INDIGO Disk Brackets (Lack of)
Nathan F. Janette
sens...@ricky.brainlab.utsa.edu (David M. Senseman) writes:
> From my personal exprience, I would say that SGI's
> position towards 3rd party hardware in general and disk
> drives in particular is admirable. Anyone who have
> followed comp.sys.sgi for any length of time knows how
> hard people like Dave Olsen labor to make 3rd party stuff
> work with the various SGI models. Try and get this kind of
> authoritative AND FREE support from IBM, Sun, HP or NeXT.
I'm not going to get into a NeXT vs. SGI argument, because I
like both vendors, although for different machines and
reasons. We have some Crimsons which are just great, and
many NeXTs which have a (IMHO) vastly superior GUI. Suffice it
to say that in general, "service" is not free from either
SGI or NeXT after the warranty period, which by the way is
only 90 days for SGI machines :-(
I think the presence of key SGI folks in this group is
wonderful, and I hope the folks at NeXT will take the hint.
> At the sales and field service level, SGI's attitude
> towards 3rd party stuff is also reasonable in my opinion
> -- a sort of laissez faire approach. They don't encourage
> you to buy 3rd party stuff (why should they?) but then
> again they don't threaten to void your warranty is you do.
> In the case of the Indigo you can EASILY hook-up a 3rd party
> drive to the external SCSI connector if you want to save a
> few bucks. The fact that you are demanding that SGI sell
> you proprietary hardware (i.e. the mounting brackets)
> is assurd. Next you will be demanding SGI sell you REX
> chips so you can build your own Indigo ;-) There's no doubt
> that internal disk drives are more convient than
> external ones, but don't expect this convience to come
> for free. Besides, we aren't talking about a simple $10
> device here either. If you ever LOOKED at the mounting
> hardware you would see that besides the mechanical
> aspects, there's 4 connectors and cabling as well.
There has been little difficulty, and no special mounting
requirements for 3rd party drives in NeXT machines.
> If you (or anyone else) can point-out a major workstation
> vendor that has a more liberal policy towards 3rd drives I
> would like to hear about it.
Glad to share my experiences.
If only SGI would port NeXTstep...
If only SGI would port NeXTstep...
If only SGI...
--
Nathan Janette "I'm a NeXTstep man,
Dept MB&B, Yale Univ I'm a NeXTcube guy"
New Haven, CT
nat...@laplace.biology.yale.edu (NeXT)
Scott Bennett
>
>If only SGI would port NeXTstep...
>If only SGI would port NeXTstep...
>If only SGI...
>
unless, of course, you mean they should port the operating system, too.
Attempting to port NeXTStep to IRIX would be somewhat less sensible
than trying to get your monitor illuminated by enclosing a kerosene
lamp inside of it.
Scott Bennett, Comm. ASMELG, CFIAG
Systems Programming
Computer Center
Northern Illinois University
DeKalb, Illinois 60115
**********************************************************************
* Internet: ben...@cs.niu.edu ben...@netmgr.cso.niu.edu *
* BITNET: A01SJB1@NIU *
*--------------------------------------------------------------------*
* "It appears that the rioters and looters [in L.A.] are going to *
* suffer for their misdeeds: the government intends to "help" them! *
* Seems a bit harsh to me, but I've always had these bleeding-heart *
* tendencies." --Paul Zrimsek in private correspondence *
**********************************************************************
Barry Merriman
> If only SGI would port NeXTstep...
> If only SGI would port NeXTstep...
> If only SGI...
SGI is going to steal it, not port it. That is, if you watched
the job newsgroup, you would have seen SGI some months back
assembling a team to develop an object oriented GUI and
development environment.
Gee, what a great idea! I'm glad I've had one of those for 3 years...
:-)
--
Barry Merriman
UCLA Dept. of Math
UCLA Inst. for Fusion and Plasma Research
ba...@math.ucla.edu (Internet; NeXTMail is welcome)
Dave Olson
| In article <1992Jul8.2...@cs.yale.edu> nat...@laplace.biology.yale.edu writes:
| >If only SGI would port NeXTstep...
| NeXTStep ported to a System V kernel?? Nathan, you're dreaming...
IRIX isn't a system V kernel in many respects. Does NeXTstep really
need some Mach specific features (seems unlikly if they are porting
to Intel systems)?
--
Let no one tell me that silence gives consent, | Dave Olson
because whoever is silent dissents. | Silicon Graphics, Inc.
Maria Isabel Barreno | ol...@sgi.com
Jeff Weinstein
> In <1992Jul9.0...@mp.cs.niu.edu> ben...@mp.cs.niu.edu (Scott Bennett) writes:
>
> | In article <1992Jul8.2...@cs.yale.edu> nat...@laplace.biology.yale.edu writes:
> | >If only SGI would port NeXTstep...
> | NeXTStep ported to a System V kernel?? Nathan, you're dreaming...
>
> IRIX isn't a system V kernel in many respects. Does NeXTstep really
> need some Mach specific features (seems unlikly if they are porting
> to Intel systems)?
The many mach'isms that crept into nextstep gave IBM real fits when they
were trying to port it. Its not a problem on the intel systems, since
next is supplying the entire OS, not just the UI.
--Jeff
--
Jeff Weinstein - X Protocol Police
MicroUnity Systems Engineering, Inc.
j...@microunity.com
Any opinions expressed above are mine.
Jeff Smith
>IRIX isn't a system V kernel in many respects. Does NeXTstep really
>need some Mach specific features (seems unlikly if they are porting
>to Intel systems)?
NeXTstep uses Mach message passing heavily. I believe NeXT is porting
thier version of Mach to Intel alone with NeXTStep.
Plus I don't think SGI has enough money to buy NeXTStep. IBM paid *alot*
of money to NeXT, and never shipped a fully supported product.
d.wagley
> If only SGI would port NeXTstep...
> If only SGI would port NeXTstep...
> If only SGI...
>
I have it from a reliable source that when he mentioned this in a
meeting with one of the higher ups at SGI he was told that at one
point SGI approached NeXT about a NextStep port and was turned down
cold.
Doug
Jeroen van der Zijp
Rumor has it that they tried, but that NeXT didn't want to
license NeXTSTEP to SGI :-(
An R4k Indigo + NeXTSTEP would have been awesome ......
--
+-----------------------------------------+----------------------------------+
| Jeroen van der Zijp | |
| Phone : (205)730-1284 | Al draagt een aap |
| Email : jer...@tparty.b17a.ingr.com | |
| Mail stop: IW17A4 | een gouden ring |
| Address : INTERGRAPH Corp. | |
| Madison Industrial Park | 't is en blijft een lelijk ding |
| Huntsville AL35895 | |
+-----------------------------------------+----------------------------------+
Derek Collison
> If only SGI would port NeXTstep...
> If only SGI...
>
> --
> Nathan Janette "I'm a NeXTstep man,
> Dept MB&B, Yale Univ I'm a NeXTcube guy"
> New Haven, CT
> nat...@laplace.biology.yale.edu (NeXT)
>
I could not agree with you more!!!!
=derek
--
Derek Collison <---> de...@oceania.com
Oceania Health Care Systems
325 Lytton Avenue Palo Alto, CA 94301
(NeXT Mail accepted)
Patrick Guelat
I hope that SGI will NEVER buy it. NeXTStep is always praised as the ultimate,
great UI. Well it has some nice features but all in all I prefer a REAL window
system like X11 with a good UI like Motif (or the derrivated 4Dwm).
It's terrible to have these menues hanging around in front of all windows,
the networking capabilities of NeXtStep are terribly primitive. With SGI's
Workspace and toolchest you have everything you need, what do you want more ?
--
Patrick Guelat, Improware AG Switzerland, CH-4414 Fuellinsdorf
Internet: pa...@imp.ch
Robb Aley Allan
NeXTStep ported to a System V kernel?? Nathan, you're dreaming...
unless, of course, you mean they should port the operating system, too.
Attempting to port NeXTStep to IRIX would be somewhat less sensible
than trying to get your monitor illuminated by enclosing a kerosene
lamp inside of it.
NeXT held discussions with Unix Systems Labs about porting to SvR4 in 1991. No dream. But
no port, either.
--
Robb Aley Allan, The Gulfstream Group
220 Sunrise Avenue, Palm Beach, FL 33480
voice: 407/ 832-4013 email: Robb_...@helical.com
Co-founder/director, Gotham Users of NeXT (GUN)
Barry Merriman
> NeXTStep ... has some nice features but all in all I prefer
> a REAL window system like X11 with a good UI like Motif
Ba Ha, Ha, Ha Haaaaaa!!!
Thats a _great_ joke, man!
Sean Luke
[refering to NeXTSTEP]
>I hope that SGI will NEVER buy it. NeXTStep is always praised as the ultimate,
>great UI. Well it has some nice features but all in all I prefer a REAL window
>system like X11 with a good UI like Motif (or the derrivated 4Dwm).
>
>It's terrible to have these menues hanging around in front of all windows,
>the networking capabilities of NeXtStep are terribly primitive. With SGI's
>Workspace and toolchest you have everything you need, what do you want more ?
Well, ease-of-use might be nice. And being able to write powerful programs
without pulling out all your hair.
X11/Motif has only two advantages:
It's cheap.
It's common.
Funny, but that seems to be the same advantages the text-based 8088 IBM PC had
in 1984 when everybody said, "Why buy a Mac? With an 8088 PC, why would you
need anything more?" And now the entire PC world is moving to Windows.
--
Sean Luke
Brigham Young University MILK: It Comes From Cows
se...@digaudio.byu.edu
NeXTmail and nifty Mac stuff welcome
Kenny Leung
ultimate,
> great UI. Well it has some nice features but all in all I prefer a REAL
window
> system like X11 with a good UI like Motif (or the derrivated 4Dwm).
>
> It's terrible to have these menues hanging around in front of all windows,
> the networking capabilities of NeXtStep are terribly primitive. With SGI's
> Workspace and toolchest you have everything you need, what do you want more ?
>
> --
> Patrick Guelat, Improware AG Switzerland, CH-4414 Fuellinsdorf
> Internet: pa...@imp.ch
I'm really getting tired of hearing these unsubstantiated pokes against NeXT
and NeXTstep. I can respect a difference of opinion. If you don't like
NeXTstep and tell me why, I'll go off into the NeXT people's corner and never
bug you again, but this "I like it just because" really burns my butt. So,
let's make this into an argument instead of just letting it be a tantrum.
Patrick, would you be so kind as to reply to these points?
0) Nice Features? Which ones are these? Do they exist in 4Dwm? If they don't,
then why is 4Dwm better?
1) What makes NeXTstep an UNREAL window system, unlike X11?
I grant you that you hate to have menus floating on top of everything.
2) What networking capabilities is NeXTstep missing?
3) Tell me what SGI's Workspace and toolchest can do, and I'll tell you if I
want more. And Why.
--
Kenny Leung |
kenny...@mts.ucs.ualberta.ca |
Michael Zyda
academic now seeing how the designer of neXTStep is not even
with neXT anymore but with Sun Microsystems? Don't we all
expect neXT to blow away any day with people bailing out
like their key software designer?
mz
David M. Senseman
Hmmmmmm.....Being from Texas, I seem to remember that Ross Perot was
a big investor in NeXT. If that's true, maybe Jobs might get some
large government contracts after Jan 1 ;-)
--
"Is this the party to | David M. Senseman (sens...@lonestar.utsa.edu)
whom I am speaking?" | Brain Research Laboratory
--Ernestine | The University of Texas at San Antonio
Jonathan Dursi
> In article <Br4Hr...@impch.imp.ch> pa...@imp.ch (Patrick Guelat) writes:
> > In article <n2m...@zuni.esd.sgi.com>, je...@soul.esd.sgi.com (Jeff
Smith)
> writes:
> > I hope that SGI will NEVER buy it. NeXTStep is always praised as the
> ultimate,
> > great UI. Well it has some nice features but all in all I prefer a REAL
> window
> >
> > It's terrible to have these menues [sic] hanging around in front of all
windows,
> > the networking capabilities of NeXtStep [sic] are terribly primitive.
> > Workspace and toolchest you have everything you need, what do you want
more ?
> >
> > --
> > Patrick Guelat, Improware AG Switzerland, CH-4414 Fuellinsdorf
> > Internet: pa...@imp.ch
>
>I'm really getting tired of hearing these unsubstantiated pokes against NeXT
>and NeXTstep.
And I'm just constantly amused by people like Patrick who post
things like that in advocacy groups, where they must surely realize
that it's just going to generate flamage.
I doubt you'll get an intelligent response, Kenny. And to all you
people who have *nothing* better to do than post flame bait in
*.advocacy groups - find something more productive to do with your
time.
Mildly irritated, but mostly bemused
- Jonathan
----------------------------------------------------------------------
Jonathan Dursi | "Never attribute to malice
Programmer, Physics Dept. | what can adequately be explained
du...@Clavius.StMarys.CA | by stupidity." - Hanlon's Razor
Jim Barton
Perhaps. But you didn't read the whole article. Guy Tribble left after the
software engineering community took a "vote of no confidence" in him as
a leader.
It is never true that one person makes a product. You may be insulting
the people left behind.
Of course, at SunSoft, they aren't letting him manage any people either ...
-- Jim Barton
Silicon Graphics Computer Systems
j...@sgi.com
Darcy Brockbank
You can't even grant him that. If he's so useless that he can't move the
menues (sic) with Preferences to a place where they're not visible, then
he deserves to have menues (sic) "all over the place."
Anyway, you're right. I'm sick of people saying, "Oh, it's just like the
Macintosh." or "I want a REAL windowing system."...
Sigh.
- db
(hostile from using System V all week).
--
-------
I have my own phone here at work, a see-through ruler and my own mechanical
pencil. I'd hope that is the definition of success (I don't want to look
it up in a dictionary though).
-- Hugh Richards
Scott Byer
> IRIX isn't a system V kernel in many respects. Does
> NeXTstep really need some Mach specific features (seems
> unlikly if they are porting to Intel systems)?
Remember, NeXTstep 486 includes Mach.
NeXTstep needs messaging so bad that it hurts.
Scott Byer NeXTMail: by...@adobe.com
Adobe Systems Incorporated These are *my* opinions, and
1585 Charleston Road, P.O. Box 7900 do not necessarily reflect
Mountain View, CA 94039-7900 the opinions of my employer.
---------------------------------------------------------------------
Jim Stukas
> (Nathan F. Janette) writes:
> In article <1992Jul10.1...@cs.mcgill.ca> sam...@uriel.cs.mcgill.ca
> > In article <1992Jul10....@kakwa.ucs.ualberta.ca>
> > kenny...@mts.ucs.ualberta.ca (Kenny Leung) writes:
> > > I grant you that you hate to have menus floating on top of everything.
> >
> > You can't even grant him that. If he's so useless that he can't move the
> > menues (sic) with Preferences to a place where they're not visible, then
> > he deserves to have menues (sic) "all over the place."
>
> kludge to get around the fact that there isn't a
> preference option to turn-off main menus.
I'll admit that it is a kludge, and maybe this should be a preference. But
actually I've gotten quite used to having the name of the application down in
the bottom right hand corner in my screen, so I always know which app is the
"key" app.
Every once in a while, I drag the menu up to click on too - whenever my right
mouse button finger gets tired. :-)
Jim
--
Jim Stukas (219) 239-8292
User Services Bitnet: jstukas@irishvma
Office of University Computing internet: stu...@mbolo.cc.nd.edu
University of Notre Dame NeXT mail cheerfully accepted
Scott Hess
>was a big investor in NeXT. If that's true, maybe Jobs might get
>some large government contracts after Jan 1 ;-)
Doubtful. He only has a piddling $20 million or so invested in
NeXT, out of the $3.3billion he has. He has much more money (orders
of magnitude more) invested in government bonds and municipalities.
Besides, I have no doubt that the government is _already_ probably
one of NeXT's biggest customers, in the guise of the Invisible
Agency With a Three Letter Name.
Later,
--
scott hess <sh...@ssesco.com> <Who's now in a programming Berserker-frenzy>
12901 Upton Avenue South, #326 Burnsville, MN 55337 (612) 895-1208 Anytime!
<I want to become so famous that people buy tapes of me reading source code>
Sean Luke
>Of course, at SunSoft, they aren't letting him manage any people either ...
What? Could you elaborate on this, please?
David M. Senseman
who grew up using a NeWS-based 4D/80GT at the lab, thinks the
NeXT's icons look "babyish". I use a NeXT now and then but the
resolution of the stupid mouse is so bad that it drives me nuts!
BTW, the guys in the NeXT lab claim that NeXT is about to release
a dual cpu machine based-on the Motorola 88000 risc chip set.
If that's true, Jobs is more wacked-out than I thought. Motorola
is having trouble with its IBM/Apple/Motorola PowerPC work
from I read, so what kind of resources can it put behind 88000
development? Is there any validity to this dual 88000 rumor
or will the next NeXT machine sport an Intel P5 (586?)? Inquiring
minds want to know ;-)
Brad Nelson
> who grew up using a NeWS-based 4D/80GT at the lab, thinks the
> NeXT's icons look "babyish". I use a NeXT now and then but the
> resolution of the stupid mouse is so bad that it drives me nuts!
I do sincerely hope that NeXT doesn't base too many of its OS
decisions on what an 8 year old thinks.
Maybe you should learn about the mouse settings in Preferences?
> BTW, the guys in the NeXT lab claim that NeXT is about to release
> a dual cpu machine based-on the Motorola 88000 risc chip set.
> If that's true, Jobs is more wacked-out than I thought. Motorola
> is having trouble with its IBM/Apple/Motorola PowerPC work
> from I read, so what kind of resources can it put behind 88000
> development? Is there any validity to this dual 88000 rumor
> or will the next NeXT machine sport an Intel P5 (586?)? Inquiring
> minds want to know ;-)
The "Dual 88110 NeXT" rumor is exactly that, a rumor. Note that
it's an 88110 rumor, not 88000. The 88110 should be faster by a
substantial margin than Intel's P5, even if you choose to believe
Intel's (usually) wildly exaggerated claims of what its chips will do.
It's my understanding that the "trouble" Motorola is having with
the Power PC venture has much more to do with the politics of their
corporate relationship with Apple and IBM than anything else. It's
also a much younger enterprise than the 88xx0 effort. There was a
report (rumor?) on the net a few months back that every new Ford
vehicle will have an 88110 on board in a year or two. Sounds like
a fairly strong commitment of resources to me. Besides, Motorola's
involvement with the Power PC has nothing to do with the 88xx0. Using
your logic, the Ford Mustang should have failed because the Edsel did.
Adoption of the Intel line of processors by NeXT would be a step
sideways, not ahead. If NeXT is going to change processors, some
sort of RISC chip is the only thing that makes sense, IMHO. The
88110 is probably as good a choice as any.
Marc Albert Ullman
>You have to admit that hiding the menus is a horrible
>kludge to get around the fact that there isn't a
>preference option to turn-off main menus.
Amen. I've been complaining about that for years. Anybody know if they
fixed this in 3.0? Figuring out how to get rid of those "damn" menus was
a big step toward making NeXTstep a useable window system.
--Marc Ullman
Stanford University
Andrew Loewenstern
As far as I know, Bud Tribble was not instrumental in the design of
NeXTSTEP. I had never even heard of him till the announcement.
Regardless, NeXTSTEP was not 'designed' by one person, or even 10
people. Many people at NeXT (and there are a LOT of really sharp
people in the software engineering department at NeXT) contributed
important things.
NeXTSTEP is way too complicated a beast to be created by a single
person even though it may have originally started as the vision of 7
people in 1985. If it weren't, Sun, Apple, IBM, and Microsoft would
all have something comparable now instead of in maybe 3 years.
andrew
--
and...@cubetech.com
Andrew Loewenstern | "If I am not for myself, who will be for me?
Cube Technologies, Inc. | If I am only for myself, who am I?" -Hillel
Andrew Loewenstern
>I grant you that you hate to have menus floating on top of everything.
Actually, that's so you don't loose them. However, you can pretty
easily set the menu to appear off-screen (or very nearly so) so they
stay out of your way (then you can use the other mouse button to pop
up the menu...).
Dave Griffiths
>In article <55...@aquarius.cs.nps.navy.mil> zy...@cs.nps.navy.mil (Michael Zyda) writes:
>>Aren't all these questions about neXTStep on the IRIS somewhat
>>academic now seeing how the designer of neXTStep is not even
>>with neXT anymore but with Sun Microsystems? Don't we all
>>expect neXT to blow away any day with people bailing out
>>like their key software designer?
>
>As far as I know, Bud Tribble was not instrumental in the design of
>NeXTSTEP. I had never even heard of him till the announcement.
>Regardless, NeXTSTEP was not 'designed' by one person, or even 10
>people. Many people at NeXT (and there are a LOT of really sharp
>people in the software engineering department at NeXT) contributed
>important things.
>
>NeXTSTEP is way too complicated a beast to be created by a single
>person even though it may have originally started as the vision of 7
>people in 1985. If it weren't, Sun, Apple, IBM, and Microsoft would
>all have something comparable now instead of in maybe 3 years.
>
Coherent pieces of software are not designed by large committees. There must
have been two or three key people responsible for the design philosophy.
Stop beating about the bush and name the guilty men! (*) (Or have they
left too? ;-)
Dave Griffiths
(*) PC disclaimer: men=people
Andrew Loewenstern
>In article <1992Jul10....@ringer.cs.utsa.edu>, sens...@ricky.brainlab.utsa.edu (David M. Senseman) writes:
>|>
>|> Hmmmmmm.....Being from Texas, I seem to remember that Ross Perot was
>|> a big investor in NeXT. If that's true, maybe Jobs might get some
>|> large government contracts after Jan 1 ;-)
>|>
>
>time ago in NextWorld that it was no longer the case. I am not sure whether
>he is still a big investor (I don't think so).
He is no longer on the board, but he still is a very good friend of
Steve Jobs... He still owns a fairly large percentage of NeXT due to
his early involvment. Steve Jobs and Canon own more than him, though,
but I think his share is still around 7% I believe...
If he were to become President, it might actually be bad for NeXT to
get some large government contracts. Many people would cry 'foul.'
Andrew Loewenstern
>BTW, the guys in the NeXT lab claim that NeXT is about to release
>a dual cpu machine based-on the Motorola 88000 risc chip set.
>If that's true, Jobs is more wacked-out than I thought. Motorola
>is having trouble with its IBM/Apple/Motorola PowerPC work
>from I read, so what kind of resources can it put behind 88000
>development?
GRRRRR I **hate** it when people starting thinking that Motorola is a
small processor design company like MIPS.
Motorola is a HUGE multibillion dollar company that makes everything
from seconductors to cellular phones. In fact, it was the largest
semiconductor company in at least the U.S. until very recently (and
they are still the second largest).
However, it really doesn't matter which processor is in the box unless
you code in assembler (which is appropriate for some situations, but
few these days and mostly in imbedded stuff).
Andrew Loewenstern
>Besides, I have no doubt that the government is _already_ probably
>one of NeXT's biggest customers, in the guise of the Invisible
>Agency With a Three Letter Name.
I think there are actually three 3-letter agencies who like NeXT
computers and they are most likely the three top purchasers of NeXT's
(in the 4 digit range)... I know the DOD has a bunch (see some of
those guys on the net every once and a while). The CIA must have a
lot (anyone notice all those people at NeXTworld Expo from McLean,
Virginia and places like that with weird company names?). I would
also assume that the NSA has quite a few of them. Of course, they
don't tell anyone anything unless they can absolutely avoid it.
There are a few government contractors who have a lot. TASC has quite
a few along with Mitre.
Laurence James Edwards
|> I know that icons do make make an os, but even my 8 year son, Trevor,
|> who grew up using a NeWS-based 4D/80GT at the lab, thinks the
|> NeXT's icons look "babyish". I use a NeXT now and then but the
|> resolution of the stupid mouse is so bad that it drives me nuts!
|>
The mouse mapping on the NeXT can be set to just about anything you want ...
although you do have to do some digging in the doc's to figure out how to
set it to something other than the 3 default settings.
NeXT's icons are certainly no worse than SGI's ... as far as style go
I'd say the Mac's icons are the clear winners.
Larry Edwards
edw...@sunrise.stanford.edu
John Mashey
>In article <1992Jul11.1...@ringer.cs.utsa.edu> sens...@ricky.brainlab.utsa.edu (David M. Senseman) writes:
>>BTW, the guys in the NeXT lab claim that NeXT is about to release
>>a dual cpu machine based-on the Motorola 88000 risc chip set.
>>If that's true, Jobs is more wacked-out than I thought. Motorola
>>is having trouble with its IBM/Apple/Motorola PowerPC work
>>from I read, so what kind of resources can it put behind 88000
>>development?
>GRRRRR I **hate** it when people starting thinking that Motorola is a
>small processor design company like MIPS.
>Motorola is a HUGE multibillion dollar company that makes everything
>from seconductors to cellular phones. In fact, it was the largest
>semiconductor company in at least the U.S. until very recently (and
>they are still the second largest).
Perhaps some facts are in order, to avoid apples-to-oranges comparisons.
1) Without knowing any of the details, I think that the rumored NeXT
machine was to use a pair of 88100s, not 88000s. The 88100 seems like
a pretty nice design ....
a) It was described well by Keith Diefendorff at Microprocessor
Forum, November 1991.
HOWEVER:
b) To my knowledge ( and correct me if I'm wrong), no one has yet:
-announced a system product based on the 88100
-demoed such a product in a public event
MUCH LESS:
- delivered production systems based on it.
[On the other hand, MIPS R4000s, which you'd think were no more than a
few months ahead, began being seen in demoes in late Summer, and shipping
in production systems (Crimsons) in March 1992.] R4000s compete
pretty well with anything else out there you can buy in systems...
2) Consider the interval from chip tapeout until production systems.
A good rule of thumb, for an aggressive, brand-new design, but which does
have useful existing software in the family, but where things are
going OK, is 10-12 months from tapeout to production (virtual-memory)
systems. The 88000 was announced in April 1988, and it was late Summer
of 1989 before the bug cleanup was OK enough to ship production systems
(i.e., the FP exception bugs, I think).
Note, also, that it sometimes takes longer to achieve a desired clock rate
than expected. This certainly happend with 88000, where it took quite a
while to get to 33Mhz; 68040s likewise.
Finally, based on (for example, SPEC89 data), the fastest 88000 numbers
are still lower than the fastest R3000 numbers...
3) Good leading-edge chip designs require:
- an experienced, world-class design team
- good set of tools and big computing resources for design
- good input from close-coupled software folks at key times
- good semiconductor processes to aim at
Now, the surprise is that MIPS/SGI has:
- 4 very good chip design teams in-house
(R4000/R4000A/R5000; VRX; TFP; T5)
+ our various semiconductor partners have a bunch of their
own doing new or derivative designs. The last I counted,
there were something like 10 different R3000-variant designs
on the market in addition to the baseline R3000s. (How many
88000-variants are there? 0, I think....)
+ Intel may have as many teams to cover the space,
and *maybe* IBM+Moto.
+ good leading-edge teams don't grow on bushes; you do not
materialize one just by printing money...
- good tools ... although I'm not sure of the comparison with IBM+Moto,
even MIPS by itself has often had *more* compute power actually
available to the engineers (on R4000 for example) than
was usually available to the same sort of group at Intel
and maybe Motorola. (yes, I know this seems counter-intuitive;
on the other hand, neither Intel nor Motorola were themselves
actually building the right sorts of big fast servers to
use for some of the simulations. The R4000 team had,
just before tapeout, something like ~2000 vax-mips worth of
computers, including a dozen RC6280 ECL 50-mips servers
cranking away, simulating UNIX boots at the RTL-level.)
- good input from close-coupled software folks: yes.
- access to good semiconductor processes: yes.
(BTW: designing & building fabs is where the big money
really is, not in the microprocessor design itself.
Leading edge micro ~ $30M; big new fab ~$400M - $500M.
4) Finally, just because a company is *huge* doesn't mean that it
can afford to throw money at everything, and dumb people don't build
big companies like Motorola. Inside large companies, there are *always*
priority calls going on with regard to money and other resources.
As the posting notes, they do build a *lot* of different things....
Now, inside Motorola, somebody has to make a priority call among:
a) 68060 & 68K derivatives, which are losing market share in
systems, but do have a large base, and will likely keep a big chunk of
that in embedded control.
b) PowerPC, which IBM is providing substantial resources to,
and which both IBM and (the major big desktop desktop customer of
68Ks) wish to buy in reasonable numbers.
c) The 88100, whose design-in numbers don't seem too high.
If you were a smart Moto executive, with concern for Return-On-Investment,
what would you do? Moto is responsible - I'd expect they'll live up to
their commitments, but I find it hard to believe that the 88100 will be
rushed out at the highest-priority, getting wafer starts before the other
efforts, etc.
5) When you compare MIPS with Motorola, you have an apples-to-oranges
comparison; you have to compare MIPS Technology, Inc + relevant semiconductor
partners' efforts to Motorola-microprocessor business (not counting
cellular, etc). I.e., MIPS itself is just the tip of the iceberg.
>However, it really doesn't matter which processor is in the box unless
>you code in assembler (which is appropriate for some situations, but
>few these days and mostly in imbedded stuff).
Well... I'd assume, that if they switch CPUs, they'll do everything
possible to make it easy to recompile, and this is certainly easier
than it used to be. On the other hand, unless they do some binary
translation scheme, a non-68K NeXT won't start life with a whole ton of
software, and you have to wonder where it would be on the vendor port list,
even as good as some of that software is.
SUMMARY: it is a mistake to assume that all it takes to do
leading-edge microprocessor design is a big company with lots of money.
Sometimes smaller companies, or combinations of companies, do just as
well, or even better .... and there's a lot more resource behind the
MIPS designs than you'd ever expect (thank goodness :-)
--
-john mashey DISCLAIMER: <generic disclaimer, I speak for me only, etc>
UUCP: ma...@mips.com [soon to be ma...@sgi.com, but not quite moved yet].
DDD: 408-524-7015, or 524-8253
USPS: (soon) Silicon Graphics, 2011 N. Shoreline Blvd, Mountain View, CA 94043
Royce Howland
>[...stuff about Motorola...]
>1) Without knowing any of the details, I think that the rumored NeXT
>machine was to use a pair of 88100s, not 88000s. The 88100 seems like
>a pretty nice design ....
> a) It was described well by Keith Diefendorff at Microprocessor
> Forum, November 1991.
> HOWEVER:
> b) To my knowledge ( and correct me if I'm wrong), no one has yet:
> -announced a system product based on the 88100
> -demoed such a product in a public event
> MUCH LESS:
> - delivered production systems based on it.
>[...lots of stuff about Motorola & MIPS...]
Data General is probably the world's leading user of the 88K series for
computer system purposes. They have definitely announced and demoed 88100
products (up to & including quad-processor servers), and as far as I know,
these products have been shipping for some time.
But a good post on Motorola vs. MIPS, nonetheless.
--
Royce Howland -- Everything is IMHO | "I can eat enormous quantities
ro...@splunge.uucp (NeXTMail OK) | of ice-cream without being sick,
or royce%spl...@atlantis.uucp | Mrs. S-C-U-M." --Mr. Neutron
or alberta!atlantis!splunge!royce or ... | C86Net: Mr._N...@RT.Alta
Bryan Strickland
<bunch of other stuff deleted>
>Well... I'd assume, that if they switch CPUs, they'll do everything
>possible to make it easy to recompile, and this is certainly easier
>than it used to be. On the other hand, unless they do some binary
>translation scheme, a non-68K NeXT won't start life with a whole ton of
>software, and you have to wonder where it would be on the vendor port list,
>even as good as some of that software is.
>
John, in the 90's, most people are using high level programming languages
to write commercial apps. Porting is as simple as recompiling on the new
machint does not matter what hardware is running, the important thing
is that the operating system,unning on top of this hardware looks the same
to the application programs. The port really becomes porting the O.S. (with
a few exceptions, like compiler backends) to the new hardwareThis is done by
NeXT, this would undoubtedly include the cc compiler. Therefore, the
number of apps, from day one of a new introduction (after a couple hours
of recompiling) would be almost the same as it is now.
bryan
Michael Woodacre
So do you have the source code to all the software you run?
Try counting up the number of applications you run and then see
how much source code you have for them.
--
Michael S. Woodacre | Phone: (408) 524-8120
UUCP: {ames,pyramid,decwrl}!mips!woodacre OR wood...@mips.com
USPS: MIPS Technologies Inc, 928 E. Arques, Sunnyvale, CA 94086
Bryan Strickland
>So do you have the source code to all the software you run?
>Try counting up the number of applications you run and then see
>how much source code you have for them.
>
>--
> Michael S. Woodacre | Phone: (408) 524-8120
>UUCP: {ames,pyramid,decwrl}!mips!woodacre OR wood...@mips.com
>USPS: MIPS Technologies Inc, 928 E. Arques, Sunnyvale, CA 94086
>
When you buy an application program for a computer, in almost all cases you
are purchasing it for one machine, or a set of specific machines.
The argument here is not, can I take
my diskette, and copy the software onto my new machine, and have it run
(which is illegal by the way), but rather, will the number of software
packages offered by venders, for my new machine, be the same as that for
the old machine. My point was that it would be the same, since all the
venders have to do is recompile the source. Yes guys, you would have to
purchase the new software for the new platform, but the amount of software
offered would be just as abundant (if we can call the current offering of
NeXT software abundant).
bryan
Nathan F. Janette
writes:
> Well, I didn't wanna get involved in this, but...
I can see why - you have a poor understanding of many "facts"
you claim below.
> I have used Next computers at Brown... as a matter of fact,
> I was using one with pretty tiny serial number, if memory
> serves me correctly. NeXT machines are, well,
> interesting... but that's about it. Also, having been in
Those early NeXTs were 68030 machines, many of which were
8 MB RAM machines that didn't have a hard disk as a primary
storage device. The hardware didn't have the
performance necessary to support the demands of the
NeXTstep OS/GUI.
This has since changed - NeXT uses 68040 (25 and 33
MHz) CPU machines now, with other hardware improvements
which support NeXtstep very nicely. Unfortunately,
some people are content to judge the current platform
solely upon old impressions.
> both the planetary and meteorology research
> communities, I can tell you that no one I have ever
> encountered in these fields uses them...from the
> literature I have been reading about them, I think that is
> true for most scientific fields. Although they are used
> in computer science circles, their use is generally the
> exception, and not the rule.
There are some in the Astronomy department in my building,
not to mention math, physics, mol bio, etc. There are
some universities that have hundreds of NeXTs on campus.
> By the way, Switzerland is not the only place where NeXTs
> are too expensive. I was *at* Brown when Mr. Jobs waggled
> his little butt up on stage and answered the question:
> "What does someone do to get a NeXT if they aren't
> connected to a University?"
> with the haughty:
> "Enroll!"
>
> About 9 months of red ink must have convinced him that his
> whizzo back-of-the-napkin financial analysis that
> Universities were "this country's secret Fortune 500
> companies" was slightly in error.
This isn't about whether Steve Jobs is or isn't a popular
God, or whatever. He definitely has had some half-baked
ideas. The NeXT is not one of them.
As for NeXT pricing, it is quite competitive for both business
and educational customers. One example: NeXTstation mono Turbo
(33 MHz 68040, 8 MB RAM, 250 MB HD, 2.88 MB FD): $5,995 list,
$4,300 educational around here, 17" display, keyboard, mouse
included.
> I also remember Mr. Jobs promising *color* NeXTs in his
> ads and promos *long* before they materialized, I also
Support this vague claim.
NeXTstation color machines (16-bit color), by the way,
will support Pantone color matching and drive the
Cannon/NeXT color printer with the release of NeXTstep
3.0 in August. WYSIWYG for color as well as postscript!
> remember Mr. Jobs thinking that optical drives were such
> a neat idea that NeXT machines would use them
> exclusively,
Agreed, a poor idea. They are just fine as secondary
storage units.
> I also remember Mr. Jobs thinking that
> laser printers (really expensive, propriatary laser
> printers) were the only hardcopy device a computer
> engineer might need, *AND* I also remember Mr. Jobs doing
The NeXT Laser Printer sells for $1,795 (list), $1,395 (edu),
and offers 8 PPM, 400 DPI printing. It rips postscript with
the host CPU, and therefore is "upgraded" whenever the CPU
is upgraded. I believe it was the first 400 DPI laser printer
widely available. The printing quality of 400 DPI can be
noticeably better than 300 DPI standard printers.
NeXT has always supported other printing devices connected
to serial ports. A 3rd party vendor offers a program to
drive many of the most popular other printers.
> his best to keep NeXTStep on the NeXT machines only.
> Financial reality (as well as practicality) caught up
> with *all* of the above ideas...
NeXT does not appear to have any financial problems,
they seem to be doing fine with their current plan.
The port to Intel CPUs is a good idea, and I'll bet the
first of many ports.
> There are enough reasons to not take NeXTStep too
> seriously besides the religious X vs. NeXT wars. X isn't
> perfect, but as you put it:
>
> > X talks to X only, NeXTstep talks to NeXTstep only. The >
> only difference is that there exist more X machines.
>
> Very true. Even if NeXTStep was a billion times better
> than X (which I don't think it is)...if you are a corporate
> developer, which machine would you pick for your target
> market? In addition, if you are an academic researcher
> with only $50,000 in your grant, and you need to pass your
> results and programs off to other researchers in your
> field, which one would you get?
There is no problem running X on NeXT machines, if that's
what you really want to do.
There is no problem using NeXT machines in heterogeneous computing
environments.
> The moral of the story is that where you are is *not* the
> rest of the world...that applies to Brown as well as
> Switzerland.
The moral of the story is get your facts straight.
--
Nathan Janette "I'm a NeXTstep man,
Dept MB&B, Yale Univ I'm a NeXTcube guy"
New Haven, CT
nat...@laplace.biology.yale.edu (NeXT)
Michael Gold
>John, in the 90's, most people are using high level programming languages
>to write commercial apps. Porting is as simple as recompiling on the new
>machint does not matter what hardware is running, the important thing
>is that the operating system,unning on top of this hardware looks the same
>to the application programs. The port really becomes porting the O.S. (with
>a few exceptions, like compiler backends) to the new hardwareThis is done by
>NeXT, this would undoubtedly include the cc compiler. Therefore, the
>number of apps, from day one of a new introduction (after a couple hours
>of recompiling) would be almost the same as it is now.
Ideally, maybe, but not in the real world. Supporting another version
of a product requires resources, including QA, documentation,
manufacturing, packaging, and customer support. This also assumes
that no bugs will be found in the ported OS and more importantly, the
compilers. Yes, NeXT and Motorola test their compilers, but believe
you me, software companies find *lots* of OS/compiler bugs.
So if a software house exec has to make a call, s/he might look at the
relatively small installed base, and the struggles NeXT has faced, and
decide to wait before porting.
--
Michael I. Gold You go your way, I'll go mine,
Silicon Graphics Inc. I don't care if we get there on time,
go...@sgi.com Everybody's searching for something they say,
(415) 390-1709 I'll get my kicks on the way...
Vernon Schryver
> In <l638ql...@aludra.usc.edu> stri...@aludra.usc.edu (Bryan Strickland) writes:
> >John, in the 90's, most people are using high level programming languages
> >to write commercial apps. Porting is as simple as recompiling on the new
> >machint does not matter what hardware is running, the important thing
> >is that the operating system,unning on top of this hardware looks the same
> >to the application programs. The port really becomes porting the O.S. (with
> >a few exceptions, like compiler backends) to the new hardwareThis is done by
> >NeXT, this would undoubtedly include the cc compiler. Therefore, the
> >number of apps, from day one of a new introduction (after a couple hours
> >of recompiling) would be almost the same as it is now.
>
Exactly.
Some of us still at Silicon Graphics were around for the change from
68000's to MIPS chips. (Similar to going from 68000's to 88K's?) Many
of us were around for the change from NeWS to X. Most of us have been
around for other less traumatic changes. All were supposed to be trivial
given source for the applications.
Somehow, it never quite turns out to be "trivial". Somehow, "a couple
of hours of recompiling" is not quite enough, even if you happen to
have fast machines and faster disks for the recompiling. And if you
just happen to have convinced all of your third party Value Added
Resellers and application vendors that you can be trusted with their
source. That last point is the killer--you just don't force your VAR's
and the important application houses to give you source, at least if
you're no bigger than SGI and Sun--it may be different for IBM.
Yes, such conversions get done, but no one, from the appplication
vendors to anyone at the computer vendor, enjoys the months of supid,
mind-numbing, brain-off work. Of course, in really big companies,
unlike relatively small ones like SGI and Next, it's a matter of years,
not months. (Is Next about the size of SGI in 1986, during the
68000-MIPS conversion?)
There are always a few ignorant polyannas in the company that predict
it will be trivial, but fortunately, there are always enough wet
blankets to add a touch of realism.
Vernon Schryver, v...@sgi.com
Ron Pomeroy x(Coop)
sens...@ricky.brainlab.utsa.edu (David M. Senseman) writes:
The PowerPC is a first-go-round for Moto. New projects of this magnitude are
bound to have problems [good thing, keeps engineers working :-)]. The 88110 is
continuing work (from the 88000) and was already in-the-works long before
PowerPC. Moto is a pretty damn big company - they make a lot more than
microprocessors. My perception is that they have the resources to support both
efforts. Also, if the P5 is reverse compatable with the 486 we'll get it for
free since NeXT is porting to the 486 architecture.
--
Ronald Pomeroy | NeXT, It's a black thang' - you wouldn't
Advanced Micro Devices | understand.
CAM Applications Group |(NeXTMail happily accepted)
rpom...@aunext3.cam.amd.com | Disclamer: I con't speek fer any1 :-)
Bryan Strickland
>Exactly.
>
>Some of us still at Silicon Graphics were around for the change from
>68000's to MIPS chips. (Similar to going from 68000's to 88K's?) Many
>of us were around for the change from NeWS to X. Most of us have been
>around for other less traumatic changes. All were supposed to be trivial
>given source for the applications.
>
>Somehow, it never quite turns out to be "trivial". Somehow, "a couple
>of hours of recompiling" is not quite enough, even if you happen to
>have fast machines and faster disks for the recompiling. And if you
>just happen to have convinced all of your third party Value Added
>Resellers and application vendors that you can be trusted with their
>source. That last point is the killer--you just don't force your VAR's
>and the important application houses to give you source, at least if
>you're no bigger than SGI and Sun--it may be different for IBM.
>
>Yes, such conversions get done, but no one, from the appplication
who said anything about source code??? The applications venders are the only
ones with the source code. They are basically porting it to the new machine,
which, as I stated earlier, envolves basically recompiling the source,
provided that the OS on top of the new hardware has the same interface
to the high-level language that the old machine had. If NeXT (rather when)
next comes out with a new machine based on another processor, I would hope
that the system level interface would look the same, if it doesn't, than
it isn't done right. For instance, the NeXT machine uses a Mach kernal,
however, the interface to this OS is also bsd4.3 compatible, now why do you
think that was done?
Also, at NeXTWorld Expo, Jobs showed NeXTStep running on a 486 machine, and
made comments on how several software packages were ported to the new
machine in a several hours. I would say that the 486 and 680xx are fairly
different beasts.
I'm sorry if you think I am over simplifying this, but when a company
introduces a line of machines, based on another architecture, these changes
should be well hidden from the application programmers. If they are not
then the company screwed up. If porting software from the old SGI
machines to the new machines was painful for the applications venders,
SGI screwed up. period.
bryan
Nick Kline
...
>also a much younger enterprise than the 88xx0 effort. There was a
>report (rumor?) on the net a few months back that every new Ford
>vehicle will have an 88110 on board in a year or two. Sounds like
>a fairly strong commitment of resources to me. Besides, Motorola's
>involvement with the Power PC has nothing to do with the 88xx0. Using
>your logic, the Ford Mustang should have failed because the Edsel did.
Ford decided that they would put off adoption for one year on the 88k
and think about out chips. I don't know of any major people using
the 88k, except for data general.
nick
Michael Gold
>Also, at NeXTWorld Expo, Jobs showed NeXTStep running on a 486 machine, and
>made comments on how several software packages were ported to the new
>machine in a several hours. I would say that the 486 and 680xx are fairly
>different beasts.
>I'm sorry if you think I am over simplifying this, but when a company
>introduces a line of machines, based on another architecture, these changes
>should be well hidden from the application programmers. If they are not
>then the company screwed up. If porting software from the old SGI
>machines to the new machines was painful for the applications venders,
>SGI screwed up. period.
Geez, not to get into a flame war, but have you ever ported software
from one architecture to another? Does the expression "byte order"
mean anything to you? Did you know that not all machines agree on
sizeof(int)? Did you know that some machines pad structures, and
others do not? Do you know what the phrase "unaligned access" means?
Did you know that some inner loops are written in assembler for
optimal performance?
A common API will not fix ANY of these issues, which may be sprinkled
throughout the code. Not so much if the code is written in a portable
manner, but guess how many people write good portable code. Bzzzz,
wrong answer, try 1/3 that number. Also, lots of applications
inadvertantly count on bugs in the OS/compilers. Finally, you seem to
have totally ignored the points in my previous message.
Clearly there are some applications that will come across cleanly if
they are well written, and if the vendor chooses to support an
unproven market (i.e. the next-generation NeXT machines), possibly at
the expense of a larger market (i.e. the current installed base).
Remember, small companies often have limited resources. And is NeXT
going to supply each and every software vendor with several new
machines at no cost; for engineering, SQA, and customer support? If
you believe that all applications on the old machine will immediately
be available on the new system, before any end-user systems have even
shipped, you're only fooling yourself.
Thomas Insel
>2) What networking capabilities is NeXTstep missing?
Well, if I'm running 4dwm, and I want to pop up a program from another
computer, I can just use rsh and the -display flag. If I'm on a NeXT, and
I want to bring up a window from some mainframe, I have to stop what I'm
doing, startup X, etc. And as far as I can tell, my local programs (NeXT) and
my X windows, can't coexist on the screen at once.
--
Thomas Insel (tin...@nyx.cs.du.edu, tin...@uiuc.edu)
Chris King
> I do sincerely hope that NeXT doesn't base too many of its OS
>decisions on what an 8 year old thinks.
This is exactly the aproach that was taken by the guys at Xerox PARC.
They were trying to build a computer that could be easily used by 8
year olds and even younger.
Chris King
Geoff Brunkhorst
> machine was to use a pair of 88100s, not 88000s. The 88100 seems like
> a pretty nice design ....
> a) It was described well by Keith Diefendorff at Microprocessor
> Forum, November 1991.
> HOWEVER:
> b) To my knowledge ( and correct me if I'm wrong), no one has yet:
> -announced a system product based on the 88100
> -demoed such a product in a public event
> MUCH LESS:
> - delivered production systems based on it.
go out and buy a NCD 19c Xterminal. You will then have a delivered
system based on the 20mhz 88100. Granted, it's not Unix, but you didn't ask
for that. Been shipping for months (March 1992 first revenue ship).
Demoed in January at UniForum.
It'll even run against a SGI box ;-).
And, yes, at the same time, NCD ships a 19r (monochrome) Xterminal, based
on the less costly R3000. The 19c (color) needed more bandwidth and power,
so they used the 88100. Why they didn't use a R4000, I don't know.
And yes, MIPS and Motorola are completely different operations. Comparing
the two is rediculous.
> -john mashey DISCLAIMER: <generic disclaimer, I speak for me only, etc>
- Geoff
--
"The only thing obsoleted by Windows 3.1 is poking myself in the
eye with a pointed stick." - Gene Lee
---------------------------------------------------------------
Geoffrey Brunkhorst, RCF brunk...@Mayo.edu
Guggenheim 10 (507) 284-1805
Mayo Foundation, Rochester MN, 55905 USA fax (507) 284-5231
Barry Merriman
> And as far as I can tell, my local programs (NeXT) and
> my X windows, can't coexist on the screen at once.
> --
> Thomas Insel (tin...@nyx.cs.du.edu, tin...@uiuc.edu)
Gee---mine can. I guess you don't have Co-Xist from Pencom, Inc.
--
Barry Merriman
UCLA Dept. of Math
UCLA Inst. for Fusion and Plasma Research
ba...@math.ucla.edu (Internet; NeXTMail is welcome)
Thomas Funke
(Ronald C. Antony) writes:
> >It's terrible to have these menues hanging around in front of all
windows,
> >the networking capabilities of NeXtStep are terribly primitive. With
SGI's
> >Workspace and toolchest you have everything you need, what do you
want more ?
>
> Hear this person. Is living in Switzerland where NeXTs are so
> expensive that noone ever gets to touch one, but then they make
> judgements on the basis of "what is generally known to be true".
>
... well, they're a bit slow in Switzerland, don't worry. It took them
quite a while to notice that mechanical watches are less accurate than
electronic ones, so who is wondering about those remarks regarding
NeXTSTEP <-> Motif ?
:-)
--
------------------------------------------------------------------
Thomas Funke, Gasteinerstr. 29, 1000 Berlin 31, Germany.
E-mail: t...@zelator.in-berlin.de Phone: +49 30 8616224
------------------------------------------------------------------
Garance A. Drosehn
> stri...@aludra.usc.edu (Bryan Strickland) writes:
> >>I'm sorry if you think I am over simplifying this, but when a company
> >>introduces a line of machines, based on another architecture, these
>
> And go...@puck.esd.sgi.com (Michael Gold) replies:
>
> >Geez, not to get into a flame war, but have you ever ported software
> >from one architecture to another? Does the expression "byte order"
> >mean anything to you? Did you know that not all machines agree on
> >sizeof(int)? Did you know that some machines pad structures, and
> >others do not? Do you know what the phrase "unaligned access" means?
>
> These are the technical challenges. Bryan, you could start refuting
> some of these. And in a sense, I agree with you somewhat - *if* the
> applications had been written portably, then they should be a breeze
> to recompile. But some of them are genuine concerns (byte alignment
> and struct padding, for example, could affect binary structures
> written to a file or a socket).
First off, apologies to SGI readers who are probably wondering what this
whole argument is doing in the c.s.sgi newsgroup. I don't mean to inflame
any wars...
Second, NeXTSTEP for the 486 is ready (at least in the "beta-testing" sense
of the word). Some companies have already ported their applications from
the 680x0 universe to the '486 version. I don't know just how extensive the
testing has been, but there are a number of people saying that the
conversion has been fairly painless for them. On the order of a week to get
their application working on the '486 version (and these are for significant
applications, not "Hello World" done via NeXTstep). This is encouraging, I
would think.
Third, note that some of the differences mentioned above are differences in
the compiler, not the underlying hardware architecture. Padding of structs
or size(int) are compiler issues, not hardware. Byte-ordering is hardware,
of course... NeXTSTEP 680x0 vs. NeXTSTEP '486 has already crossed that
bridge though.
Fourth, while I'm mainly a NeXT bigot (well, aficionado, at least), I have
heard a lot of good stuff about SGI machines. I, like others in the c.s.n.*
newsgroups, would love to play around with an SGI machine running NeXTSTEP,
just to see what it would be like. My comments are not meant to put down
anything about SGI offerings.
If NeXTSTEP does do a better job at supporting development on multiple
hardware architectures, it will undoubtably be thanks to people who have
tried it before. I'd like to see this succeed, just because that would mean
we could separate software flamewars from hardware flamewars. I'd really
like to write programs for one software environment (whatever that is), and
have it possible to run it on many hardware architectures.
I say this as a person who used to develop for an operating system that only
runs on 370-mainframes. I don't want to see software environments run into
a brick wall just because some hardware chip (or the company making it) runs
into a dead end. While I understand the technical hurdles to that ideal, I
hope they can be addressed by software environments.
--
Garance Alistair Drosehn = g...@eclipse.its.rpi.edu
ITS Systems Programmer (handles NeXT-type mail)
Rensselaer Polytechnic Institute; Troy NY USA
Peer Landa
Laurence James Edwards writes:
> NeXT's icons are certainly no worse than SGI's ... as far as style go
> I'd say the Mac's icons are the clear winners.
I rather go for SUN's icon.
-- peer
David Lemson
> > I also remember Mr. Jobs thinking that
> > laser printers (really expensive, propriatary laser
> > printers) were the only hardcopy device a computer
> > engineer might need, *AND* I also remember Mr. Jobs doing
>
> The NeXT Laser Printer sells for $1,795 (list), $1,395 (edu),
> and offers 8 PPM, 400 DPI printing. It rips postscript with
> the host CPU, and therefore is "upgraded" whenever the CPU
> is upgraded. I believe it was the first 400 DPI laser printer
> widely available. The printing quality of 400 DPI can be
> noticeably better than 300 DPI standard printers.
You can do a head to head with a NeXT Laser Printer and any Apple
LaserWriter and find that the NeXT will produce faster results with
graphic images in every case. With the IINTX, it was usually around 2-3
times faster. I haven't tested it with a IIg with a stopwatch, but it
noticeably faster. With text, it's much less apparent.
Please don't quote educational prices - you will be wrong in every
instance, because the discounts vary so greatly from institution to
institution. In this case, many institutions have a price under $1200 for
the NeXT Laser Printer, which is a hell of a deal for that printer.
One more thing about the NeXT Laser Printer - in our labs at UIUC, we have
one that is over 3 years old and has survived plastic melting in it,
hundreds of thousands of copies, and being turned on almost all the time
it has been in service. Its design is so simple that it can hardly fail.
--
David Lemson (312) 732-4741
FNBC Sys Admin (Summer) UIUC NeXT Campus Consultant(rest of the time)
E-mail to: lem...@fnbc.com NeXTMail accepted
mark.d.wuest
>In <l63s0b...@aludra.usc.edu> stri...@aludra.usc.edu (Bryan Strickland) writes:
>>I'm sorry if you think I am over simplifying this, but when a company
>>introduces a line of machines, based on another architecture, these changes
>>should be well hidden from the application programmers. If they are not
>>then the company screwed up. If porting software from the old SGI
>>machines to the new machines was painful for the applications venders,
>>SGI screwed up. period.
>
>Geez, not to get into a flame war, but have you ever ported software
>from one architecture to another? Does the expression "byte order"
>mean anything to you? Did you know that not all machines agree on
>sizeof(int)? Did you know that some machines pad structures, and
>others do not? Do you know what the phrase "unaligned access" means?
>Did you know that some inner loops are written in assembler for
>optimal performance?
>
>A common API will not fix ANY of these issues, which may be sprinkled
>throughout the code. Not so much if the code is written in a portable
>manner, but guess how many people write good portable code. Bzzzz,
>wrong answer, try 1/3 that number. Also, lots of applications
>inadvertantly count on bugs in the OS/compilers. Finally, you seem to
>have totally ignored the points in my previous message.
Hear, hear! I was reading the post in astonishment myself. Geez, the only
*real* problem I have with Intel is their funky byte-ordering held over
from ancient processor designs.
Portable code example:
#IFDEF I86
switcheroo();
#ENDIF
;-)
Mark
--
Mark Wuest
mark....@att.com
m...@cheshire.att.com (NeXT Mail Welcome!)
Michael Portuesi
hun...@bacchus.esa.oz.au (James Gardiner [Hunter]) writes:
|> You can do some amazing things in Nextstep. All the apps
|> are built on it and can take advantage of it without the
|> programmer even trying.
|>
|> If you want this funtionality on a Indigo, you would have
|> to go redesign the servers and underlying windowing system.
|> Ie. copy nearly exactly what nextstep is. A little silly
|> when nextstep already exists.
You are incorrect. The primary advantages of NeXTStep as a
a software development environment or as a graphical user
interface have little to do with the servers or the imaging
model which it is based upon.
Those "servers and underlying window system" you refer to may
have been designed to make it convenient for the implementation
of NeXTStep, but it is by no means a requirement for a
NeXTStep-like software environment. There are public-domain
toolkits available which are fully object-oriented, support
dynamic object linking, support object transcription,
have interactive application builders, and are window system
independent to boot.
|> I would love to have nextstep intereface on my indigo.
|> In real terms a combination. IE nextstep with inbuilt
|> X11 support to keep it all very fast.
Apparently you do not have a clear grasp of the technical
issues involved. X11 on the Indigo has extremely high
performance because it talks directly to the Indigo display
hardware, and has been finely tuned and optimized to do
that in an efficient manner.
NeXTStep requires its own proprietary windowing system,
built as extensions to Display PostScript, to run.
|> If it had X11 as a crutch to lean on when it lacks in
|> application, it would have been affected by X and products
|> as good as they are now may not have beenn developed as open
|> minded as they where.
NeXT could have developed NeXTStep using X. Instead of writing
their own windowing system, they could have spent the effort on
developing extensions to X for the services they needed. The
result would have been just as good as it is now, plus it would
have the advantage of interoperability and be easier to port to
other platforms without having to port the entire operating
system it runs upon. Even companies like SGI might have been
interested in licensing it.
--
Michael Portuesi Silicon Graphics, Inc. port...@sgi.com
Scott Byer
> You are incorrect. The primary advantages of NeXTStep
> as a a software development environment or as a graphical
> user interface have little to do with the servers or the
> imaging model which it is based upon.
Actually, the imaging model has quite a bit to do with
some of NeXTstep's ease of use and programming.
True WSIWYG display is achievable only when the imaging
model for the screen and the printer match. The X imaging
model is so tied to display architechtures that printing
requires an application to contain it's own printer
driver.
Good typography and page layout is a snap given the right
imaging model. X lacks such things as fractional
character widths, arbitrary text transformation, and the
ability to get outlines back for text effects which are
typical in page layout.
Arbitrary transformations as part of the imaging model
becomes important when trying to place images or graphics
within a confined or other well-defined area.
A good color model frees applications from worrying about
such things as color table allocation, windows crossing
mono/color display boundaries, and image depth (for such
things as icons) having to exactly match screen depth.
To do these things under X is possible, but you would be
implementing an imaging model on an imaging model. Better
to just use the Display PostScript extension that's in
the SGI instead. It's not as easy or smooth as having
the windowing system / appkit built upon a good imaging
model to begin with, but it at least makes applications
like Illustrator feasable.
Scott Byer NeXTMail: by...@adobe.com
Adobe Systems Incorporated These are *my* opinions, and
1585 Charleston Road, P.O. Box 7900 do not necessarily reflect
Mountain View, CA 94039-7900 the opinions of my employer.
---------------------------------------------------------------------
Clark L. Coleman
>ma...@mips.com (John Mashey) writes:
The confusion in these postings is about product numbers. A little review:
The 88000 is a product family. The first CPU is the 88100; it has shipped for
3 years now. The 88200 was an MMU/cache controller/cache chip, or something
like that. Don't correct me; who cares.
The 88110 is the upcoming superscalar CPU. It has not shipped yet. Where John
Mashey said "88100", he meant "88110", hence the confusion.
Systems based on the 88110 have been announced by Dolphin Technologies, but
that is vaporware (pre-announcement, a la Sun and DEC and IBM.) They have
not shipped anything to customers.
The problem with Motorola is their design tools, not their wafer fabrication
facilities or their engineering talent. Witness the whole sordid tale of
the 68040/68050/68060 and their various missed deadlines and even outright
cancellations (e.g. the 68050.) Motorola is lagging the market, and that
is empirical fact, not opinion. Intel used to trail Motorola in CPU
performance; no more. They will ship 66 MHz 80586/P5 chips by the end of
the year, and clock-doubled 33/66 MHz 80486 chips in a couple of months
that will far exceed the integer performance of the NeXT boxes, and match
them or better in floating point performance.
When I talked to my boss at work about the new NeXT boxes, his immediate
reaction (about the 88110) was "Oh, you don't want to depend on Motorola
for your CPUs."
Motorola needs to get its act together or continue to become irrelevant
in the CPU business (or just be a chip house for Apple and IBM.) Who is
going to buy the 68060 when it finally comes out 100 years from now?
Big customers like HP/Apollo, Sun, Apple and NeXT all used previous 680x0
chips, and none have any interest in the 68060. Sun stopped after the
68030; the 68040 looks like the end of the road at NeXT and Apple; and
HP has already announced that it has no interest in the 68050/68060 chip
development. Single-board VME-bus computers will be their last market.
The announced 50 MHz 88110 was hinted to be about 63 SPECmarks of performance.
By the time it finally comes out, there will be systems with more than
100 SPECmarks using 100 MHz HP PA 7100, 75 MHz MIPS R4000, and 50 MHz
IBM RS 6000 chips. Even the Sparc architecture, the laughingstock of
the RISC CPU world, will do better than 63 SPECmarks this year. So, NeXT
will be trailing the pack again.
As I said, if Motorola wants to play with the big boys, it needs to get
its act together. FAST.
--
-----------------------------------------------------------------------------
"It is seldom that any liberty is lost all at once." David Hume
||| cl...@virginia.edu (Clark L. Coleman)
Sean Luke
>|> If it [the NeXT] had X11 as a crutch to lean on when it lacks in
>|> application, it would have been affected by X and products
>|> as good as they are now may not have beenn developed as open
>|> minded as they where.
>
>NeXT could have developed NeXTStep using X. Instead of writing
>their own windowing system, they could have spent the effort on
>developing extensions to X for the services they needed.
Sure. They could have set themselves afire too.
>The
>result would have been just as good as it is now, plus it would
>have the advantage of interoperability and be easier to port to
>other platforms without having to port the entire operating
>system it runs upon. Even companies like SGI might have been
>interested in licensing it.
I see. So it was a bad idea for NeXT to implement an advanced GUI on an
advanced operating system. They should have gone the Windows route. Oh.
When new technology is ever introduced, the first thing the industry says is
"it's not compatible with our old stuff". Of course: old ideas are...old.
It reminds me of the joke about the company that introduces a brand new
computer. It's 10000 MIPS, has huge memory storage, is the easiest computer in
the world to learn, comes with tons of applications, and is free. So what does
the customer say?
"Is it compatible with 1-2-3?"
--
Sean Luke
Brigham Young University MILK: It Comes From Cows
se...@digaudio.byu.edu
NeXTmail and nifty Mac stuff welcome
John Mashey
>ma...@mips.com (John Mashey) writes:
>>machine was to use a pair of 88100s, not 88000s. The 88100 seems like
>>a pretty nice design ....
>> a) It was described well by Keith Diefendorff at Microprocessor
>> Forum, November 1991.
>> HOWEVER:
>> b) To my knowledge ( and correct me if I'm wrong), no one has yet:
>> -announced a system product based on the 88100
>> -demoed such a product in a public event
>> MUCH LESS:
>> - delivered production systems based on it.
>>[...lots of stuff about Motorola & MIPS...]
>
>Data General is probably the world's leading user of the 88K series for
>computer system purposes. They have definitely announced and demoed 88100
>products (up to & including quad-processor servers), and as far as I know,
>these products have been shipping for some time.
Eeek! Thanx!. Everywhere above, please change 88100 -> 88110.
88100 is the CPU, which with 2 or more 88200 cache-mm is what was
announced in 1988, and of course they've been shipped. 88110 is the
next-generation chip described by K.M. last Fall.
Sorry for the screwup.
Michael Portuesi
Byer) writes:
|> Michael Portuesi writes
|>
|> > You are incorrect. The primary advantages of NeXTStep
|> > as a a software development environment or as a graphical
|> > user interface have little to do with the servers or the
|> > imaging model which it is based upon.
|>
|> Actually, the imaging model has quite a bit to do with
|> some of NeXTstep's ease of use and programming.
|>
|> True WSIWYG display is achievable only when the imaging
|> model for the screen and the printer match. The X imaging
|> model is so tied to display architechtures that printing
|> requires an application to contain it's own printer
|> driver.
Yes, you are correct. The X imaging model is not appropriate
for a wide range of applications (which is one reason why
we have GL).
SGI originally chose NeWS as its windowing system because
it provided a superior imaging model than X. But eventually
market pressures forced us to move to X as our primary
windowing system.
|> Good typography and page layout is a snap given the right
|> imaging model. X lacks such things as fractional
|> character widths, arbitrary text transformation, and the
|> ability to get outlines back for text effects which are
|> typical in page layout.
You are correct. However, it is my understanding that
at least some of this functionality is now being implemented
as extensions to X.
|> To do these things under X is possible, but you would be
|> implementing an imaging model on an imaging model. Better
|> to just use the Display PostScript extension that's in
|> the SGI instead. It's not as easy or smooth as having
|> the windowing system / appkit built upon a good imaging
|> model to begin with, but it at least makes applications
|> like Illustrator feasable.
Which was my point exactly. The DPS extension on the IRIS
is not as tightly integrated with the rest of the system,
but it would be possible to build a NeXTStep-like toolkit
on top of it. The same statement applies for the IRIS
GL (and indeed, we have developed a toolkit on top of it,
though it is focused only on solving problems unique to
applications involving 3D graphics, rather than being a
generalized applications toolkit).
m.
John Mashey
NOT.
I talk *often* to 3rd-party software vendors. Recompiling isn't the issue,
it's everything else: QA, distribution, support. This has been
bludgeoned into my head by *numerous* ISVs... you must talk to different
ones than I do to believe this in general. Of course, some of them have
terrific automated test suitesthat help, but some require much more work.
I am of course not under the slighest illusion that people are doing the
bulk of their code in anything but high-level langauges these days,
thank goodness.
Patrick Guelat
|> In article <Br4Hr...@impch.imp.ch> pa...@imp.ch (Patrick Guelat), I wrote:
|> >I hope that SGI will NEVER buy it. NeXTStep is always praised as the ultimate,
|> >great UI. Well it has some nice features but all in all I prefer a REAL window
|> >system like X11 with a good UI like Motif (or the derrivated 4Dwm).
|> >
|> >It's terrible to have these menues hanging around in front of all windows,
|> >the networking capabilities of NeXtStep are terribly primitive. With SGI's
|> >Workspace and toolchest you have everything you need, what do you want more ?
|>
|> Hear this person. Is living in Switzerland where NeXTs are so
|> expensive that noone ever gets to touch one, but then they make
|> judgements on the basis of "what is generally known to be true".
Well, I have a good old slow 68030-NeXT Cube in my office ... I don't have to
find someone who owns a NeXT. We have a lot of other machines, including
Suns, SGI and PCs... On all these machines I can run X-Windows, mainly
because the raw MIT X11 is free and nearly all unix manufacturers
ported it to their platform. X11 is a standard, it's not perfect, but it's
robust, thousands of applications run on top of X11 and you can get a lot of
PD Software for X-Windows. BTW: I'm also running X11 on top of MS-Windows
on the PC.
Then I have this NeXT, the first thing I do when log in is to start the
public domain X11R4-Port (from Der Maus) on it.....
I seldom use NeXTStep, mainly because I'm used to X11.
NextStep will never be the big success : Why ?
1. Think of all the applications running on top of X11, if a vendor suddenly
converts to another window system all applications must be ported to it...
(Who wants to write a transition guide for motif-programmers ????)
2. A user that worked with X11 is probably happy with it and don't want
to start over with a new GUI.
3. MIT-X11 is free, vendors can port it to their platforms and sell it.
Mainly this point lead to the success of X11.
If vendors have to decide between a free base to start from (X11) or
from an expensive encumbred licensed source-code (NeXTStep) they'll
probably choose X11. (most of them have already selected X11 as their
Window System).
4. IBM gave up NeXTStep. (There are still companies looking at IBM as THE
trend-setter..)
|> Fact is that I have no single menu on my screen, however Motiv has
|> it's stupid menu bars taking up space in all windows, even those that
|> are no even active. This I call waste of screen real estate. The fact
|> that NeXTstep allows Mac and Windows bozos to have their menus
|> floating around because those people would otherwise forget that such
|> menus exist, does not mean you have to have them there.
|>
|> The NeXT OS supports about any networking standard there is.
Well I didn't talk about the NeXT OS ! It has some nice features, sure,
I only hate the people that call Mach a micro-kernel, but that's another
story.
We're talking about the window-system & GUI here..
|> So your arguments are quite mute. Try to find someone who owns a NeXT
|> and knows the system (I know this is difficuilt in CH) and tell them
|> to give you a decent demo. Then come back to the Net with your
|> criticism. Have fun...
Sorry ... I'm still very happy with X-Windows.
BTW: NeXT computers are not that expensive in Switzerland. I know a lot
of people who have NeXTs as private home machines, even students.
--
Patrick Guelat, Improware AG Switzerland, CH-4414 Fuellinsdorf
Internet: pa...@imp.ch
Patrick Guelat
|> John, in the 90's, most people are using high level programming languages
|> to write commercial apps. Porting is as simple as recompiling on the new
|> machint does not matter what hardware is running, the important thing
|> is that the operating system,unning on top of this hardware looks the same
|> to the application programs. The port really becomes porting the O.S. (with
|> a few exceptions, like compiler backends) to the new hardwareThis is done by
|> NeXT, this would undoubtedly include the cc compiler. Therefore, the
|> number of apps, from day one of a new introduction (after a couple hours
|> of recompiling) would be almost the same as it is now.
It's not always that easy. Just think of pieces in your code where you didn't
consider that the byteorder might not be the same on a future machine, or
think of field alignments in structures, different handling of bitfields, ....
Ok, if you're software is well designed and takes care of such dependencies
it's not a REAL problem....
A lot of companies have ported their os to different CPUs, Sun did it,
SGI did it, not to forget Altos (remeber these nice little XENIX boxes with
8086 processor ?)... It didn't take long and the applications were ported to
the new platforms, but sometimes it's not just recompile.. it's dbx, recompile,
dbx, recompile, dbx, ...........
Patrick
Andrew Loewenstern
>kenny...@mts.ucs.ualberta.ca (Kenny Leung) writes:
>
>>2) What networking capabilities is NeXTstep missing?
>
>Well, if I'm running 4dwm, and I want to pop up a program from another
>computer, I can just use rsh and the -display flag. If I'm on a NeXT, and
>I want to bring up a window from some mainframe, I have to stop what I'm
>doing, startup X, etc. And as far as I can tell, my local programs (NeXT) and
>my X windows, can't coexist on the screen at once.
Sure you can. coXist does this very nicely.
andrew
--
and...@cubetech.com
Andrew Loewenstern | "If I am not for myself, who will be for me?
Cube Technologies, Inc. | If I am only for myself, who am I?" -Hillel
Max Stevens-Guille
> <bunch of other stuff deleted>
>
> >Well... I'd assume, that if they switch CPUs, they'll do everything
> >possible to make it easy to recompile, and this is certainly easier
> >than it used to be. On the other hand, unless they do some binary
> >translation scheme, a non-68K NeXT won't start life with a whole ton of
> >software, and you have to wonder where it would be on the vendor port list,
> >even as good as some of that software is.
> >
>
> to write commercial apps. Porting is as simple as recompiling on the new
*well deserved flame on*
Boy bryan, you certainly sound smuggly self-assured. Ever ported software to
a new architecture?
I think not.
> machint does not matter what hardware is running, the important thing
> is that the operating system,unning on top of this hardware looks the same
> to the application programs. The port really becomes porting the O.S. (with
> a few exceptions, like compiler backends) to the new hardwareThis is done by
> NeXT, this would undoubtedly include the cc compiler. Therefore, the
And of course, this is just as trivial as typing 'make OS' and off you go.
Seconds really. It's frankly amazing that they pay those guys anything.
Gosh, it's so trivial.
Welcome to the real world, buddy.
*flame off*
> number of apps, from day one of a new introduction (after a couple hours
> of recompiling) would be almost the same as it is now.
>
> bryan
>
Rather than just spit invective, I'll mention a couple of obvious problems.
1) developers will have to get the new machines (i.e. pay more money). Mom
and Pop shops find money hard to find. Remember that recession thing?
2) new architectures bring with them all sorts of interesting software problems.
Things you would never suspect suddenly become big problems. How about byte
order? How about every device driver in the kernel being rewritten? How
about every criteria you profiled against in order to get good performance
becoming a non-issue and everything you didn't slowing you to a crawl?
3) you blindly assume that NeXT won't introduce any bugs in their port. Frankly,
while the folks at NeXT are good, they aren't saints. They'll be in a rush
(just like everyone else) to get stuff out the door and make money. I'll let
you in on a state secret:
there will be bugs in the next version of the NeXT OS. (don't tell anyone)
Assumptions that were made for the last version of the OS may no longer hold.
Work arounds for the last version may not work.
4) Just typing 'make' isn't enough in the real-world. You then have to test
the crap out of the product to make sure it still does what it did before.
This frequently costs as much as developing the product.
I was in school not so long ago. I know where you're coming from bryan.
But please, don't go trivializing the software development process until you've
got a few battle scars yourself.
regards,
Max
--
Max Stevens-Guille Research & Development __
Alias Research Inc. 110 Richmond St. East, /
Toronto, Ontario, CANADA M5C 1P1 / Alias
m...@alias.com / \ Research
vox: (416) 362-9181 fax: (416) 362-0630 / \_ Incorporated
Rui Pedro Salgueiro
: In article <l61pn9...@spim.mips.com> ma...@mips.com (John Mashey) writes:
: > 1) Without knowing any of the details, I think that the rumored NeXT
: > machine was to use a pair of 88100s, not 88000s.
As far as I know there is no such thing as a 88000. What is usually
called a 88000 is one 88100 (CPU + FPU) + one (or more) pair of 88200
(MMU + cache).
This is the original processor announced several years ago.
What I think you meant was :
machine was to use a pair of 88110s, not 88100s.
: > The 88100 seems like
: > a pretty nice design ....
The big problem with 88100 (I dont know if it has been fixed in the
88110 but I doubt (see below) is a lousy floating point performance
(compared to other RISC processors (everything is faster than a 486 :)
)).
I think this is due to lack of registers (the 88k has a unique
register file for the integer unit and the floating point).
So if NeXT is changing processor to have a fast machine the 88xx0 is a
bad idea (IMNSHO). Use a R4000 os something faster if you can find it.
: go out and buy a NCD 19c Xterminal. You will then have a delivered
: system based on the 20mhz 88100. Granted, it's not Unix, but you didn't ask
: for that. Been shipping for months (March 1992 first revenue ship).
: Demoed in January at UniForum.
Not surprising (see above).
Max Stevens-Guille
these comments about porting to new architectures have been
rather interesting.
Anybody else think it's a little strange that the only folks from
a workstation vendor who had anything to say on the topic was SGI?
(I know it was cross posted, but...)
Your's in writing bug-free, self-documented, easily-ported code,
Clark L. Coleman
>As far as I know there is no such thing as a 88000. What is usually
>called a 88000 is one 88100 (CPU + FPU) + one (or more) pair of 88200
>(MMU + cache).
True. But then you use the term "88k" below in a way that only applies
to the 88100 and not the 88110.
>
>The big problem with 88100 (I dont know if it has been fixed in the
>88110 but I doubt (see below) is a lousy floating point performance
>(compared to other RISC processors (everything is faster than a 486 :)
>)).
>I think this is due to lack of registers (the 88k has a unique
>register file for the integer unit and the floating point).
No, the "88k" does not have a single register file, the first generation
implementation (the 88100 CPU) has a single register file. This is not
part of the 88000 family architecture, but part of the implementation
of the 88100. The 88110 has two register files, one for floating point
and one that is mixed (for compatibility with the old 88100 code.) And
floating point performance is greatly improved over the 88100.
The problem is that Motorola is so SSSLLLOOOOWWW these days. If the 88110
were shipping in quantity right now, they would have a market window in
which they had faster chips than the currently available SPARC and Intel
chips. By the time they actually get the 88110 out the door in quantity,
however, it will only be about equal in speed to the SuperSPARC and the
Intel 586. So, NeXT will be erasing a negative (trailing the workstation
market in raw speed) but not leapfrogging anybody and making headlines
with their new CPU.
>
>So if NeXT is changing processor to have a fast machine the 88xx0 is a
>bad idea (IMNSHO). Use a R4000 os something faster if you can find it.
Perhaps. But, I think NeXT needs to differentiate itself with its own
hardware, while porting to Intel hardware in order to appeal to a wider
mass market. Using MIPS CPUs would prompt the question, Why didn't you
just port to the SGI machines and become a VAR for them? Likewise for
using a SPARC or PowerPC or HP CPU. Alpha was mentioned in the latest
NeXTWorld as being too expensive (and it is really only cost effective
for vectorizable codes, IMHO.) The 88110 will be either a good differentiator
(if Motorola gets its act together) or a bad one (if Motorola can't crank
up the MHz on it to keep up.) We'll see how it turns out. In the meantime,
they can sell their software on 486 and 586 boxes as a hedge (and as an
important revenue source.)
Volker Herminghaus-Shirai
Portuesi) writes:
[stuff deleted]
> NeXT could have developed NeXTStep using X. Instead of writing
> their own windowing system, they could have spent the effort on
> developing extensions to X for the services they needed. The
> result would have been just as good as it is now, plus it would
> have the advantage of interoperability and be easier to port to
> other platforms without having to port the entire operating
> system it runs upon.
... and NOT have a unified, standard imaging model...
... and NOT use industry-standard, high-quality fonts...
... and NOT be compatible with standard X-Terminals because they don't
know about the extensions....
IMHO, NOT relying on braindead X is one of the greatest advantages on NeXT.
You can have X if you want it. Implementing NextStep on top of X would not
have raised the lowest common denominator to a windowing system that can
PRINT what it can DISPLAY. Always. Look at what SGI and SUN are doing with
NeWS: dropping it. Not because it's technically inferior to X, but because
it wasn't used by ISVs enough. Why wasn't it used by ISVs? Because they had
X as an alternative. If NeXT had offered X compatibility from the beginning,
do you think anybody would have cared the extensions they built into the
server?
(E.g. does FrameMaker on SGI make active use of SGI's X-extensions for
their 2D/3D-rendering hardware?)
--
Volker Herminghaus-Shirai (v...@darkcube.radig.de)
If I had intel inside(tm), I'd throw up...
Darcy Brockbank
> NextStep will never be the big success : Why ?
>
> 1. Think of all the applications running on top of X11, if a vendor suddenly
> converts to another window system all applications must be ported to it...
> (Who wants to write a transition guide for motif-programmers ????)
Who cares about Motif programmers?
Anyway, if anyone is bound to a machine by religion, then they're
silly. I just roll downhill to where I get the most bang for my
programming minute.
As well, most of the apps running on top of X11 can just stay there.
They're too expensive, and they're weak in areas that the NeXT
is typically strong.
I don't want any X11 apps, and I don't want any DOS apps either.
(Could do with a couple more Mac apps though, and though I'm no
fan of Windows, I could do with Excel...)
>
> 2. A user that worked with X11 is probably happy with it and don't want
> to start over with a new GUI.
>
Let them stay. The computer world is growing rapidly, and there will
be more and more fish in the sea every day. Go catch the fresh ones,
and worry less about the fish in the other guy's net is what I think...
> 3. MIT-X11 is free, vendors can port it to their platforms and sell it.
> Mainly this point lead to the success of X11.
> If vendors have to decide between a free base to start from (X11) or
> from an expensive encumbred licensed source-code (NeXTStep) they'll
> probably choose X11. (most of them have already selected X11 as their
> Window System).
How about "MIT-X11" is *all there was*. X seems good to me if all you're
doing is commandline oriented stuff. But when it comes to an app that
exibits more usefullness then a terminal window (no offense to Scott
Hess :-), then NeXTSTEP is the way to go by far...
X11 can't match NeXTSTEP for application
quality, and application quality is what I'm concerned about if I'm
an end user. I don't really care if my vendor has the source code to
the OS or not.
GIVE ME A KILLER APP OR GIVE ME DEATH!!!!!
>
> 4. IBM gave up NeXTStep. (There are still companies looking at IBM as THE
> trend-setter..)
Yeah, if they want to know how to make their stock slide. :-).
- db
--
-------
I have my own phone here at work, a see-through ruler and my own mechanical
pencil. I'd hope that is the definition of success (I don't want to look
it up in a dictionary though).
-- Hugh Richards
Thomas Insel
>In article <1992Jul13.2...@mnemosyne.cs.du.edu> tin...@uiuc.edu writes:
>>Well, if I'm running 4dwm, and I want to pop up a program from another
>>computer, I can just use rsh and the -display flag. If I'm on a NeXT, and
>>I want to bring up a window from some mainframe, I have to stop what I'm
>>doing, startup X, etc. And as far as I can tell, my local programs (NeXT) and
>>my X windows, can't coexist on the screen at once.
>Sure you can. coXist does this very nicely.
I appologize profusely for this statement, and wish to thank everyone who
has brought coXist to my attention. Now, if the NeXT could support SGI's
graphics library, I'd swear never to even look at a different computer
again.
Chris Lloyd
Heh heh. I also seem to recall that NeXT beat IBM in workstation sales last
year. Need more be said?
In the mean time,
--
:: Christopher Lloyd :: a74...@titan.ucc.umass.edu :: Yrrid Inc. ::
:: AAAAAAAAAAAAAAAAAAAIIIIIIIIIIIIIIIIIIIIIIIGGGGGGGGHHHHH!!!!!!! ::
Michael Portuesi
|> >>>>> "Patrick" == Patrick Guelat <pa...@imp.ch> writes:
|>
|> Patrick> 2. A user that worked with X11 is probably happy with it and
|> Patrick> don't want to start over with a new GUI.
|>
|> WRONG! I worked with X11 before NeXTSTEP. As a user,
|> I was less than pleased. As a would-be programmer, I was
|> definitely unhappy. I've got both kinds of boxes on my desk,
|> and I sit in front of the NeXT all day while the other box
|> crunches numbers in the background.
You are yet another person who cannot differentiate between
a windowing system (X) and an applications toolkit (NeXTStep).
There is nothing about the X11 protocol, or its extension
mechanism, which would prohibit the development of a
NeXTStep-like toolkit on top of it. The implementation
may have been more difficult than it would have been
otherwise, but the rewards for NeXT, and the industry in
general, would have been more than worth the investment.
If NeXT had based its toolkit on X11, Motif probably
wouldn't exist today and NeXTStep would likely be the way
applications are developed on a wide variety of platforms,
including SGI. As it is, I'm not sure NeXTStep is going
to survive as a commercially viable software system, which
(to me) is a crying shame because it means that we all
lose, not just NeXT.
My opinions are my own, not necessarily those of SGI.
Michael Wasfy Nicolas Ibrahim
>
> How about "MIT-X11" is *all there was*. X seems good to me if all you're
> doing is commandline oriented stuff. But when it comes to an app that
> exibits more usefullness then a terminal window (no offense to Scott
> Hess :-), then NeXTSTEP is the way to go by far...
>
> X11 can't match NeXTSTEP for application
> quality, and application quality is what I'm concerned about if I'm
> an end user. I don't really care if my vendor has the source code to
> the OS or not.
I hate to get dragged into the middle of a thread but I think I have
experiences to share :^). I am a NeXT station owner and am currently
doing an internship where I get to use HP Apollo's. I never realized
until now how spoiled the NeXT interface has made me. Sure HP's have
balls but who cares, the guy who made the open panel was an idiot. That
is just an example of the many little things that NeXT thought of, and
they deserve credit for that. On the HP it all comes down to this: it is
easier to use the command line than the the GUI. Every day I see people
using the HP's for basically one purpose: to have 4 terminal windows.
What's the point? If that's what you want go buy 4 vt100's.
Michael Ibrahim
before August 5: wa...@ibrahim-michael.eigenmann.indiana.edu
after August 5: wa...@nic.gac.edu
NeXT mail welcome at both
Barry Merriman
> Portuesi) writes:
> [stuff deleted]
> > NeXT could have developed NeXTStep using X. Instead of writing
> > their own windowing system, they could have spent the effort on
> > developing extensions to X
No they couldn't. The Next development occured during the mid 80's.
At that time the most that would have been available would have been
an alpha-alpha version of X10!! Developing any major GUI on top of that
would have been ridiculous. Even the X consortium totally scrapped X10
and went on to X11, which had major changes and incompatibilities.
--
Barry Merriman
UCLA Dept. of Math
UCLA Inst. for Fusion and Plasma Research
ba...@math.ucla.edu (Internet; NeXTMail is welcome)
Barry Merriman
>
> 1. Think of all the applications running on top of X11
Ok...gee, that was a quick thought. All that passed through my mind were
a bunch of half-baked PD stuff and a few very expensive niche market
apps. Did I miss anything? Where's the cheap WYSIWIG word processor
for X?
>
> 2. A user that worked with X11 is probably happy with it and don't want
> to start over with a new GUI.
Wrong. I started with SunTools. It was no good, so I switched to
X10 when it came out. I used X for 4 years, was something of an Xpert,
but gladly switched to NeXTStep when it came out---what a breath of fresh
air. I would say this: users who don't really care about the quality
of their windowing environment (if its got windows and menus, thats enuf)
will stick with X (or whatever else they are trained on) rather than switch
if given the option.
>
> 3. MIT-X11 is free
And you get what you pay for.
>
> 4. IBM gave up NeXTStep. (There are still companies looking at IBM as THE
> trend-setter..)
Yes, the trend there being that their inferior unix was ill-equiped
to run NeXTStep, and their archaic management had little vision. Excellent
trends to follow, I'd say.
> Patrick Guelat, Improware AG Switzerland, CH-4414 Fuellinsdorf
> Internet: pa...@imp.ch
Basically, you stick with X because you learned it first, and its
free. Fine, but then I don't consider you to be a very demanding
consumer of GUIs.
James Gardiner [Hunter]
>|> If you want this funtionality on a Indigo, you would have
>|> to go redesign the servers and underlying windowing system.
>|> Ie. copy nearly exactly what nextstep is. A little silly
>|> when nextstep already exists.
>You are incorrect. The primary advantages of NeXTStep as a
>a software development environment or as a graphical user
>interface have little to do with the servers or the imaging
>model which it is based upon.
You are right but, I think there is more to NextStep then just
a good development platform. Because it is such, some interesting
products have come out of Next developers. Because of these interesting
products, the next has sold quite a few machines to buisness.
Doctors, Lawers, government. They do not use Nexts for development.
>Those "servers and underlying window system" you refer to may
>have been designed to make it convenient for the implementation
>of NeXTStep, but it is by no means a requirement for a
>NeXTStep-like software environment. There are public-domain
>toolkits available which are fully object-oriented, support
>dynamic object linking, support object transcription,
>have interactive application builders, and are window system
>independent to boot.
I do not clame to be a programmer but I know that Xsgi already
has some support for features like drag and drop with workspace.
Implementing next type features and keeping all old compatability
leads to obese executables and possibly conflicting problems.
NextStep type functionality could probably be easily established
but not compatability between Xsgi and NextStep servers.
>|> I would love to have nextstep intereface on my indigo.
>|> In real terms a combination. IE nextstep with inbuilt
>|> X11 support to keep it all very fast.
>Apparently you do not have a clear grasp of the technical
>issues involved. X11 on the Indigo has extremely high
>performance because it talks directly to the Indigo display
>hardware, and has been finely tuned and optimized to do
>that in an efficient manner.
Um strange, I have heard Next users in c.s.n.* complain that
the x-systems like coXist are very slow on the next.
(CoXist alows NextStep and X to run together.)
I know true X servers exist on Next and run fast but what good is a next
if its not running the nextstep interface. You through away the
hole reason you would chose it over another system.
>NeXTStep requires its own proprietary windowing system,
>built as extensions to Display PostScript, to run.
>|> If it had X11 as a crutch to lean on when it lacks in
>|> application, it would have been affected by X and products
>|> as good as they are now may not have beenn developed as open
>|> minded as they where.
>NeXT could have developed NeXTStep using X. Instead of writing
>their own windowing system, they could have spent the effort on
>developing extensions to X for the services they needed. The
>result would have been just as good as it is now, plus it would
>have the advantage of interoperability and be easier to port to
>other platforms without having to port the entire operating
>system it runs upon. Even companies like SGI might have been
>interested in licensing it.
NeXT could have but didn't. Speculate why please.
It would have been advantages for NeXT as well as the rest of the
world if they did do this. But they didn't. I don't know the
difference between the two windowing systems, but there must be some
big differences. Why develop a totally new windowing system otherwise?
Every aspect of how a windowing system works affect how things are done
and eventually affects the final product. If Frame was developed on the
next and ported to the X system, it would most probably carry over
some NeXT type user interface qualities. This is what I mean when
I say, next products would have been different if next supported X.
And they would have been.
>--
Michael Portuesi Silicon Graphics, Inc. port...@sgi.com
This is the most impressive quality of SGI. Day to day convo about
issues that affect the future of computer development. It is very
reassuring to talk to SGI people about there opinions of product
and indestry. The scaryest part about buying a Indigo was that
how could I be sure SGI does not turn around and stop development
of the products we buy the machine for.
James
--
------------------------------------------------------------------------------
James Gardiner [Hunter] [hun...@bacchus.esa.oz.au] Expert Solutions Australia
Systems Administrator ACN [003 130 434]
Sundar Narasimhan
about programming languages. They reflect the subjective results of
your experience. WRT NeXTStep: Almost all the developers I have come
into contact with who are fanatic about it have never played with or
used a better environment. In their experience the NeXTStep environment is
just about the best there is for program development. This opinion
however is relative. I know of a quite a few Lisp Machine developers
who (grudgingly) hack Motif/OpenLook/NeXTStep these days. For these people,
NeXTStep is definitely a step down, not up.
Personally speaking, I hope that X/NeXTStep don't do for window
systems/programming development environment research what Unix did
for operating systems research. :-) Commercial success != technical merit.
Nathan F. Janette
> >>doing, startup X, etc. And as far as I can tell, my local programs (NeXT)
> >Sure you can. coXist does this very nicely.
> I appologize profusely for this statement, and wish to thank everyone who
> has brought coXist to my attention. Now, if the NeXT could support SGI's
> graphics library, I'd swear never to even look at a different computer
> again.
About a year ago there was some news that SGI would make "open GL"
or something like that available for licensing. A number of vendors
were mentioned as signing up for ports, and one of them (I think)
was intel for the i860. I was hoping NeXT would support that and
make it available for NeXT Dimension system.
Ho ho.
--
Nathan Janette "I'm a NeXTstep man,
Dept MB&B, Yale Univ I'm a NeXTcube guy"
New Haven, CT
nat...@laplace.biology.yale.edu (NeXT)
Larry D. Pyeatt
|> Perhaps some facts are in order, to avoid apples-to-oranges comparisons.
|>
|> 1) Without knowing any of the details, I think that the rumored NeXT
|> machine was to use a pair of 88100s, not 88000s. The 88100 seems like
|> a pretty nice design ....
|> a) It was described well by Keith Diefendorff at Microprocessor
|> Forum, November 1991.
|> HOWEVER:
|> b) To my knowledge ( and correct me if I'm wrong), no one has yet:
|> -announced a system product based on the 88100
|> -demoed such a product in a public event
|> MUCH LESS:
|> - delivered production systems based on it.
Data General has an Aviion based on two 88100 chips. It is a nice,
powerful workstation.
...stuff deleted...
|> 2) Consider the interval from chip tapeout until production systems.
|> A good rule of thumb, for an aggressive, brand-new design, but which does
|> have useful existing software in the family, but where things are
|> going OK, is 10-12 months from tapeout to production (virtual-memory)
|> systems. The 88000 was announced in April 1988, and it was late Summer
|> of 1989 before the bug cleanup was OK enough to ship production systems
|> (i.e., the FP exception bugs, I think).
|>
are you sure you aren't talking about the 80486?
--
Larry D. Pyeatt The views expressed here are not
Internet : pye...@texaco.com those of my employer or of anyone
Voice : (713) 975-4056 that I know of with the possible
exception of myself.
Mark Thomsen
from one X implementation to another or b) there was a NeXTSTEP adapter for
common X toolkits, as a layer that mapped AppKit objects onto the X stuff.
But there are many, many problems. The timing between X versions and NeXT
starting their engineering project makes it clear why NeXT skipped X. X was
basically a terminal-network arrangement at that time with little industry
support. (This timing situation explains the choice of Objective-C over C++
too). The layering of NeXTSTEP today has Display Postscript underneath the
window system underneath the AppKit. DPS is pretty integral to NeXTSTEP. The
amount of surgery required to make a NeXTSTEP-X merge that has the quality of
NeXTSTEP is probably measured in man-decades. That's assuming it is truely
possible.
The adapter was tried by my group here in 1990 (for OpenLook XView). It can
work but it depends upon having the full Objective-C toolset for porting and
maintaining the adapter to versions of NeXTSTEP and versions of X and versions
of the particular toolkit. Serious product versionitis is guaranteed. We chose
to let the (partially completed) stuff die rather than pump more $ into it.
So we are left with competition. I have grave concerns over X. I have yet to
hear developers tell me it is well designed, makes GUI applications easy to
develop, it is wide open for advanced graphics, the UI is good to use, or
anything else of merit. My personal opinions based on limited use and
programming follow this. The only reason we work it (and we work it big time at
TRW) is it is a 'standard'. In my more cynical moments I tell senior management
that X is really a plot by programmers to employ lots more programmers and
guarantee lifetime employment for hackers - that software engineers learned of
the lawyer's ability to create employment for one another and adapted the
lesson.
NeXTSTEP is, at the moment, the opposite. Developers single it out as the best
(provided they are not Symbolics vets). It is a quality environment for users.
Development is fast ... a feature of the level of reuse, the simplicity of the
interfaces, the InterfaceBuilder, and the positive attitude that a quality
application will result. But it is not a 'standard' so senior management and
customers take more convincing.
What I advocate is NeXTSTEP in it's entirety offered on other platforms: SPARC
2, Indigo, DEC 5000, RS/6000 (I know), HP 700, Mac Quadra, and 486 clones. We
are getting the last one over the next nine months.
Mark R. Thomsen
MK...@psuvm.psu.edu
Stevens-Guille) says:
>In article <l638ql...@aludra.usc.edu>, stri...@aludra.usc.edu (Bryan
>Strickland) writes:
... discussion deleted ...
>1) developers will have to get the new machines (i.e. pay more money). Mom
> and Pop shops find money hard to find. Remember that recession thing?
>2) new architectures bring with them all sorts of interesting software .
>problems
> Things you would never suspect suddenly become big problems. How about
>byte
> order? How about every device driver in the kernel being rewritten? How
> about every criteria you profiled against in order to get good performance
> becoming a non-issue and everything you didn't slowing you to a crawl?
>3) you blindly assume that NeXT won't introduce any bugs in their port.
>Frankly,
> while the folks at NeXT are good, they aren't saints. They'll be in a rush
> (just like everyone else) to get stuff out the door and make money. I'll t
>le
> you in on a state secret:
> there will be bugs in the next version of the NeXT OS. (don't tell
>anyone)
> Assumptions that were made for the last version of the OS may no longer .
>hold
> Work arounds for the last version may not work.
>4) Just typing 'make' isn't enough in the real-world. You then have to test
> the crap out of the product to make sure it still does what it did before.
> This frequently costs as much as developing the product.
All good points. It occurred to me that there is probably a great business
opportunity here for the right people. As NeXTstep appears on more and more
(we hope) platforms, it will indeed become harder and harder for developers
to assure quality products.
However, rather than require every developer to test their product on
numerous platforms (duplication of effort) there should be a small number
of firms that specialize in such testing, and offer advice on debugging,
porting, etc. Perhaps even neXT should offer this service themselves.
Perhaps some of the new NeXT power-company-wannabies will get together
on this. Imagine Pages, Stone Design, Lighthouse, etc etc getting together
and jointly funding a company to offer this service.
Of course, I speak for no one, hardly even myself.
Jon Rosen
>Michael Portuesi writes
>
>>|> If it [the NeXT] had X11 as a crutch to lean on when it lacks in
>>|> application, it would have been affected by X and products
>>|> as good as they are now may not have beenn developed as open
>>|> minded as they where.
>>
>>NeXT could have developed NeXTStep using X. Instead of writing
>>their own windowing system, they could have spent the effort on
>>developing extensions to X for the services they needed.
>
>Sure. They could have set themselves afire too.
The effect of either would have been more positive than using X :-)
X is so totally brain-dead AS A GUI ENVIRONMENT that it utterly
defies human comprehension. I spent 14 months managing an X project
at Locus Computing Corporation. We managed to build parts of a
database query product during that time, using a Motif "gooey builder"
(aptly named IMHO) and X. In 2 months, I cloned the prototype on
a NeXT workstation. But it was much better. The X version couldn't
print WYSIWYG in Postscript because I didn't have the time or energy
or resources in my group to build a complete X-to-Postscript driver
for our access window. On the NeXT, it was nothing more than
adding a [Window printPSCode] line in the program (and adding a
few methods to handle pagination. It is hard to believe that X,
designed several years after the Macintosh was released, did not
have scalable fonts built into its architecture (no, I don't mean
Display Postscript, just simple scalable fonts, even grungy ones
like the original Mac).
>>The
>>result would have been just as good as it is now, plus it would
>>have the advantage of interoperability and be easier to port to
>>other platforms without having to port the entire operating
>>system it runs upon. Even companies like SGI might have been
>>interested in licensing it.
>
>I see. So it was a bad idea for NeXT to implement an advanced GUI on an
>advanced operating system. They should have gone the Windows route. Oh.
>When new technology is ever introduced, the first thing the industry says is
>"it's not compatible with our old stuff". Of course: old ideas are...old.
>
>It reminds me of the joke about the company that introduces a brand new
>computer. It's 10000 MIPS, has huge memory storage, is the easiest computer in
>the world to learn, comes with tons of applications, and is free. So what does
>the customer say?
>
> "Is it compatible with 1-2-3?"
Sorry, on this one I disagree. This so slurs 1-2-3 (to compare it to X)
that I must object. 1-2-3 was a GOOD (in fact, an exceptional) product
for the time and place it was introduced. I have NO objection to trying
to provide compatibility with older GOOD products/designs/architectures
where something is to be gained by doing so. X, on the other hand, was
inferior to several existing architectures or directions (Mac, Postscript,
Windows (uggh :-), etc.) and was invented by DEC solely for the purpose
of defeating Sun (NeWS was pretty good and was an early version of
Display Postscript). X was still-born, and deserved to be given a
dose of RU486.
Jon Rosen
"If Al Gore spells trouble for the GOP, that puts him a step up on
Dan Quayle, who can't spell trouble for anybody." --- Arsenio Hall
james dipalma
So we are left with competition. I have grave concerns over X. I have yet to
hear developers tell me it is well designed, makes GUI applications easy to
develop, it is wide open for advanced graphics, the UI is good to use, or
anything else of merit. My personal opinions based on limited use and
programming follow this. The only reason we work it (and we work it big time at
TRW) is it is a 'standard'. In my more cynical moments I tell senior management
that X is really a plot by programmers to employ lots more programmers and
guarantee lifetime employment for hackers - that software engineers learned of
the lawyer's ability to create employment for one another and adapted the
lesson.
NeXTSTEP is, at the moment, the opposite. Developers single it out as the best
(provided they are not Symbolics vets). It is a quality environment for users.
Development is fast ... a feature of the level of reuse, the simplicity of the
interfaces, the InterfaceBuilder, and the positive attitude that a quality
application will result. But it is not a 'standard' so senior management and
customers take more convincing.
--------------
Currently I am a student who WANTs to one day code something important. I
get so frustrated with X cause I feel like I am using 4 vt100s (reference
to an earlier post) on this 17 inch color display with odies pasted on the
root window. I program in C and I sys admin (an awesome job for a bio
major without a degree). The stupid thing is that my programs would NOT be
any different if I was sitting in front of a vt100.
I use a NeXT during the school year, and it just amazed me how easy it
appeared to be for programmers to utilize the windowing environment with
their aplications (from watching Steve Jobs announce Improv in Boston and
from various BCS NUG meetings). The SUN diehards laugh at my NeXT
enthusiasm and tell me that X and/or SUN have plenty of tools that do just
what a NeXT does. "Go look at InterViews (sp?)," or ,"You have to look at
SPARCworks. The OpenWindows Developer's Guide lets you drag buttons and
stuff." Of course these people have never developed on anything, let alone
used the programs that they are recommending.
Is the 'joy' of programming with NeXTSTEP a reality? What about all those
toolkit things that aid in developing software in other environments?
My primary curiousity is what people, who have developed stuff with NeXTSTEP
and have developed stuff using other environments and what people, think.
Yes, I am young; I am a student; I have never compiled on a NeXT; I am a
Biology major; and I waste a lot of time yoyo-ing. If you think that my
first post is just a rehash of old news, please send mail to
dip...@csa.bu.edu with the header "Why Should I Bother When you could FTP
a comparative study by ???" and include the ftp address in the body of the
message. OR just send mail with the title "I don't want to make this
thread any longer."
-jim
Nick Kline
>has brought coXist to my attention. Now, if the NeXT could support SGI's
>graphics library, I'd swear never to even look at a different computer
>again.
I have read about a public domain library that can output code for
different windowing systems (GL is one of them).
Also recently SGI begain encouraging others to implement GL as a standard.
I feel soon there will be implementations of this for the next.
It appears that one day, you might never be able to look
at other computers :-)
John Mashey
>|> 2) Consider the interval from chip tapeout until production systems.
>|> A good rule of thumb, for an aggressive, brand-new design, but which does
>|> have useful existing software in the family, but where things are
>|> going OK, is 10-12 months from tapeout to production (virtual-memory)
>|> systems. The 88000 was announced in April 1988, and it was late Summer
>|> of 1989 before the bug cleanup was OK enough to ship production systems
>|> (i.e., the FP exception bugs, I think).
>are you sure you aren't talking about the 80486?
Like I said in another post, I screwed up, consistently at least.
Every place that said 88100, replace it with 88110, which is what I meant.
Every place it say 88000, it really meant 88100+88200s...
I didn't mean 486, although it had some similar problems.
The *486* was taped out 2/89, announced 2Q89, and in production systems
around 12/89 - 1/90.
I don't know the 88100/88200 tapeout dates, but the early history, for sure,
was (first instance of mention):
1) Electronics, Feb 18, 1988, p83:
"Motorola Semiconductor, Inc" is about to spring a big surprise on the
electronics industry. While it's no secret that the U.S. chip leader has
been working for a couple years on a reduced-instruction-set-computer chip,
only a few of its customers know that the chip maker saw silicon last year
and expects to unwrap its product-a three-chip set-during the second quarter
of 1988... Jack Browne, Jr., marketing director...
...will burst out of the gate with architectural horsepower meant to leave
the competiion eating dust... `Working silicon is here and we have measured
its speed at 34,000 Dhrystones,' Browne says. `We are projecting a
rough 17 working VAX mips tostart, but we will show our customers how
they can hit 50 mips this year.' .... And he is quick to mention the more
than $20 million Motorola will spend this year alone in RISC
development, not counting process development. ... Browne wonders if
`$100 million companies building RISC chips have deep enough pockets for
all the necessary support.'"
2) 4/18/88: 88K announcement.
3) 8/15/88 (From CSN Article): 88K & Roger Ross move under Tom Gunter in Moto.
4) 8/29/88 (from EE Times): Roger Ross leaves Motorola.
5) 9/12/88 (From EE Times): Motorola sues Roger.
During the Fall of 88, there were supposed to be yield problems on the
CMMU parts, and difficulties fixing the synthesized logic; also,
there was something weird with FP, where if you had an exception during
some FP operation, bad things happened, so FP benchmarks were being
run with exceptions turned off.
6) July 1989: MIPS magazine (nothing to do with MIPS Computer):
"The Motorola 88000 is still in its very early days. Although current
numbers are impressive, they have been generated with prerelease versions
of the silicon... D.5 step of the silicon, which still had problems that
significantly slowed floating-point performance. ...
Data General expects to ship production systems with E.2 silicon and
the new compilers in July, about when this issue appears."
So, here it looks like tapeout (4Q87) to production (3Q89) is
something like 18-21 months....
Now, why is any of this relevant to this newsgroup?
Please recall the earlier posting where someone got irritated that
anyone would compare Motorola to a small design house like MIPS....
that Motorola was a huge company with vast resources......
CHIP TAPEOUT PRODUCTION SYSTEMS INTERVAL (MONTHS)
R3000 1Q88 4Q88 [MIPS M/2000] ~ 7- 8
R4000 1Q91 1Q92 [CRIMSON] 11
68040 4Q89 2Q91 [HP 9000/4xx] ~ 15-18
88100 4Q87 3Q89 [DG Aviions] ~ 18-21
Now, I do not mean here to criticise Motorola, which is a quality company
with many good people, merely to use some reasonably comparable chips
(R3000+R3010 == 88100; R4000 == 68040 in terms of transistor counts,
if not in many other ways) as an example to show that this leading edge
chip stuff is hard work.
Just saying:
"Company X has vast resources, and therefore company Y cannot compete"
is demonstrably *bogus*.... (and I explained why earlier in the sequence
of postings.) No matter how big a company is, it *still* comes
down to a fairly small team of engineers and the resources provided them,
and the team can only be *so* big for a given kind of project,
and if it's bigger, they just get in each others' way; only so many engineers
can fit into one chip :-)
Peter Kocks
>>>>>> "Patrick" == Patrick Guelat <pa...@imp.ch> writes:
>
>Patrick> 2. A user that worked with X11 is probably happy with it and
>Patrick> don't want to start over with a new GUI.
>
>WRONG! I worked with X11 before NeXTSTEP. As a user, I was less than
>pleased. As a would-be programmer, I was definitely unhappy. I've got
>while the other box crunches numbers in the background.
>
I would also add that X11 based computers I have used have taken too
long to set up. I worked with some dec3100s that came with X11 and
it took us (chemists) two days to get the things to even boot up. It
was a month before we compiled emacs from the tapes. When I bought my
NeXT all I had to do was:
1) Open the box
2) Plug in the machine, monitor, and keyboard
3) Turn it on.
4) Find and launch the terminal app. (took me 30 seconds).
5) type "emacs"
and I was off and working. This was worth to me at least $1000 per
machine. I wish our department had bought NeXT, so we students could
work instead of maintain dec 3100s.
I know this has little to do with X11, but they are related. X11
systems tend IMHO to be on machines which take far too much time and
effort to maintain. NeXT saves me hours in the long run (much more
time than having a faster cpu would).
Upgrades of X11 from R3 to R4 to R5 have been almost useless as far as
I can tell. Maybe some new features for the programmer, but nothing
for your typical chemist. NeXT, on the other hand, is giving a great
upgrade from NeXTstep 2 to 3. For chemist's we get the new renderman
based 3d imaging system. SGI model is probably better for chemist's,
but I believe that this will change with a faster machine from next
and possibly some Open GL to renderman portability tools -- not here yet
but I bet they will come.
IMHO X11 will die because it is free and a standard. This means that
small changes in interface design and new tools will require
compatibility checks and general approval between several companies.
For example, if a standard 3d image model (or even a scalable font
model) were to be implemented on X11 there would be at least a 2 year
battle between SGI, DEC, IBM, HP, SCO, etc etc for a) what model and
b) debugging across several vendor's hardware. It will take forever.
On the NeXT, it takes just one company to make the change. It is
easier for me to follow what is happening at NeXT, than gleen the
information on future upgrades from the X11 compatible vendors.
My hope (prayer) is that NeXT will port NeXTstep to an OSF version of
Mach/Unix. To run NeXTstep on multiple hardware configurations would
be great. From what I can see, it will be easy as pie to recompile a
program on one NeXTstep based machine to another. X11 on the other
hand always seems like a nightmare. Is the program you are trying to
compile going to work on X11 R3, R4, or R5? If you don't have an
Imakefile, you have to edit a long and horrible makefile and then it
probably will not work without at least a couple hours of work.
Another exampke, if the program was compiled on ultrix 4.2 it may not
work on utltrix 3.1. If you don't have ultrix 4.2 you have to spend
$$$ for a new disk drive to hold the huge ultrix 4.2. That's the dec
problem. I'm sure that for all other vendors of X11 based machines,
there is a whole set of compatibility issues. If NeXTstep were on all
of them, it would be as easy as double clicking a package icon. Want
X11 to have a package icon like the NeXT? Dream on. It will take
years for them to do it across all the vendors.
Peter Kocks
ko...@chemistry.stanford.edu
PS. SGI is my second choice. DEC or IBM second. SUN third. Sun does
not come with an ANSI c compiler (you have to get gcc, which NeXT
incidently gives you compiled and ready to go).
Robert Munyer
In article <n8p...@zola.esd.sgi.com> go...@puck.esd.sgi.com (Michael Gold)
writes:
> ... And is NeXT going to supply each and every software vendor with several
> new machines at no cost; for engineering, SQA, and customer support?
Sounds like a good idea to me. Well, at least every software vendor that they
consider _significant_. They could lend the machines, not give them outright.
They could even do it _before_ they start selling the new machines to the
general public.
______________________________________________________________________
Robert Munyer | "I may be synthetic, but I'm not stupid."
rob...@metropolis.com | -- Bishop, _Aliens_
Izumi Ohzawa
>
>| 1) Open the box
>| 2) Plug in the machine, monitor, and keyboard
>| 3) Turn it on.
>| 4) Find and launch the terminal app. (took me 30 seconds).
>
Wow!
I didn't know that SGI systems have evovled into such
intelligent machines -- it comes out of the box and plugs itself
togeter and auto-power on!
NeXT is way behind.
--
Izumi Ohzawa [ 大澤五住 ]
USMail: University of California, 360 Minor Hall, Berkeley, CA 94720
Telephone: (510) 642-6440 Fax: (510) 642-3323
Internet: iz...@violet.berkeley.edu NeXTmail: iz...@pinoko.berkeley.edu
Jonathan Hendry
>4. IBM gave up NeXTStep. (There are still companies looking at IBM as THE
> trend-setter..)
Two possible reasons for this, that I've heard. Take your pick:
1) IBM gave up NeXTStep because it ran like a dog on their machines, due
to their relctance to use a mach foundation.
2) IBM internal politics killed it. They knew that it was superior to what
they had and didn't want the DOS/Intel base (read: cash cow) to be
challenged by a powerful Unix/Risc line. " According to an IBM insider,
'IBM simply couldn't afford to put that good a solution on Unix.' NeXTstep
was just such a good solution." (Upside magazine, May 1992).
Now look at IBM. They are desperately trying to establish OS/2 as a better
alternative to MSWindows. (And not doin a great job of it.) Now everyone's
faced with Windows NT, and many people aren't happy with more years of
domination by Bill Gates. IBM should have gone with NeXTStep.
--
Jonathan Hendry Anderson Financial Systems
j...@afs.com (Nextmail Welcome!) or hend...@duvm.ocs.drexel.edu
"You can't run a country by a book of religion. Not by a heap or a bunch
or a smidgen."-F.Z. "Dumb All Over"
Thomas Insel
>I have read about a public domain library that can output code for
>different windowing systems (GL is one of them).
>Also recently SGI begain encouraging others to implement GL as a standard.
>I feel soon there will be implementations of this for the next.
I would be interested to know if NeXT's have the hardware to run gl well.
For example, sgi's have an extra 24-bit z-buffer for hidden surface stuff,
some have hardware gamma correction, and different windows can run in
different graphics modes simultaneously. Also, I would venture most Iris
have better cpu performance. None of this is meant to knock NeXT, it's
in a different niche, and has a nicer pricetg.
Dave Olson
| In article <nd2...@zuni.esd.sgi.com> ol...@anchor.esd.sgi.com (Dave Olson) writes:
| >| 1) Open the box
| >| 2) Plug in the machine, monitor, and keyboard
| >| 3) Turn it on.
| >| 4) Find and launch the terminal app. (took me 30 seconds).
| >
| >You don't even have to do this on the SGI system; it comes up itself
|
| Wow!
| I didn't know that SGI systems have evovled into such
| intelligent machines -- it comes out of the box and plugs itself
| togeter and auto-power on!
Nah, that's the next generation robotics option!
(For those who didn't get it, I was of course referring to item #4,
not the whole sequence.)
--
Let no one tell me that silence gives consent, | Dave Olson
because whoever is silent dissents. | Silicon Graphics, Inc.
Maria Isabel Barreno | ol...@sgi.com
James Clifton Elam
> In article <nd2...@zuni.esd.sgi.com> ol...@anchor.esd.sgi.com (Dave Olson)
writes:
> >
> >| 1) Open the box
> >| 2) Plug in the machine, monitor, and keyboard
> >| 3) Turn it on.
> >| 4) Find and launch the terminal app. (took me 30 seconds).
> >
> >You don't even have to do this on the SGI system; it comes up itself
>
> Wow!
> I didn't know that SGI systems have evovled into such
> intelligent machines -- it comes out of the box and plugs itself
> togeter and auto-power on!
>
> NeXT is way behind.
Yes, when will NeXT finally realize that what millions of X users want
is really just a bunch of auto-launching terminal applications. It
probably is smart enough to come up in 40 character mode and automatically
dial prodigy!
--
- Cliff "I want my NeXTstep 3.0" Elam
(mail:j...@glv.com)
Mark Neidengard
>| I would also add that X11 based computers I have used have taken too
>| long to set up. I worked with some dec3100s that came with X11 and
>| it took us (chemists) two days to get the things to even boot up. It
>| was a month before we compiled emacs from the tapes. When I bought my
>| NeXT all I had to do was:
>
>with X vs NextStep. The SGI systems all come up directly with nothing
>to configure (but plenty to customize, if you wish); presumably so
>do some other company's systems.
>
>| 1) Open the box
>| 2) Plug in the machine, monitor, and keyboard
>| 3) Turn it on.
>| 4) Find and launch the terminal app. (took me 30 seconds).
>
>
>
>OK, so here you would have to use vi, or take the time to install
>emacs :)
>
>--
>Let no one tell me that silence gives consent, | Dave Olson
>because whoever is silent dissents. | Silicon Graphics, Inc.
> Maria Isabel Barreno | ol...@sgi.com
I have been watching the flaming from the sidelines, and am frankly
amazed at the narrowmindedness of most of the commentary. I have used a
NeXT in the past, as well as Sparcstations and the like. I have the
following to say:
1: X11 is compatible. I can go anywhere and find X-Windows, and I can also go
anywhere to find X-APPS! It was said that compatibility was undesirable,
and that new technology requires a break from the compatible way of doing
things, which leads me to point 2:
2:Revolution vs. evolution. The NeXT has some very attractive elements in
its design and in the design of its OS. However, it should be noted that
these changes are not RADICAL departures from the norm. The CPU's that NeXT
has been using are GROSSLY ordinary, both in terms of interested vendors and
in terms of performance. The operating system is, indeed, highly objective
and tightly coupled with its Postscript Imaging Model and Mach kernel.
However, this is merely an EXTENSION of what already exists. With OSF/1
coming out, NeXT's Mach advantage will be largely obliterated when it goes
into wider use on other platforms. Also, it seems likely to me that the
upcoming versions of OSF should implement Mach 3.0 microkerneling, a feature
that NeXT hasn't implemented yet in 3.0 of the OS.
3: Ah, the infamous GUI. Yeah, it's a good looking system, and for the MOST
part I like how it's organized, and the features like the directory tape and
recycler. But, all this bunkum about how X can't do this, and X can't do
that, shows me that some of the NeXT moguls have been reading too much of
their own marketing propaganda. If you want a 3D toolkit that is REALLY
useful, it has to be device independent. PHIGS is one of the greatest
toolkits around for this, and who uses it? X11R*5* does, as well as
scalable fonts, improved network security, and a slew of other features.
Furthermore, I can RUN X11R5 on more than one kind of computer (including
the NeXT from what I hear) which allows me to not worry as much about
writing machine-dependent code. Sure the NeXT is tightly coupled and all
that, but X11R5 will IMHO succeed because it provides all the features
required for high-performance network graphics (in 3D no less) with the
option of using different graphics libraries (like GL).
4: System power. We all know that the NeXT is taking it in the shorts in
this category. I've used the Turbostations, and they're REASONABLY
responsive, but they simply aren't FAST. When I think of the HP9k systems
and they're raw power, or even an RS6k, and then think of the NeXTs, I sort
of cringe. I think it's really a shame that the wonderful NeXTOS doesn't
run on a machine that has the power to really move data. Unfortunately, for
compiling, imaging, or raw numerics, I must say that while the NeXT is a
CONVENIENT platform, the HP9k and other like systems are PRACTICAL
platforms. With VUE (and hopefully X11R5 soon) I can get just as much work
done on a HP, and FAR FASTER, as I can on the NeXT. What I find myself
doing is using the NeXT for its native apps, but using coXist to fire up X
on a more powerful machine if I REALLY need to get some heavy work done (ray
tracing). What would be nice is to implement RPC's or something similar to
make NeXT apps use more powerful machines as backend compute engines, and I
hope this will become easier as OSF/1 gains acceptance.
4: Money. The NeXT is still one of the best computing values per dollar out
there. Yeah, I doubt many of us here (myself included) is in a position to
purchase a HP9k/750. The NeXT ships with more of the useful bundled
software than any other platform I know of, and it does it at a very low
price. In this respect, the NeXT is worth having not only for the KILLER
development environment (which I still haven't seen equalled) but the low
price to buy it with.
In fine, the NeXT is a good machine for using, for software development, for
writing and maintaining in-house applications, and for various graphic-arts
and publications pursuits, all of which NeXT is marketing for. In the realm
of SCIENTIFIC computing, where people evidently are not SCARED of a
commandline (I know that was cheap, couldn't resist :) ), it seems like the
NeXT is good for prototyping and personal use, but not necessarily for the
actual scientific crunching. I know that the proposed NRW might correct
this, but being that the chip in question is the 88110 in PARALLEL (an area
that NeXT is somewhat unproven in), I'm not gonna hold my breath too
tightly.
Mark Neidengard
(no NeXTmail please)
mnei...@nike.calpoly.edu
"The other 'P'??? Well, that's not that simple..."
Naughty by Nature
Scott Byer
> X11R*5* does, as well as scalable fonts
Simple side note:
X11R5, while having scalable fonts, is certainly not up
to any simple typographic tasks. It's font interface is
very primitive, not even having fractional widths, much
less non-horizontal writing. You just can't write a good
word processor very well with X, without doing quite a
bit of stuff yourself..
Scott Byer NeXTMail: by...@adobe.com
Adobe Systems Incorporated These are *my* opinions, and
1585 Charleston Road, P.O. Box 7900 do not necessarily reflect
Mountain View, CA 94039-7900 the opinions of my employer.
---------------------------------------------------------------------
Sherwood Botsford
>
> In article <n8p...@zola.esd.sgi.com> go...@puck.esd.sgi.com (Michael Gold)
> writes:
>
> > ... And is NeXT going to supply each and every software vendor with several
> > new machines at no cost; for engineering, SQA, and customer support?
>
> Sounds like a good idea to me. Well, at least every software vendor that
they
> consider _significant_. They could lend the machines, not give them
outright.
> They could even do it _before_ they start selling the new machines to the
> general public.
>
Or if they have the courage of their convictions, trade the machines for n% of
the gross sales of Next Software of the company....
Sherwood Botsford
We have RISC 6000's and Nexts, (and this and that else)
The RS6000's sit, unloved, lonely, and furiously crunching fortran code day and
night.
The nexts have people in front of them, usually with a Stuart window or two
open onto the RS6000's
This has actually worked out to be a very good combination; at least until next
comes out with a machine that has 16 R4000's in it. (<evil grin>)
Oh, we do use the RS6000's to play dogfight after hours.
Mark Thomsen
> My primary curiousity is what people, who have developed stuff with NeXTSTEP
> and have developed stuff using other environments and what people, think.
A coworker of mine has watched NeXTSTEP development and has been impressed. A
couple of weeks ago we got together on a tiny project and I wanted to do it on
NeXT to make the UI easy to develop. So he logged onto the NeXT for the first
time. In one day he had his algorithm up and running (image generation and
compression), had results displayed, and was ready to integrate. He now tells
everyone of how easy it is to program. He had done something similar on a Sun a
few months earlier, under X. He swears he will never do that again.
I get this sort of story from high and low, large serious projects to tiny ones
like the one we did. I am beginning to sense that the difference is so huge
(beyond my direct experience and that of the people around here) that there is
a moral obligation to turn people on to NeXTSTEP. A huge number of people every
year are expended writing code to use X. Projects are behind schedule, the
results usually are inelegant, and the potential for writing new applications
is unrealized.
It is not just good business for NeXTSTEP to migrate across other platforms -
it is the right thing to do. This industry has to get moving forward. X (and
Windows) are not steps forward based on all the experience I have at hand.
NeXTSTEP is.
So, how do we get these ports?
Mark R. Thomsen
Christopher Nagel
itself to doing ports of other peoples' software. I contacted a few companies
that had key productivity apps, but no one was interested in having their
software run on the NeXT. (We were also going to provide mass-duplication of
ODs, but Canon was dead-set against that.) Now, Microsoft is getting accolades
for its NT PortShop--and advance bug reports as well!
Perhaps the time for this idea has come...
(BTW In csn.misc there's a poster who wants a satellite app ported...)
Remo Williams
(Michael Zyda) writes:
> Aren't all these questions about neXTStep on the IRIS somewhat
> academic now seeing how the designer of neXTStep is not even
> with neXT anymore but with Sun Microsystems?
No. I assume you presume that Bud Tribble designed and coded all of NeXTSTEP
from IB to the appkit. Guess again.
>Don't we all
> expect neXT to blow away any day with people bailing out
> like their key software designer?
No. Bud was not the key software designer. And in any case NeXTSTEP 3.0 was
done with no involvement from him.
Do we expect SGI to blow away with NeXT passing them by with number of boxes
shipped (certainly not revenue) ? No, of course not.
Do we expect IBM to blow away with NeXT breathing down their necks and passing
them with the number of unix boxes shipped? No of course not.
Do we expect SUN to blow away with NeXT supplanting them in the financial
marketplace? No, of course not.
>
> mz
So its more than an academic exercise--though on usenet its hard not to treat
all discussions like academic exercises.
--
David Williams -- d...@atherton.com -- NeXTMail welcome