NetNews Usenet Archive 1992 #19/NN_1992_19.iso/spool/comp/os/os2/misc/28038

Internet Message Format | 1992-08-23 | 8.0 KB

Xref: sparky comp.os.os2.misc:28038 comp.os.os2.advocacy:4408
Newsgroups: comp.os.os2.misc,comp.os.os2.advocacy
Path: sparky!uunet!caen!hellgate.utah.edu!jaguar.cs.utah.edu!brian
From: [e-mail] (Brian Sturgill)
Subject: Re: I've seen NT......and I'll take OS/2 2.0 thanks!
Date: 23 Aug 92 11:59:32 MDT
Message-ID: <[e-mail]>
Organization: University of Utah CS Dept
References: <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]>
Lines: 142

In article <[e-mail]> [e-mail] writes:
...
>>Windows NT forces programs to go through intercepted hardware accesses that
>>are bounds checked, and then performed for the program.  This of course
>>is slower, and only those device registers directly accounted for in the
>>DOS subsystem can be used.  As I've said before OS/2 has NT beat hands
>>down in this ONE area.  Again, referring back to my recent post proposing
>>the direction OS/2 should go in.... OS/2 should seek the middle ground.
>>Keep the DOS compatibility, and concentrate specifically on the Desktop
>>user's needs.  OS/2 would need to lessen it's DOS support too if it were to
>>try to fill the same area that NT is trying to fill (i.e. "real" security).
>>NT has at least a two year lead getting symmetric multiprocessing and heavy-duty
>>networking, and other high-end features in... OS/2 is better off not trying
>>to be an NT-wannabe.
>>
>Sorry, I disagree with all you are posting here.
>Are you suggesting that IBM should drop Lan Server (the 3.0 beta test
>just started this week) and Extended Services?
I'm not suggesting IBM drop anything... see below.
Would you like to post a feature list?
How much will it cost?

>Come on, you must be joking. Also, I fail to see why you cannot have
>compatibility with DOS apps and security. The fact that NT doesn't
It is impossible to do what IBM did to give OS/2 DOS compatibility and
be C2-level secure.  Quite simply OS/2's DOS box allows unrestricted
access to hardware devices, which while it makes for good DOS compatibility
allows any DOS program to override any security measures that would be put
in OS/2.  OS/2 could have security, but it would have to go to the same
approach as NT... validation of all hardware accesses, thus incurring
the same problem... not being able to allow many/most installable DOS device
drivers.  This is NT's only DOS BOX weakness (in design) relative to
OS/2's design.  (Really, see the PC Week article... it's a very good examination
of both systems.)

>give it, doesn't mean it cannot be done. The DOS VDM's in OS/2 2.0
>are capable of more than you seem to know about it.
>About SMP: You must have missed the IBM/Parellan announcement.
Yes I did miss it... would you like to provide details?

>What MS lead are you talking about?
>Also don't forget the parallel database technology demo, IBM showed
>on Comdex, fall 1990, running on OS/2.
I also saw a demo of Windows 3.1 at the OS/2 roll-out at COMDEX...
I still don't have that.

>Finally you seem to forget all about additional security that can be
>obtained using client/server technology.
>An 'NT wannabe': that will be difficult. It's more NT that tries to
>become an 'OS/2 wannabe'.
Not at all, I think you're being irrational, please calm down and look
at things as they stand.  3 years ago, Microsoft and IBM had a fight,
Microsoft decided to have their own OS, which they designed from the
ground up to be in the style of OS/2 1.X at the API level with a Windows
GUI.  From the start this code was: portable, had symmetric multiprocessing,
security, POSIX support, etc.  This project is nearing completion.
IBM has not spent nearly the time doing this.  As of two months ago they
had only just decided to use Mach as a basis for OS/2 3.0 (unless IBM was
lying, but I doubt they did).  They themselves say this is how OS/2
will get multiprocessing.  While (according to rumor) it seems that
IBM has been writing portably the sections of OS/2 1.X they recoded for
OS/2 2.0, this does not account for symmetric multiprocessing, and the
fine RPC and networking features of NT, nor security, nor a mail-api,
nor.... sigh the list is just too large.

I am not suggesting IBM give up current features of OS/2... that's not
it at all.  What I am suggesting is they FOCUS on serving the desktop user.
This is an area where NT has a problem (due to DOS incompatibilities,
lack of WPS, integrated REXX, etc).  IBM _IS_ TRYING TO IMITATE NT.
For example at the April COMDEX Roll-out of OS/2, Reiswig in response to
questions to a previous announcement of NT, promised to have NT features,
such as POSIX.  This is rediculous... why is IBM FOLLOWING Microsoft?
(Not that Microsoft's direction is bad, but they have a big lead... surely
you can believe 1 year... I, taking into account's OS/2 current state and
IBM's slowness to move, believe the lead is more like 2 years.)
What I'm hoping is that IBM will quit worrying about NT getting the high-end
market (that's virtually a done deal), and start worrying about getting OS/2's
rightful piece.  OS/2 is a nice personal OS... why fatten it with high-end
features?

What I see happening is that IBM is failing to improve OS/2's current state,
while squandering resources on high-end features.  Why should I believe
in promised C2-level security, POSIX, symmetric multiprocessing, etc..
from a company that still has not delivered the 32-bit GDI that "just
barely missed" its PROMISED March delivery date, and its JUNE CSD date.
It still doesn't have super-VGA drivers!

Did anyone else notice in the July OS/2 Monthly the article that had the
interview at COMDEX with John Soyring?  Here's the question that upset me:
    JOEL:
    I understand that sometime in June, you're coming out with a 32-bit
    graphic engine upgrade.

    JOHN SOYRING:
    ... (denial that it's 16-bit omitted -- it's a "hybrid")
    ...
    So we got to the point in March where the 32-bit graphics engine
    was done. They finished the coding, we finished the internal
    testing, and it was meeting specifications. We also had a decision.
    Both were operating very well. One had extensive external testing
    as well. Which one did we ship initially? Which one would you
    ship if quality was you top objective?
    ...

Now if this is true... why do we still not have that 32-bit engine?
Surely this could have been ready by the end of June?  Why, if this is true,
did an IBM employee call me up in response to the posting I made in
comp.os.os2.programmer (complaining about lack of support for ISV's) and
tell me that the 32-bit GDI is STILL (as of a few weeks ago) so buggy that
they're refusing to install it internally?

If getting these problems fixed are so hard for IBM, then they have no
prayer of directly competing with NT.  Thus I suggest that they take
OS/2's strengths and optimize those.  Compete by differientiation, rather
than head-on.

>
>>>
>>>- Microsoft will no longer give any release date for NT (What a
>>>surprise - it seems like it is delayed!)
>>What an odd thing to say, Microsoft has never given a release date for NT...
>>to this point, not even for the beta.  The program development schedule
>
>(lines deleted)
>Perhaps not, but for how long are we now hearing about NT?
>And maybe there never was an official announcement, but MS folks
>have been publicly mentioning all kind of dates.

Big deal!  These are not Microsoft spokespersons.  Just because IBM
has such draconian "security" measures that IBM employees can't do likewise
isn't a fault with Microsoft, but surely of IBM.  If IBM truely had any
marketing sense, they too would be discussing openly their future directions and
approximate time frames.  I'm glad I know about Microsoft's future plans.
I'm glad to know about Cairo, even if it's largely vapor at this point.
It's good to know what they have in mind.
On the other hand I take projected dates as projected dates... they are not
firm, and will most likely be overly optimistic.

...
>Anton Versteeg - IBM - Uithoorn Netherlands - TEAMOS2 NL

Brian