Slashdot Mirror
Copland/Gershwin vs. NeXT
Etcetera writes "David K. Every (of MacKiDo fame) has written an interesting article at iGeek about Copland vs. NeXT and the decisions that Apple made back in '95-96. Although most agree that bringing Steve Jobs back was a Good Thing, a lot of cool Apple-invented technologies got left by the wayside without a fair shot at proving themselves once NeXT came in. Was it always the right call? Functions as a cautionary tale about management vs. engineering as well."
I worked with a Software Engineer who worked on OpenDoc, which was about the only good thing aside from QuickTime that Apple had put out in that era. Some of you might remember the demo of an ActiveX control running inside an OpenDoc container. When NeXT took over, they hacked OpenDoc to run within NeXT's ojbect model. The Steve (himself) just said no, NeXT's object model is better. That's when my friend left Apple. Can't say that I blame him. OpenDoc had a lot going for it.
Karma: It's not just a good idea. It's the law.
Another good one about how a project evolves into a "death-march", but it's probably preaching to the choir here.
The tragedy is that those who most desperately need to learn the lesson presented here (that would be the uninformed, inflexible PHBs) will never read this kind of article, or if they do they will never realize that we're talking about you, here!
It's just amazing how many technology "managers" cannot comprehend the silliness of mandating a schedule and a feature set. You might get it by then, but it will be a piece of #@%$.
"Send an Instant Karma to me" - Yes
"(A common engineering practice is to take your best guess at how much time it will take, and then multiply that by 3)."
that way, you look like a miracle-worker =] (or so Scotty told LaForge)
Still, it would be an interesting universe had Be been chosen over NeXT. the Be APIs were super nice, but from what I heard they had a very hard time in larger projects. I think Gobe Productive was the largest app ever for BeOS. I used BeOS exclusively (when i had a peecee) for about a year, and I loved it. All the applications felt oddly 'light', I think from the quick responsiveness of it all.
The Best OS is Dead! Long Live the Best OS!
"I would say that 99 per cent of what my father has written about his own life is false." - L. Ron Hubbard Jr.
Given a choice between OLE and COM on Windows (95% of the desktop market) and OpenDoc on MacOS (5% of the desktop market), most developers chose OLE and COM for their components.
So Carbon on a new kernel (NuKernal) was done long before the rest of OSX was ready. It took years to get NeXTSTEP and Cocoa and the rest of the OS time to catch up
This is just absolutely false. The NeXT kernel (Mach/BSD) and Cocoa were ported very quickly. Apple bought NeXT at the end of 1996, the first Rhapsody developer release was ready less than a year later, and Mac OS X Server 1.0 shipped in early 1999. The reasons why the "real" Mac OS X took longer were that Apple had to implement Carbon for developers unwilling to convert to Cocoa, and write a brand new display system (Quartz) after Adobe dropped Display PostScript.
NeXT delivered on its promises, it's just that Apple's requirements changed. And it's also worth pointing out that Mac OS X is a far better system than what was envisioned for Copland. Aside from the much better adherence to standards, it is a much cleaner architecture. From what I remember of Copland's documentation, it had a weird form of partial memory protection where the entire GUI ran in a single process, so any app could take down all other UI apps, although server processes would be protected. The transition to a fully buzzword-compliant OS wasn't going to happen until Gershwin, and I seriously doubt that could have shipped by now.
How to solve most of our problems: 1.Lots of nuclear plants. 2.Cure aging.
This article is one big rant. I wish he had more information to back up his claims. The article just seemed to be overly bitter calling NeXT "liars" and claiming that the engineers were hit with unrealistic expectations.
Hey, I'm an engineer and I think it takes everyone on a project to make it fail. On successful projects you have only 10% of the people doing all of the work and fixing the other people's mistakes... most projects succeed in spite of bad management or bad subsystem X.
Mac OS was on a death march because of the fundamental underlying technologies and a culture that was stuck in only one way of doing things... and windows is on a similar death march. OS X takes technology from the UNIX way of thinking and from the NeXT way of thinking to make a platform that is a developers dream. I think this is what puts OS X beyond Windows and will eventually lead to new "killer apps" that will save Apple. I really doubt that if Apple had went with Copland so many alpha geeks would be flocking to Mac... I doubt slashdot would have added an Apple page or O'Reilly would have a macdevcenter.
There are 10 types of people in this world, those who can count in binary and those who can't.
Ain't hindsight great? It allows you to sound wise without producing any testable arguments.
I would remind people that David, in my mind, is a Mac zealot. His old Mackido website was full of vitrole and over-advocacy. (Mind you, it was a great place to find a computer joke, but I took little of the site very seriously.)
As I mentioned in other topics, Mac OS 9 users are among the most stubborn to change their way of doing things on a Macintosh. That's understandable--the original Mac OS was easier to use than most operating systems and developed quite a fan following. This resistance to change, however, causes finger-pointing and blame-making over a matter that, now, isn't really a point of conversation anymore. I think David falls in this trap of "Apple's changed with OS X and I don't like anymore." I also give this problem a name: whining. I'm a Macintosh advocate by trade but I've never been fond of zealotry--its a blinding thing in helping a customer.
I disagree with David about his being "sick of hearing that Copland project failed because of engineering. From what I know of engineering and people inside of Apple, it was mostly because of bad management decisions inside of Apple." He makes the incorrect assumption that the technology was sound enough and needed only enough money and time to push it through. Apple creates the environment that R&D works in, true. But faulty engineering is faulty engineering. Steve Jobs likely killed many projects not only because he didn't find them practical but because they just plain didn't work or didn't follow the business plan. Seedless corn or instant water sounds like a cool idea, but a bad idea is a bad idea.
Apple nearly died because they didn't have a business plan. Does David want Apple to revive old projects at the cost of the company's existence? Makes no sense to me. I think David is grousing over spilled milk.
We were all uncertain about what became OS X, and, thanks to a strong business plan, focused R&D, and listening to customers and developers (things that Apple did very badly in the '90s), Apple has a sound product with a good future.
Vos teneo officium eram periculosus ut vos recipero is.
As I remember the events of the summer of 1997, Apple would have fallen apart if Steve Jobs had not come on board. At the time, people were leaving Apple left and right. Scott McNealy suggested that Apple should abandon MacOS and start selling java based network computers. The pundits were suggesting that Apple should abandon all things Mac and sell PC clones. I like OS X. I use it every day, but I have to wonder if Apple hadn't abandoned Copeland, if today I would be using an operating system more in the spirit of the original Macintosh. OS X feels like Unix with a bunch of pancake makeup on top. Instead of being simple like pre-X MacOS or BeOS, it hides its complexity from most of its users. For all the strength that OS X brings to the Mac platform, it brings along a fair amount of baggage too. It just doesn't feel quite right. There is something disharmonious about it. I'm having a bit of trouble explaining it, but I wonder if an OS built from the ground up for the Mac wouldn't have been better,
There is nothing important that Smalltalk has, that Objective-C lacks.
In fact, Objective-C is superior to small talk for its given market: compiled applications that can be shipped in shrink wrapped form.
Call C++ a pale shadow of smalltalk and I agree with you, but don't knock Objective C. Its wonderful, and has no problems or poor aspects that from what I see.
There's a lot of whining about OS X and I just don't get it-- it is far and away the best OS I've ever worked with as a user, or developed for as a developer. Far better than Classic Mac OS, better than windows, better than Unix.
I think people want to complain about some idealized impossibly perfect idea of what something could be... but in terms of reality-- shipping products-- its the best there is right now, and well ahead of everything else out there.
Yeah, and you guys panned the ipod too: http://apple.slashdot.org/article.pl?sid=01/10/23
One of the reason why NeXT failed is that because of Motorola and then because of Sun and HP.
The Motorola was because they could not get a new 68K chip fast enough.
The Sun was because they went from an OpenStep based system to a java based one. HP decide to brake their deal with NeXT right after that.
OpenStep was running on both SPAC and HPPA before Apple bought NeXT.
Also before Sun went the java route, they bought a company that was making an office suite for OpenStep (see they owned an office before StarOffice).
The Mac without the Unix underpinnings would still be relegated as "toy" OS and its marketshare would be declining rather than climbing (or at least stagnating).
Unix gave the Mac credibility from some key market segments. It gave the Mac mindshare. If we chirped about OS 9 having preemtive multitasking, people would've said "about time" and rightly so. As it stands now, the Mac is now buzzword compliant and, more importantly, it has the time tested core of Unix and all of its familiar tools that scientists and sysadmins love.
Most frustrating part:
Apparently the author is too stupid to realize Apple was sinking. Fast. It took a major acquisition to prove to the board and shareholders that positive steps were being taken towards updating what was then a very unstable, inefficient Mac OS.As a Mac user since the early 1990's, I can honestly say that the 2nd-generation Power Macs (the PCI ones like the 7200/7500/8500), in the System 7.5.2 - to 8.0 era, were horrendously crashy and pricey. (I still used them because the Windows UI pissed me off, but I was beginning to envy my Wintel friends every time my 7200/75 locked up)
OS 9, the iMac, the legacy-free towers were all great products which had little to do with the NeXT team. While it's likely true that Apple's own engineers bailed themselves out of the mess they had gotten into under Sculley's leadership, it was too little too late in the eyes of shareholders and consumers.
In short: if they couldn't demonstrate that they were going to leapfrog Windows in terms of stability, Apple was dead in the water. Apple's own engineers likely had lots of credibility with Apple management, but that was not enough. Even if Copland was only 6 months from completion, Apple was in grave danger, and was wise to purchase a proven technology. It may be unfortunate that business success is a necessary evil in the development of software, but I wouldn't go calling Apple's management stupid. They bit down and did what had to be done.
Well, I did a lot of the OpenDoc design work, and yes we did file some patents. They were mostly about UI issues, particularly how the composition & layout model and event handling was done.
The spec mentioned above was the official stuff. Looking back from a vantage of ten years or so, there are a lot of things I would change, but the basic design is still sound.
Actually, my current favorite in the space of such things is KParts, which does a lot of the same sorts of things OpenDoc was meant to do.
I worked for Apple in the mid-nineties, when the PPC was new. I watched the entire (almost) death-spiral with particular interest.
I think the article is good, but it is only half the story. Apple is, and always has been, a hardware company. The thing that really screwed Apple during the Gil Amelio years was a total lack of hardware engineering. Apple tried to become a "beige box" company, and tried to have a solution for every problem. It just wouldn't work.
When Steve Jobs came back to the company, there was something like 85 SKUs for hardware systems. For those who aren't familar with marketing/retail, that's a lot. It wasn't like "You can have a 7200 with X Y or Z memory", it was "You can have a 7200 with 16MB, or you can have a 7200 with 32MB or you can have a 7200 with 64MB" and the vendor would have to stock all three. Impossible.
Plus, none of the hardware was exciting. People just plain didn't want to own it. It was like, sure, I can get a beige box that runs 7.5.2, crashes all the time, and has only Word 6 (god, what an abomination), or I can spend half as much, build my own Wintel, and get the newest Office.
The main thing the reintroduction of Steve Jobs did for Apple was put a single vision back in charge of both hardware and software. Even if Copland was further along than NeXT, it was hopelessly mired in a hardware development cycle that was just flawed. One of the main problems Copland faced was not only the need for backward compatibility of software, but the need to support 85 different configurations of non-industry-standard hardware. Impossible. Anyone remember the "Enablers"? God, what a mess.
One of Job's most controversial moves, and perhaps his smartest, was to draw a hard line in the sand at the G3, a processor that was barely even shipping when he announced the spec. Thus, he promised hardware compatibility only back to the currently brand-spanking new machines, guaranteeing that, at the end of a 3-5 year development cycle, the OS would only have to support hardware 3-5 years old. Man, did people scream and moan ("But I just bought an 8600/120!"), but now Apple is back where it needs to be. One of the biggest complaints over Win2K was its trouble with older hardware. MS was able to make Win2K fly by (1) not pitching it to home users, who were more likely to have funky sound cards, and (2) providing a lot of expensive support to hardware manufacturers to write compatible drivers in time for XP.
One pre-Jobs hardware move that Apple took that is now reaping benefits was to eschew its own good but expensive standards for adequate but cheap industry standard. Internal SCSI 4X CD Roms gave way to ATAPI; NuBUS gave way to PCI; etc. This made it even easier for the OS developers to support hardware.
Anyway, in order to understand the whole Apple picture, you have to consider the wretched state the hardware side was in in 1996, and realize that, even if no one bought the cube, they have come a really long way - and that is what made OSX possible.
I'm a lawyer with excellent karma. Something's gotta be wrong.
For those of who weren't programming Macs back in the Day (remember APDA?), Apple had an award-winning quarterly technical journal called "develop" which had lots of neat articles in it with a fun and offbeat tone. Sort of like O'Reilly and Associates has now. They stopped putting it out in '97 amid all the hemoraging, but all the issues are available here.
Anyway, there's a pretty informative article explaining all about Copland here.
Hire a Linux system administrator, systems engineer,
Here's the archive of Apple's official Hotsauce (aka Project X) homepage, which is what you're thinking of. Yeah, even Yahoo was up on the Meta-Content Framework deal.
V-Twin was the basis for the summarize feature, I think there's still an SDK up for it.
Personally, I liked PowerTalk. A system-wide, integrated mail and collaboration framework with a standardized mailing interface.
Even if the implementation is done with LDAP, PGP, and Sendmail now, I really wish they'd bring back those APIs for application code. The "digital signing" concept integrated into the API was better integrated than Apple File Security in Mac OS 9, and better than anything in OS X right now. Remember being able to verify a document's signiature from the Get Info window? The Keychain is the only concept that survived.
QuickDraw GX ( here or here) was WAY ahead of its time. Although a lot of its features found their way into ATSUI with Unicode, QDGX still had soul. I still don't know of any program that can do all the really really fancy and obscure ligatures properly.
Probably my FAVORITE technology was/is Apple Guide. There's nothing quite like a help system that draws coach marks on the screen when telling you how to perform a step. That plus hilighting proper menu options in red, and the fact that it wasn't glacially slow once it became PowerPC native, made it a really amazing conversion tool.
Balloon Help was quirky and fun. I don't really understand why they replaced something so cool with something as lame as Tooltips. =(
QuickDraw 3D and Apple Data Detectors had some cool concepts too. Maybe we should get them to re-write Data Detectors using Perl regexps
Arrgh.. the OS that Could Have Been... Try the old Apple Advanced Technology Research Group website for more stuff.
Hire a Linux system administrator, systems engineer,
I posted it above, but take a look at the Copland article in Apple's develop Tech Journal.
Hire a Linux system administrator, systems engineer,
Yes, V-Twin (AIAT) is still alive in OS X.
Objective-C has garbage collection, in fact, it has the best implementation I've used so far.
When objects are autoreleased, they are GCed at the end of the event loop.
This is much better than, say, Java, where you can never predict when GC will occur (and I've spend countless hours trying to get a product to work even though it will randomly stop and check EVERYTHING IN MEMORY to see if it should be deleted.... when doing a large server with gigabytes of objects in memory this impact is a major PITA) With objective-C memory management is semi-automatic. A better solution than Java and FAR better than C/C++.
I don't know what you mean by "real exceptions" but basically, every languages has its differences and the person I was responding to was just wrong in saying that Objective-C is a pale ripoff of smalltalk.
Yeah, and you guys panned the ipod too: http://apple.slashdot.org/article.pl?sid=01/10/23
Not all Java implementations have mark and sweep GC, which is the primitive form of GC of which you speak. However, there are still plenty of reasons to dislike Java beyond that. ;)
ObjC also has a primitive form of GC, even more primitive than mark and sweep. The best GC you can probably get is a generational system, not surprisingly found in most (all?) current Smalltalk implemenations and good Common Lisp implemenations. You get the best of both worlds- unlike the GC in ObjC, you're not stuck looking over your sholder to deal with it's primitive form of GC (found also in Python still?) and unlike Java, it doesn't do one huge mark-and-sweep at any given time- it's always doing a little GC.
Working toward a usable PDA environment in the spirit of Newton OS: Dynapad