Xref: utzoo comp.unix.xenix:2886 comp.unix.microport:1169
Checksum: 31768
Path: utzoo!utgpu!woods
From: woods@gpu.utcs.toronto.edu (Greg Woods)
Date: Thu, 4-Aug-88 01:35:47 EDT
Message-ID: <1988Aug4.013547.27285@gpu.utcs.toronto.edu>
Organization: G. A. W. Constulting
Newsgroups: comp.unix.xenix,comp.unix.microport
Subject: Re: Bell Tech 386 SysVr3
Summary: not so long, nor boring, though partly repetitive.
References: <25145@ucbvax.BERKELEY.EDU> <465@sp7040.UUCP> <11643@steinmetz.ge.com> <1988Jul30.141708.3175@gpu.utcs.toronto.edu> <86@jetson.UUCP>
Reply-To: woods@gpu.utcs.Toronto.EDU (Greg Woods)

In article <86@jetson.UUCP> john@jetson.UUCP (John Owens) writes:
>In article <1988Jul30.141708.3175@gpu.utcs.toronto.edu>, I write:
>>[ nothing, it seems :-) ]
>First, you are consistently comparing various *386* System V ports
>with *286* Xenix.

I know, and I also like to compare 3Bs, and Suns, and Vaxes, and such :-)

>Not for me.  I've run UUCP connections at 9600 baud on direct serial
>lines on Xenix 286.  [ and I've done it at 19.2 Kbps -Greg ]

This is the boring part:  Re-read my comments, and the other long boring
version I posted today.  You will lose data at 1200 baud when sharing
interrupt vectors with more than one port when using the generic
built-in serial driver of Xenix 2.2.x with some interface cards, even
when using XON/XOFF flow control.  (whew :-)

Did you calculate the effective baud rates, and count the garbled packets?

>I hope so - almost any 386 compiler is going to be better than any 286
>compiler, since you don't have to worry about segmentation.  I've had
>only one minor problem with Microsoft's 386 compiler (an infinite
>spill, easily worked around), and am happy with the code it produces.

Well, I wasn't really fair, but what I was thinking of were problems
with pointer expressions, and with register variables, that shouldn't be
related to the CPU (I don't really know, I didn't write the compiler).

Again, even SCO support suggested not using the Xenix 386 for
development until the "merged AT&T / SCO" release, or at least until
another version of the compiler is released.  Microsoft consistently
refuse to fix bugs, and ignore many pleas for help.  Even the DOS 5.1
version has lots of problems.  I'm not sure what DOS version the Xenix
386 compiler is derived from, but the 2.2 286 version is from MS-C 3.00
according to the copyright string.

>-- 
>John Owens		john@jetson.UPMA.MD.US
>SMART HOUSE L.P.	uunet!jetson!john		(old uucp)
>+1 301 249 6000		john%jetson.uucp@uunet.uu.net	(old internet)
-- 
						Greg Woods.

UUCP: utgpu!woods, utgpu!{ontmoh, ontmoh!ixpierre}!woods
VOICE: (416) 242-7572 [h]		LOCATION: Toronto, Ontario, Canada