> with a Motorola 68030 CPU running at a screaming 25Mhz, [...] this was the "Ferrari" of desktop systems!
Um, no. The NeXT cube competed mostly directly in price and market against MIPS and SPARC boxes which ran rings around it (quite literally 4x faster on typical CPU benchmarks in many cases). It was never a performance platform, and would only fall farther behind as the 68k architecture failed.
There were many things to like about the platform, but really NeXT was a very flawed system. It lacked the speed and features to compete with machines in its price range, yet was priced out of the PC and Mac world where it's UI and integration would otherwise have been attractive.
> It lacked the speed and features to compete with machines in its price range,
Having used a lot of SPARCs and NeXT boxes side by side during that period, I'm scratching my head as to what features you're referring to, at least among machines which directly competed with, say, the NeXTstation Color Turbo. Looking back I'm really amazed at how ahead of its time that machine was.
But the critical item, I think, was the OS. NeXT had NeXTSTEP. Sun had heaven help us all) SunOS and later Solaris. SGI had IRIX. There was absolutely no comparison. I think this is why NeXTSTEP still lives on in OS X, and Solaris, um...
They picked the wrong horse (680x0) at the wrong time (late 80s/early 90s) in a market that no longer exists, that niche between high end personal computers and low end workstations.
This was about the time both MIPS and SPARC were finding their legs and Intel started to get its act together with the 486. In addition, this was the time the first wave of commodity graphics accelerators were on the market and NeXT couldn't scale up the way SGI could.
If they had managed to exploit the ND boards, it may have been a different story and SGI may have failed to secure the graphics workstation market, but they really didn't have much of a chance against Windows and SPARC on technology or HP and IBM on financials.
I was referring to the workstation market, where Sun and SGI were stitching up that market. GX, LX, and Élan were arriving on the market when NeXT only had ND.
3dfx was, for obvious economic reasons, not a slice of the solid, high, or extreme IMPACT graphics that were available on the desktop at the time of its release. It was more competitive with the options for Ultra, but Sun had already conceded the high end graphics market to SGI.
The NeXT hardware wasn't just priced out of competition with PCs and Macs, NeXT wasn't allowed to compete head on with Apple, as part of the agreements that allowed Steve Jobs to leave Apple and start a new company. The workstation market was the only one NeXT wouldn't have to litigate their way into.
http://www.youtube.com/watch?v=uQ-G99rh0p8 - In case anyone has not seen it already, this is Steve Jobs here explaining where NeXT is competing and how it plans to compete (with, by the way, Sun).
The companies who enthusiastically picked up NeXT machines by the hundreds did it because the development environment was a decade ahead of its time. However there weren't enough of those companies, so NeXT was forced to reverse-takeover Apple and from there take over the world.
The early RISC machines were marketed on benchmark results which didn't reliably translate to real-world interactive performance. They got better with time though.
But yeah you wouldn't buy it because of its superior performance vs the RISC competitors, you got it because of the software and the dev tools etc. A bit like Mac vs Windows in the PowerPC days (only with the RISC vs CISC roles reversed!)
Sorry but you're totally missing something if you think you can compare to a Mac or 'PC' of it's time. Not only did it have a lot of special peripherals but the entire OS and environment is what made it special.
Like Apple today, you have to look at the entire package. The difference being that Apple mostly targets the general consumer today while NeXT was targeting a more advanced and smaller but richer market back then.
Missing what? These arguments were all had on the internet at the time, there's nothing we can add. Suffice it to say that the idea that an "environment" can make up for a severe performance handicap is a uniquely modern one informed by the existence of dual core multi-GHz devices that fit in a pocket.
If you were doing CAD work in 1990, you might not be so forgiving of a device that cost almost as much as a years salary for its user. And the market wasn't.
The entry models were always dogs regarding storage. The first cubes had those horrid MO drives and nothing else and the first stations had 105MB(?) drives that couldn't hold a full OS installation. No Shakespeare for you!
Scroll bars are on the left, and the scroll buttons are at the same end of the scroll bar.
NeXTSTEP GUI programs adhered to a well-defined concept of an application: they're stored on-disk as self-contained app bundles that appear in the file manager as a single file, there's one Dock tile per application, one menu per application, and zero or more windows per running application. OS X has somewhat weakened this.
Menus are vertically arranged, and can be re-positioned. Sub-menus can also be torn off and re-positioned.
Unlike OS X's Dock, when you fire up an application that isn't already in the Dock, the app icon goes in the bottom-left corner of the screen (which is also where minimized windows go). The app icons in the Dock are only the ones that you have selected to stay in the Dock even when the application isn't running.
The window close and minimize buttons are in different corners, and the close button provides visual indication of a window containing an unsaved document.
The display is grayscale, but the software supports color.
I used to have a cube with a NextDimension board in it: it was a dedicated i860 with 32MB of backing store. Whoa. 24bpp color with 8bpp alpha. Hot shit!
Unless I'm misunderstanding what's being described here, I believe they were kept up through 10.6. I'm running 10.6.8 currently, and there are two scroll buttons at the same end of the scroll bars, as in NeXT. But from 10.7 the scroll buttons were scrapped entirely.
Whether to put the buttons together or at opposite ends has been an option for as long as I have been using OS X, but I don't recall the history of the default setting.
The other scrollbar-related behavior is what happens when you click on an empty part of the scroll bar. On NeXTSTEP, the scroll thumb jumps to where you clicked. On Windows, it scrolls down one page/screenfull. OS X has defaulted to the Windows behavior for at least the past several major releases.
What you're referring to as "Windows" behavior is actually Mac behavior.
Smalltalk scroll bars were hidden entirely when the mouse pointer was outside a scrollable view. When shown they were on the left, they were proportional and went up/down with a left/right button mouse press. The code for this is probably still in Squeak somewhere - it's certainly how old versions of VisualWorks Smalltalk worked.
Lisa made scroll bars always visible, moved them to the right, made the thumb/elevator non-proportional (users didn't get why it changed size), and added arrows to the ends - and then added page-scroll buttons too. Lisa also came up with the resize corner, rather than resize-by-any-edge or resize-by-explicit-command.
The Mac streamlined this further by getting rid of the page-scroll buttons and making it what happened when clicking in the bar background. Later the Mac also came up with zooming a window to a proper size to fit its contents. (Which isn't the same as maximizing it to fill the screen.)
NeXT moved scroll bars back to the left, put the arrows together at the bottom, made them proportional again, and made clicking in the the bar background scroll to a point.
Mac OS X kept the arrows together and proportionality but matched the positioning of the Lisa and Mac. Scroll-to-point instead of scroll-by-page remained an option in System Preferences though, and holding the option key inverts the behavior.
OS X Lion hid the scroll bars once more, at least for users with touch-scrolling devices like trackpads and the Magic Mouse, and got rid of the arrows.
I had a Microway that I ran NeXTSTEP / OPENSTEP on that was an amazing machine at the time.
I loved the left scroll bars and still feel the menu on NeXTSTEP is better than the Apple menu (UI laws be damned) because I didn't have to constantly scroll to the top of the screen for menu options. The ability to tear off parts of the menu was amazing.
The Library app was something they should have kept. The NeXTSTEP file view / shelf still is better than Finder.
There was a lot of awesome things about NeXT machines and NeXTStep but there were also crappy things. The endless list of nested menus for example was nice in that it was standard and you knew what to expect in every application, but also could be less than ideal in some applications.
Nostalgia favors remembering the good things and forgets the frustrations of daily usage.
In the end NeXT failed not because the machines were too expensive (it was competing against Sun, DEC and SGI workstations which were also very expensive, so the price was not really out of line in that context -- only if you compare it to PCs, which you shouldn't.) It lost for a variety of other reasons, and I'm sure if you talk to 20 different people who placed the PO's for workstations in the early 90's you'd get 20 different reasons.
A lot of the times it just came down to "Our guys are used to SunOS/Irix/etc. and don't want to adjust." Or "Our software works best on X." or only works on X. etc.
Well, NeXT got bailed out by being bought by Apple. I don't think you can really call it success. I mean you can call it that but it's not by any yardstick I'd use. It's like you're in last place in the race and your Mom hands you a trophy.
NeXT basically took over Apple from within. Their technology ended up in basically every Apple product, nearly completely displacing what was there before. Their top people became Apple's top people. Their CEO became Apple's CEO.
Apple watchers from the time often joked that NeXT bought Apple for negative $400 million. Buying NeXT is what turned Apple from a perpetually "beleaguered" niche computer company (remember that word?) to one of the most valuable companies on the planet.
The modern Apple, Inc. is much more NeXT Inc. than it is Apple Computer, Inc.
Subsequent disclosures suggest that Apple had screwed the pooch over Taligent/Pink, and was desperately shopping for a next-generation OS. The contest was down to NeXTStep, with Steve Jobs, or BeOS, with Jean-Louis Gassee. As I understand it, the BeOS folks held out for too much money (Apple was much closer to broke than they realized) while Jobs settled for power. All else is history.
If Apple's successive post-Jobs CEOs (Sculley and Amelio) hadn't maneuvered the behemoth into a blind alley, NeXT would indeed have faded eventually. But with full access to Apple's resources, the NeXT team were able to turn things around. Sitting on the top shelf in my office, gathering dust (I turn it on every couple of years) there's a cube. A nice shiny silver cube that runs something that looks eerily like NeXTStep, if you squint at it, with a classic MacOS emulation bag on the side. If that's not a statement by Steve Jobs (and a trophy of success, to have the elbow-room to make such a statement) I don't know what is!
Pink and then Taligent definitely caused Apple to lose traction, but Copland put them in a serious skid that forced them to choose between Be and NeXT.
While Apple saved NeXT, it was a near thing for both companies. Without the 'loan' Jobs was able to secure from Microsoft, it's doubtful Apple would still exist as a separate entity. Whether it was Sun, Oracle, or another company, it would probably remain today as an object framework and not much else.
The Microsoft investment wasn't financially significant at the time; the main benefit was the end of the IP wars and the agreement to keep selling Office for the Mac.
It was absolutely financially /significant/, but you're correct that the software support helped shore up investor confidence.
Remember at the time that Apple had been running at a loss for the last year or two before this and had started to cut R&D funding. This is when Apple was struggling to hold onto about 2-4% of the personal computing market, dropping from about 10% just a year or two before. They had experienced a precipitous drop across the board, in technology, sales, and financials.
It was quite feasible to entertain thoughts of takeovers, hostile or otherwise, from Sun and Oracle. Either of them (and MSFT) had the ability to make it happen at the time. Oracle priced the effort at about $700m but expected a worst case scenario of about $1.5-2b. The motivation was mainly a "FU" move by Ellison to put Jobs back in control of Apple, but floated as a way of accelerating their JavaStation and client computing vision.
This was also when Apple laptops were burning people before burning people was cool. :-)
150 million dollars wasn't going to make of break Apple at the time. Amelio and Andersen floated a bond issue of many times that amount the past year to resolve cash flow concerns and 1997 was when Apple regained profitability. It was mostly a move to settle investor confidence by signifying Microsoft's bet on Apple's future success.
That has had me rethinking your what-if. I don't think the unix is a big deal - the Be and Haiku terminals feel unixy, and BeOS offered a big subset of the unix system calls even into v3 (time of Apple purchase).
By v4.0 (two years after), BeOS had become a remarkable system for single-user, close-to-the-metal computing. The user experience was very fast and stable. The system gave a skilled developer easy access to video and audio.
I've been trying out haiku again this morning. They've done good work. It now boots on my thinkpad, but, like OSX or Windows, feels sluggish in a way that Be never did. It needs some serious performance tuning. They need to get their out-of-the-box development experience sorted out too. OSX has done an excellent job of that.
I really think it was the NeXT engineer's tech demos showing that they could run Openstep on different processors and also that they could complete the 'Blue Box' in time.
Actually it was the Newton that made Apple loose so much money. It was never profitable with an average annual loss of 1 billion USD. But if it wasn't for that loss, Steve Jobs would never have been requested to be a consultant at Apple and turn himself into the CEO later.
The Newton was hardly a money-maker, but IIRC, Apple invested somewhere north of $500 million over the course of the Newton project, and sold about $150 million in units. Many of its sales were the last models, notably the MP2100, largely because these units had such impressive processors for their time (162MHz StrongARM). That is, the Newton got Steved just when it started picking up.
More importantly, as part of the Newton project, Apple invested $1.5M in ARM, and made some $800M in profit from that investment. That ARM investment and early Newton-driven design is still reaping dividends in Apple's current low-power devices.
Fuck you. The Newton was a HUGE(!!!!!!) loser. You are just too fucking lazy to look it up. I was working in the 90's. I remember this and I'm not going to waste my time learning you easy to find information. FUCK OFF AND DO YOUR OWN HOMEWORK.
OT: I wish their was a way to delete my fucking account. This site is a bunch of armchair CTO's who don't know their eye sockets from their assholes.
Citation? This is an interesting factoid, so I googled for about 10 minutes, and couldn't find any numbers on how many total newtons were sold, and how much apple lost on them. Really interested in any supporting data you might have.
That link suggests a total loss of about 750 million over 11 years from start of development to discontinuation. That's an order of magnitude less than the 1 billion annual loss you originally claimed.
Wasn't me - I actually was really interested in your comment (as I indicated).
The link you provided, btw, states, "Although the exact scale of the Newton losses remains unconfirmed, some sources say that the company sunk a billion dollars into the Newton and recouped only about one quarter of that amount in sales."
"Some Sources" might have merit if that was the Wall Street Journal, or New York Times (though, in both cases, those "sources" might have ulterior motives") - but they aren't useful in this context.
Also, as a later poster indicated, losing $750mm over the lifetime of a multi-year project, is a lot (lot) different than losing $1B/year.
I'm still intrigued, and would love if (anyone) could track down some data on the Newton, and it's losses. Or even how many units it actually sold.
NeXT got a second chance handed to it, and then took advantage of it to build the world's most valuable company.
It's similar to the break Microsoft got when IBM inadvertently handed it the PC OS business. It was a great opportunity, but not every company would have leveraged that opportunity to dominate computing for two decades.
If Jean Louis Gassee got the same opportunity for BeOS that Jobs got for NeXT, do you think he would have gone on to create the world's most valuable corporation?
When I was at CERN Apple was giving sessions selling UNIX heart of Mac OS X as a way to attract UNIX fans to the platform, which many enjoyed given the OS's pedigree.
Not sure if the same would have worked wit BeOS as base, although I really enjoyed it as OS and follow Haiku's development.
To be fair, the NeXT machines were not as fast as competing RISC boxes from Sun, HP, IBM and DEC, and were equally costly. Also, as you pointed out, it was not Unix-ish enough. It was the most Mac-like Unix machine at a time being Mac-like was not an advantage in the high-end computing market.
> (it was competing against Sun, DEC and SGI workstations which were also very expensive, so the price was not really out of line in that context -- only if you compare it to PCs, which you shouldn't.)
Except that it is the only comparison that ended up mattering. PC obliterated the workstation market.
My two favorite days with my 030 cube were similar to those with my boat. The day I bought it and the day I traded it for a 3/80. The cube, that is. I traded the boat for cash. :-)
ISTR Next ads in Byte, and it being targeted at the high-end of the PC market. I had one at Chicago and loved it, but only because I didn't have to try to live with the WORM as my only storage. Great computer, great, great OS.
Those who want to experience NeXTSTEP without the hurdle of installing a VM should try GNUstep. This project aims to faithfully recreate the NeXT's software environment, with the omission of Mach and its IPC mechanisms.
The problem is that their development is almost stalled.
I hardly see any difference since 2004, the last time I used WindowMaker as my desktop.
As for GNUStep the web site wiki is very confusing, as it is not easy to find out how close they are for Objective-C language version and NeXTSTEP/Cocoa libraries.
Again, many of the pages look the same as back in 2004.
WindowMaker split off from the GNUstep project quite some time ago, IIRC. They don't even use Objective-C, they use their own framework for widgets built on WM called "WINGs."
There's still activity on the GNUstep front. Just a few things I can think of,
- Their runtime, libobjc2, has done a great job keeping pace with Clang and modern Objective-C. It now has a Cmake based build system and doesn't require any bootstrapping with GNUstep-make.
- GNUstep-base (the Foundation alternative) is solid, though some parts go back as far as OS X 10.6 and others OS X 10.4. You'll have to check the headers to see what's up to date. They do have an extremely capable implementation of distributed objects, last I heard they were planning to make it integrate with Apple's distributed objects implementation, but that project still seems to be pending.
- GNUstep-gui (AppKit alternative) just added support for 10.6's NSCollectionView very recently. Most activity there has been in the form of refinements. As I recall, drag and drop is still on their todo list... they really could use some manpower there.
- GNUstep-back (the backend renderer) still uses an API that's largely based on Display Postscript. They are planning to pitch a GSoC project to replace the Display Postscript API with one based on GNUstep-Opal (a Core Graphics alternative), and they already have a largely proof of concept Core Animation alternative running on it (GNUstep-QuartzCore.)
The project's current vision is to focus on making it easier to port Objective-C to other platforms, rather than to be a desktop environment. But like most small, mature open source projects, there's a lack of manpower keeping them from easily achieving those goals. Would be nice to see more activity there, in the future.
As for the current website, I think Scott Stevenson designed it a decade ago long before he joined Apple. I like the design, personally, but I do understand that it's a bit difficult to navigate... and several of the docs haven't been updated in some time.
1. They aren't going for "faithful NeXT", they are going for modern Cocoa.
2. Holly cow, I didn't know it could look nice (screenshot on the link). I mean I did, but all of the screenshots in the RSS/email announcements on gnu lists have a theme that looks like the screenshots in the article!
I owned a NeXT Cube (since sold) and a NeXTStation Turbo Color, which I still have and which still powers up nicely. Monitor and NeXT laser printer still work as well. Pretty impressive for 20-year-old hardware. And still a beautiful, sleek design, if you can get past the 100-pound color monitor. Owning a NeXT in the early 1990s and the first days of the Web was like having super powers.
Ah, the memories, the memories. Yes kids I was there and it was beautiful. My company built systems for financial traders. There was always a battle between the techs in the banks who wanted something architecturally elegant e.g. object oriented and the traders who wanted a UX that worked in an incredibly high pressure environment. In some deals the techies won and in some deals the traders won and in both cases the bank lost. NeXT was the only thing that was equally good at both. We ported to NeXT and sold a few. The rest as they say is history (i.e. NeXT disappeared into the dustbin of history and remerged in Jobs's brilliant second act. Today in our real time social mobile web we are all like traders, so that is why we have NeXT 3.o aka the iPhone.
For those who'd like to get a sense of what the NeXT desktop was like, there's the WindowMaker window manager for X11.
I've actually been using it for a decade and a half (trying others from time to time as well). It's very light and fast on modern hardware, the codebase is remarkably stable (development has just re-started after stagnating for 6-7 years), and it's configurable (I've got numerous hotkey shortcuts defined which I miss terribly when I'm in another environment).
In a world where I'm coming to appreciate calming interfaces, WindowMaker is very much one such. I've only recently discovered Readability (similar to InstaPaper or Pocket / Read It Later), and there's a certain similarity of spirit between the two.
Also, as others have noted, Cocoa for OS X contains a lot of NeXTstep-isms under the hood.
I have been searching for, and failing to find a quote by Steve Jobs - I remember reading it in the late 80s. It was to the effect "UNIX is going to be the operating system for the 90's". I'd love to track down the original source for this...
There was ad by NeXT -- "In the 90's we'll probably see only ten real breakthroughs in computers. Here are seven of them." that lists making Unix usable as one of them. I have this poster framed in my office.
I wouldn't be surprised at all if he'd said that; but you might be thinking of Gates who once said that OS/2 would be the operating system of the '90s. (He was wrong but he… still did all right in the end.)
The Windows that is related to OS/2 is NT, which was very much not the operating system of the 90's, since it didn't go mainstream until NT5.x: Windows 2000 and Windows XP.
Aside from the beauty of this restore, I thought it was cool to see the UI influences the NeXT's OS had on Mac OS. I've never seen screenshots until now, but you can definitely see aspects carried through to its successor.
When looking at the terminal screen shot, I can't help but think of the classic "csh considered harmful" (http://www.perl.com/doc/FMTEYEWTK/versus/csh.whynot) (Which is itself a classic "meme" before "memes" were a misused "meme").
> All the NeXT components are displayed here, showing a clean and well engineered assembly, demonstrating a design asthetic that extended to the inside of the NeXT cube computer.
Some machines are really nice inside. IBM (and some Lenovo); some HPUX machines; etc; have really nice well laid out insides.
It's a shame that this approach is dying, even though most people don't need it and benefit more from semi-disposable hardware.
I'm not a mac fan, nor a mac apologist, but I have to say the current Mac Pro beats all of them ... that is an extremely well designed desktop computer. Really a joy to work with.
For runner-up I would say the SGI Octane2, and for a non-desktop machine, the Sun E4500 was an extremely well designed, well manufactured system. Also a real joy to work with (although weighs something like 100 lbs fully loaded)
I hope the Octane2 was better than the Octane. While I like mine, it's really loud, the incandescent bulbs (!) in the light bar don't last very long, there's no room in the large case for a single 5.25" drive, the plastic skins are really fragile, the compression connectors are very delicate, the texture ram likes to unsolder itself, and certain generations of circuit boards fail at a very high rate.
From what little I've seen, Sun gear of a similar vintage seems better built than the Octane.
And a great hardware, with the sliding boards and the pressure connectors. Its only culprit was the missing 5.25'' slot. I still have a dual R12K MXE Octane too :)
I did some NeXTStep programming 20 years ago. I converted the Tetris game to color when they finally shipped the NeXTstation Color. If someone has the Tetris source, I'd like to look at again and see how close it is to the iOS stuff I'm doing these days.
If you are using a monochrome system, you should copy the Tetris
blocks from MonoBlocks to this directory. You can design blocks of
your own by using Icon and saving them as Block1, Block2, Block3 and
Block4. Be sure to save you images as 16x16 files. NOTE: Icon will
open(and save) the Block images as 48x48 so you will have to change
the image size back to 16x16.
Also, if you would like to design a color icon for Tetris, please
NeXTmail me the TIFF at the following address.
melling@cs.psu.edu
Also, I'm interested in hearing about any features that you would like
added.
If this whets the appetite of anyone in the Seattle area, I have a couple of pizza boxes (one turbo color) with soundboxes, a mouse, keyboard and monitor all looking for a new home.
All this is most likely non-functioning, but anyone who wants it is welcome to it. I can be contacted at gmail.com
NeXT was way out of my own price range in the early 90's (I only had an Atari ST). At our university there were mostly DECstations and Sun workstations, apparently due to huge discounts (50%+) these vendors gave to academic institutions and because they were also used as servers (with NCD X-Terminals as clients).
The amazing design and focus on graphics power may have hurt sales in these circles because it would have been hard to justify a high price for eyecandy over raw cpu power and most universities outside the US were probably not adequately funded at that time (perhaps Jobs overestimated that market?).
I bought my first Mac a TiBook when OSX was still in beta because I knew it was based on the OS that the world wide web was invented on. I've got the NeXT Bible at home on a shelf. I've been looking to buy a cube for years and have only found them for $5k+ on ebay. I know they're rare but does anyone have one they want to sell for a reasonable amount or know of a place selling them?
Lots to love about NeXT. I got a kick out of the display PostScript screen. Very few other computers or workstations ever took that route. http://en.wikipedia.org/wiki/Display_PostScript tells a bit more about how it evolved into PDFish displays in Mac OSX.
Not to be needlessly pedantic, but MazeWars was originally on Macs circa ...1985? I know we played it in the dorms at Virginia Tech over phone-net (some homemade!) connectors in 1986-7. Pretty sure I had seen it long before then.
All I could think about as I was looking through the pictures and reading was, "Wow that is a good looking computer." Perhaps it's the designer in me, but it just has a completely different feel to it.
I remember going to their store on university avenue (palo alto) which is think today is the location of an apple store. I used Apollo workstations at the time, but the NeXT boxes looked really cool.
Um, no. The NeXT cube competed mostly directly in price and market against MIPS and SPARC boxes which ran rings around it (quite literally 4x faster on typical CPU benchmarks in many cases). It was never a performance platform, and would only fall farther behind as the 68k architecture failed.
There were many things to like about the platform, but really NeXT was a very flawed system. It lacked the speed and features to compete with machines in its price range, yet was priced out of the PC and Mac world where it's UI and integration would otherwise have been attractive.