more
gracefully; this has to be tested in the field first.  The fact is that
in most cases a UAE is saying that the application did something bad, typically
loading an invalid segment register, read/writing past the end, or
writing into a read-only place.  Changing the OS to OS/2 can't fix a bad
app (i.e. one that doesn't quite get its far arithmetic right).

Admittedly, there are cases where conditions cause Windows itself to
UAE, such as low memory or programs passing bad parameters to Windows.
There are also a few bugs in the Windows kernel which cause UAEs.
Undoubtedly, there are bugs in Win 3.1's kernel and OS/2 2.0s.

Even fixing all this, there's only so much any OS can do when an app
contains code like that I've written above.

Anyway, Win 3.1 and OS/2 2.0 are both cheap--you can have both for under
$100, and I imagine quite a few people will use both as the need arises.

Aaron Wallace