There is nothing wrong with working help desk in general; it's an important job. But there is something wrong with working help desk in your 30's from your own point of view if you got into the field with visions of making billions instantly with only minimal knowledge of computers. The elitism here is the elitism of the people who chose a career in computers without the right preparation.
Linus is talking about open source software development as a career path, not system administration or help desks or other standard IT functions. Open source development is special and different from proprietary software development. Administration or support are not (there is just a little less of it).
The difficulties with installing new software using a linux-based operating system arise when you want to install software from elsewhere
That may or may not be difficult. Even if you want to download applications, in many distributions, all you do is download an RPM file and double click on it.
But, importantly, you don't have to do that in Linux at all: if you run a decent distro, all software should be installed through the package manager. That's where you search for applications and tell it what you want. That's the easiest way of installing and maintaining software: you don't have to download anything at all, and you never have to worry about it; all you do is tell the computer what you want.
In fact, most people don't even have to do that: when you install one of the major Linux distributions, you pretty much get all the software you would ever want or need preinstalled: office suites, desktops, PIMs, mail readers, web browsers, scientific software, desktop publishing, image editing, etc.
The report is also wrongly optimistic about software installation on Windows or Mac OS X. On Windows, most software comes with installers. Macintosh, on the other hand, is far messier: not only do you have to download the software and extract it, after that's done and made a mess on the desktop, you have to figure out how to proceed: maybe you open a folder or a virtual drive and drag something to your application folder, maybe you run an installer, and if you run an installer, it asks questions about "what disk volume" you want to install the software on. Macintosh software installations are the most confusing of the bunch. And neither Windows nor Macintosh will automatically update software that wasn't shipped by the OS vendor.
If the main complaint about Linux is that software is hard to install, then Linux is at this point perfect for the desktop. And that's, actually, what I find with my "mom and dad": they have trouble dealing with Macintosh and I have ended up having to install software for them, but with Linux, I just put the machine there and it works, every time, all the time.
Um, no. They were making a product for the market existed, and even then, it *barely* existed.
It was a market that was going to exist whether or not Apple grabbed the lead by putting out a hyped up, heavily marketed, pretty-but-inferior product. In parallel with Macintosh development, there were several other companies that were developing inexpensive GUI-based personal computers. Barely a year after the Macintosh, Amiga started selling a machine with pretty much the same specs and price as the Macintosh, but with hardware graphics acceleration, a preemptively multitasking 32bit operating system, and a really well-designed microkernel. Workstation efforts were in full swing, which yielded graphical workstations that were no more expensive and far more powerful than a Macintosh shortly thereafter.
I suppose Apple employees genuinely believed the marketing hype of their own companies that they were the salvation for the "rest of us" and that the world came down to Microsoft and Macintosh. But the world only came down to Microsoft and Macintosh after Apple had driven all the other, innovative personal computers and workstation manufacturers out of business.
Okay, jackass, implement those "solid technical foundations" on an 8Mhz 68000 with 128K of RAM and a 400K floppy, yet still have room left for applications.
I don't have to, someone else already did: a preemptively multitasking 32 bit operating system with a microkernel architecture, programmable file system devices, and built-in scripting, running on pretty much the same hardware as the Mac (at a considerably lower price, too).
In the early 80's? On what GUI, pray tell? Apple was innovative for having a GUI, and being the first personal computer with one.
The first commercial personal computer with a GUI probably was the Xerox 8010, released in 1981. The Smalltalk GUI came into existence in the late 1970's, and the Lisp machine window system in the early 1980's, both powerful, well-architected, object-oriented systems. X came out in 1984 and rapidly evolved into X10 in 1985. The Blit was published in 1985, although it also had been in development for several years.
Everybody was developing GUIs in the early 1980's; Apple was just following an industry trend, and they cut enough corners to get to market a little earlier than the others.
NOTHING used object orientation yet outside of research; certainly no OS development - it was way too slow for the processors of the era.
Wrong. OOP doesn't require powerful processors, and there were several efficient object-oriented languages around already. Simula-67 came out in 1967 (!).
A bare 68000 did not have the hardware to support a modern operating system.
First of all, Apple had complete freedom in choosing their hardware; if you want to argue that the 68000 was a bad choice, Apple only has themselves to blame for it. They could have used one of the many alternative processors, or they could have added a trivial MMU outside the processor, like everybody else did.
But, more importantly, Apple wasn't building a multiuser system with a hard disk, they were building a single-user system with a floppy. Such a system doesn't need an MMU or VM or UNIX compatibility. But even without those capabilities, you don't have to throw any considerations of architecture out the window. A plain 68000 was a powerful machine at the time, and you can put a well-architected multi-tasking operating system on it.
Amiga, for example, demonstrated a high-quality, 32-bit preemptively multitasking operating system with a microkernel architecture on the plain 68000. Unlike MacOS, AmigaOS was a joy to program. Having the two machines side-by-side, two things were crystal clear right away: (1) AmigaOS was lightyears ahead of MacOS technologically, and (2) Apple was going to kill Amiga because Apple had better marketing. You don't have to take my word for it: you can still run the original Amiga and Macintosh as emulators today and see for yourself.
Whichever way you look at it, Apple cut lots of corners to beat others to market, and we all had to suffer the consequences.
AFAIK, Apple doesn't have much manufacturing anymore, they are outsourcing manufacturing to contract manufacturers in Taiwan and other places, the same companies that also make desktops and laptops for PC manufacturers.
Does anybody know whether there are still any Apple manufacturing facilities left at all? Do they still actually produce any hardware themselves?
The Mac team did their best with what was available at a reasonable cost. I'm not going to blame them for decisions that were suboptimal on processors that would not exist for many years.
You are saying exactly the same thing as I'm saying: Apple cut corners in order to get a cheap product out to market quickly. As a result, they created de-facto standards that the world had to live with for decades to come. You seem to think that it's a good thing that Apple managed to beat everybody to market by a year or two with this strategy, I don't. I think the world would have been better off to wait another year or two for a GUI-based machine with an operating system that was properly architected.
The original Mac ran on a 68000. A slow 16/32-bit processor with no MMU or support for VM. It also had limited memory.
Both Smalltalk and UNIX ran fine on even slower, less powerful machines at the time. You could get BSD UNIX on 16bit machines (and I was using that for several years). Both UNIX and Smalltalk were designed so that they scaled up without a hitch as machines became more powerful. AmigaOS ran on similar hardware and was a well-designed, efficient, multi-tasking operating system (although its design was more hardware dependent).
So, even within the limitations of the hardware they had, the Macintosh team could have done a better job at creating an architecture that would last, instead of a collection of pretty-looking but haphazardly thrown together libraries, which is what they actually released.
And Henry Ford was a jerk for not putting gas turbine engines in the Model T.
I don't know a lot about the history of automotive technology, so I can't judge that. I do know a lot about the history of computing, and in this case experienced it first hand (we had some of the first Lisas and Macintoshes), and that's why I call a spade a spade when it comes to Apple.
Re:So much easier to knock down than to build up
on
Top 10 Apple Flops
·
· Score: 1
Show me a single computer company (hell, any company) that's been around for 25 years or so and hasn't made any mistakes. To succeed, you have to play the game, and playing is a risk.
But the thing to remember is that some technologies were right on even 25 years ago: foremost UNIX and Smalltalk. Apple and Microsoft, effectively prevented them from catching on by dumping cheap visual imitations into the market prematurely without giving a hoot about the underlying technologies that made those kinds of environments easy to create and program.
At least Apple has admitted their mistake, at least in their actions if not in their marketing, by throwing out the mess they created and moving to a UNIX workstation architecture with a stripped down Smalltalk-like system on top of it (Objective C) a few years ago.
I just think it's a bit sad to concentrate on someone's failures. It seems such an... inadequate... response to someone/thing doing well.
Apple isn't "someone", it's a money-making enterprise. And it's good to mull over their failures and our part in them as customers (because Apple only exists because people buy their machines), and their effect on the industry.
However, it isn't worth dwelling too much on them: in the 1980's, Apple was enormously influential in every aspect of personal computing. Today, they have a few percent market share, and while people love to copy the appearance of Apple software and hardware (Apple has good designers), no mainstream desktop platform is making an effort to copy their software architecture anymore (the closest you get is GNUStep). In terms of languages, the industry is moving to Java, C#, VB.NET, Python, Ruby, and in terms of frameworks and platforms, it's.NET, Gnome, Swing, and KDE outside Apple.
Although it built the modern Apple, Apple's biggest failure was MacOS 1-9. Why, do you say? Because the company ran it into the ground. Apple had to throw it out and buy a new operating system (NeXTStep) from another company a few years back.
And the reasons are pretty clear: MacOS didn't start out with a solid technical foundations: no multi-tasking, no memory protection, no object-oriented architecture, just lots and lots of essentially procedural code, largely written in assembly language.
What people rant and rave about in OS X today was already available in the early 1980's, in more complete and better form: dynamic object-oriented programming, GUI designers, UNIX workstation technology, etc. But Apple chose not to copy that, they did a half-assed job at implementing stuff, creating something that looked nice but didn't have a solid foundation. Of course, they did that so that they could bring a low-priced unit to market a couple of years earlier than everybody else. As a result, Apple grabbed lots of market share early on, got lots of praise for how "innovative" they were, and we were stuck with poor software archtitectures for decades to come (Microsoft copied Apple's mistakes).
Unlike MacOS, the software that inspired the Macintosh in the 1970's and 1980's, like Smalltalk, has stood the test of time. It is still at the cutting edge of what is possible in software, GUIs, and still sets the standard for easy, powerful, and flexible software development. And even with all the stuff Apple is shipping now (Objective C, Interface Builder, etc.), they still aren't getting close to that.
Name one commercial application written in Java or pure managed.NET
Oh, come on, you're joking, right? Java is being used for a huge number of server-side applications by companies, and.NET is increasingly being used for the same purpose. There are commercial desktop apps, like IDEs, for example, plus lots of custom apps (what NeXT used to be used for).
The reasons for this are performance issues and the enormous footprint of JVM (or any other "sandbox" type of architecture).
Java is pretty much as fast as C at this point, and C# is catching up quickly. In terms of footprint, neither C++ nor ObjC apps are any better if you account properly for all the memory they actually use.
It is possible though to attack issues of security and reliability by improving existing architecture. Perhaps, by introducing new features into the compiler (such as the Boehm-Demers-Weiser garbage collection), or making the underlying OS less prone to buffer overflows.
Look, we tried this for 20 years, it doesn't work. You can fix Pascal, but anything with C at its core is unfixable, no matter what you do to it. That's why people finally have moved to Java and C#.
You say yes, and then you argue something about X11 and C/C++. Is the answer yes, or no? Would GNUStep have failed for the same reasons as NextStep, or would it have failed because of Objective C and Postscript?
A GUI that relies on Objective-C and/or Postscript for its development is obscure and academic (NeWS as well). That's what killed NeXT and that's what would have killed GNUStep on Linux even if it had made it out of the gate.
I don't think anyone's programmed X11 directly in C since the mid-eighties. In almost all cases, a toolkit is used.
But those toolkits are written in C and C++, not in Objective-C and they don't require DPS.
Unless you actually show a way, I can't help but feel that your "dismissal" is simply an attempt to avoid admitting you're wrong by pretending there's some criteria here that hasn't been mentioned.
It's just not worth debating you on all those points in detail: even if you were right on every single one of them, even if GNUStep were the best GUI in the world, it wouldn't have been a viable choice as an alternative to Gnome and KDE: it was based on technologies that just weren't going to make it and it wasn't even anywhere near ready when it mattered.
And today, something like Mono/Gnome/X11 is just a technically better platform than Objective-C/GNUStep/DPS in my opinion.
In that parallel universe, we'd not have twm, not have Motif/CDE, and not have people's pet desktop environment projects?
I have no idea what you are getting at. twm, Motif/CDE, Gnome, and KDE are all choices for different people, and that's a good thing. That kind of flexibility and modularity is why X11 is still around, while both Microsoft and Apple had to start over, basically from scratch.
You mean GNUStep would have made the GNU/Linux desktop absurdly expensive and virtually unknown outside of academia?
Roughly, yes. X11, C, and C++ were the standard for UNIX workstations and toolkits. Anything even remotely involving Objective-C or Postscript was DOA.
OpenStep was network transparent and GNUStep certainly is [...]
I stand by my statements. You may think that NeXTStep, OpenStep, and/or GNUStep fulfill these and other criteria, I don't.
Both KDE and GNOME upon principles designed for an operating system that doesn't resemble Unix in any real way.
From the point of view of a long-time UNIX user, KDE and Gnome do feel clunky in some areas. But they are decent, usable modern desktops for "the rest of them". They are also highly configurable. And I don't think NeXTStep would have been much better.
What we ended up with is GNU/Linux becoming a kind of Frankenstein OS.
No, what we have ended up with is a wide range of choices for desktop environments, from classic (tvm) to commercial UNIX workstation (Motif/CDE) to mainstream (Gnome, KDE). And, unlike what Apple and Microsoft had to do, Linux+X11 will support the next generation of desktop paradigm, whatever it will be, as well, without having to throw away everything. And that's altogether a really good thing.
It is this mentality, I believe, that prevented the interesting stuff like WindowMaker and GnuStep from gaining any traction at a time when it would have mattered.
What time would that have been? WindowMaker and GNUStep would have done for the Linux desktop what NeXT did a decade earlier: they would have made the Linux desktop fail, and pretty much for the same reasons. And what "interesting stuff" do you think they offered?
Even if GNUStep would have been technically better, it simply wasn't anywhere near usable when Gnome and KDE started getting traction (I still have the old GNUStep CDs that I got around that time).
It is the quality of this standard and the applications' ability to adhere to it that defines the quality of the system. It has taken the X11 community a painfully long time to figure this out.
X11 desktops have had interface standards for nearly two decades, plus a toolkit and desktop environment that implemented them (Motif and CDE).
I think many people would disagree with you, but I'd love to hear some examples of this cutting edge technology and functionality.
Compared to OS X, right of the top of my head, Gnome has XML-based GUI specifications, a network transparent window system, theming, language neutrality (so you can write GUIs in modern OOLs like Python and C#--possible but a lot harder on OS X), and a consistent look-and-feel (as opposed to the Carbon/Cocoa Metal/Glass mess on Macintosh). There is no technology in OS X/Cocoa that I can think of that doesn't have a comparable or better equivalent in Gnome, but if you have examples, please share them; if they are valid, we can add the functionality to Gnome.
He has really made a differnce by thinking different. Jobs is the Che Guevarra of the computer industry except Steve didn't get killed. If you don't him, don't click the links.
The thing that is annoying about the Jobs cult is that he is getting credit for a lot of things neither he nor his companies actually invented. The guy has flair and a knack for packaging and selling technology, but the actual technology and vision comes from elsewhere.
And, flashy as his products are, there are usually better choices. For example, NeXT was targeting rapid, custom software development (banking apps, travel agencies, etc.), but there were plenty of tools like that for UNIX and Windows workstations around that didn't require you to throw away everything and start on a new platform. The situation is not unlike today with Mac vs. everything else.
It's honestly amazing. I'm serious. Can anyone remember Windows 3.11? That's what was state of the art when this came out.
No, it was not the state of the art. The state of the art was defined by systems like the Xerox office workstations, Smalltalk, and the work done at several other companies and research labs.
NeXT, like Windows and Macintosh, was an attempt to bring some of those ideas to market. NeXT incorporated more of the technology, but they also weren't as successful, which shows you again that packing more technology into a product often is not a good idea if you want to win in the market.
Longhorn sounds like they're ripping off a ton of OS X technology, like a new display technology, hardware-accelerated window drawing, and so on.
In order to "rip off" OS X technology, that technology would have to belong to OS X. But the people who created OS X did not invent the display model they are using, they did not invent hardware accelerated drawing, etc. Apple and NeXT, like Microsoft, took existing ideas from lots of other products and research labs and put them in their product.
Adobe, Macromedia, id Software, and so on aren't going to rewrite their apps in unmanaged C++.NET code just to fit in.
No, but that doesn't matter: if you need to develop a new application, you can do it with a managed API, which will save you time and effort. If Adobe finds it simpler to struggle with creating and debugging unmanaged applications, that's their thing.
Note, incidentally, that Objective-C is "unmanaged" in the C#/CLR sense: you have to add code manually to do storage management, and you have to worry about the possibility bad pointers in Objective-C code.
And, as hardware improved, there could have been an advanced desktop built on top of Linux that would have been a very compelling alternative to Win9x, if not the leading edge of desktop innovation. Instead, we got a start menu, a task bar, and a dump truck full of skins.
A start menu and a task bar is pretty much what OS X uses (Apple menu, dock), together with a bunch of quick-launch buttons. Despite all the hoopla, the OS X GUI is not all that different from any other GUI: separate apps, file storage of documents, file system browsers, icons, desktop, etc.
The real issue is what the underlying technology is. Objective-C is a better language for building GUIs than plain C or C++. NeXT made the right choice in language for when the OS was developed.
Today, however, systems like Gnome are often programmed in Python or C#, and those are even nicer and more modern object-oriented languages than Objective-C. Furthermore, the idea of having a separate persistent and manipulable representation of GUI layouts has caught on and in Gnome, you can use XML-based representations to do that.
Software has evolved and become more standardized. Desktops like Gnome are on the cutting edge of what is done in the real world, ahead both functionally and technologically of both Windows and OS X.
The designs, ideas, and concepts were all there in the 90s waiting to implemented.
Not only were many of the designs, ideas, and concepts around before the 90s, they were already implemented, in systems like Smalltalk; they didn't originate with NeXTStep, although NeXT did a good job packaging them in a workstation system (albeit, commercially with comparatively little success).
I'm not sure what's supposed to be so impressive about the GNUstep GUI builder. Compare it with Eclipse or any other modern IDE. Not only do they give you full visual design of your GUI app, they also provide tons of other functionality, including refactoring, version control, etc. And because they compile into languages with garbage collection, runtime safety, and other features, big projects are a lot easier to create and debug as well than in GNUstep.
And, for historical interest, you might want to compare it with some of the Smalltalk IDEs from the 1980's and 1990's, where a lot of the functionality and ideas of modern IDEs were originally developed.
NextStep in 1989 was an endless series of brilliant concepts and ideas that are just now coming into mainstream operating systems.
Yes, they were brilliant concepts, but they weren't invented by NeXT. Objective-C, for example, came from Stepstone and was an attempt to bring at least some Smalltalk concepts to a C dominated world. In the transition from Smalltalk to Objective-C, a lot of important functionality actually got lost.
The organizations who developed those brilliant concepts were places like Xerox, AT&T, IBM, MIT, CMU, Stanford, SRI, etc. NeXT was a well-engineered attempt to bring those technologies into the mainstream, mostly by scaling them down and leaving out a lot of stuff. That's something Jobs is good at: picking good technologies and putting them into his company's products.
As someone else mentioned, the foundations of OS X are a lot more mature than people realize. Cocoa is truly a fantastic way to develop apps.
The NeXTStep foundations of OS X are a reasonable compromise for the late 1980's, when you have to use stripped down OO languages like Objective-C because nothing else will fit well and because C compatibility is a must. But hardware these days ships with hundreds of megabytes of memory and GHz processors. We don't have to leave out important features like safety, garbage collection, full reflection, and dynamic code generation. Furthermore, other window systems offer the same graphics primitives now as OS X (translucency, anti-aliasing, etc.).
In different words, among commercially available systems, NeXTStep was a great system compared to SunOS. It's not such a great system anymore compared to Java, Mono, or.NET.
This video was made in 1992!!! It is amazing how advanced NeXT was at that time.
Not really. Most of the technology in NeXT was invented elsewhere: Smalltalk, UNIX, Simula, Lisp, etc., at places like AT&T, Stepstone, Xerox, Adobe, Sun, and IBM. NeXT's object system, programming language, and GUI were generally more limited versions of the systems it copied so liberally from.
NeXT was primarily an attempt to bring brilliant new technologies invented elsewhere to a UNIX workstation environment. It succeeded pretty well at that and was a pretty well-engineered system for the time.
A lot of the technologies that we take for granted in MacOS X were already around at the time,
Windows and Linux (and every other platform on the planet for that matter) have perfect OSs and GUIs
I didn't say any of that. In fact, Windows and Linux have lots of problems. But Macintosh doesn't fix most of them and it creates lots of problems of its own, so there is no point switching to Macintosh. In different words, I have been on the Macintosh side, and the grass is no greener. That's why trying Macintosh was a disappointment.
I suppose or by your logic you should never buy anything until it's perfect.
No, my logic is much simpler: if you are a Linux user, don't bother with Macintosh; Linux is at least as good, comes with tons more software, and will cost you less.
Seriously, sounds like you played around with OS X once and decided that because it wasn't what you were used to it has "fundamental OS and GUI problems."
Actually, I have used it much more than that. But even if I had "just played around with it", Apple's claims for it are such that that should have been enough to convince me how intuitive it is. In reality, it's just another GUI that does things slightly differently from everybody else.
There is so much misinformation here it's astonishing.
Well, you yourself just said:
"The MacOS is NOT a Unix workstation!"
"Fink sucks. Just because MacOS CAN run linux software doesn't mean it SHOULD."
"You seem to want Photoshop for free out of iLife,"
So, the Macintosh is not a UNIX or Linux workstation, it is a pain to get the FOSS to run on it that Linux users are used to, and Macintosh users need to buy expensive commercial apps for a lot of basic functionality. None of the other things you say matter in the context of this question (wrong as many of them are) because you yourself agree on the core issue: the Macintosh is not a substitute for a Linux machine.
I found the Macintosh laptops to be a big disappointment. First of all, there were hardware problems: one died within a few weeks, the replacement didn't play DVDs properly. The processor is pretty slow, too: even in its heyday, a G4 wasn't all it was cracked up to be, and today it is really not competitive anymore. And it's not true that the thing never crashes; it's not bad, but the GUI will hang on occasion, and I have had it crash, too.
Then, your only real choice for an office suite is Microsoft Office. If you want a laptop just to run Microsoft Office, I suppose it is better than a Windows laptop. OpenOffice has too many limitations on the Mac (among other things, forget about using it for presentations) and it requires you to fiddle with X11, which isn't well integrated (and also needs to be installed). iWork isn't a serious academic or business tool either: no spreadsheet, no math, limited drawing.
The iLife applications are useless toys: iPhoto doesn't let you fix even gross problems with images, iMovie has limitations on what you can important and export (looks like they are deliberate). You probably need to upgrade to expensive commercial packages if you want anything that's more than a toy.
I thought there were going to be a bunch of nice outliners and brainstorming tools for Macintosh--lots of them are advertised with great fanfare and colorful ads, but they were pretty much a disappointment, too: proprietary formats, complex UIs, and limited functionality. There are better open source tools available than that.
Fink is supposed to be the way to install more of a real UNIX/Linux environment on Macintosh, but I had no end of trouble. Worse, for many packages, there are two versions of it: the Fink version and a non-Fink version. Some Mac applications assume one, some the other, and if you install both, you run into conflicts. Cygwin on Windows runs more reliable than Fink on Macintosh.
Macintosh network configuration is supposed to be really well done, but it's pretty cumbersome, no better than Windows. Yes, you get a pull down menu of wireless basestations in your surroundings right on the menu bar. But if you use anything other than the default network settings, you still need to dig into the network configuration dialog, which matches Windows in its obscurity. Software like Switcher-X shows that you can do better--why can't Apple at least ship decent network configuration tools with the Mac?
Macintosh also promises to integrate well into Windows and UNIX networks, but that's an unfulfilled promise, too. Yes, it sort of speaks SMB and NFS, but actually getting automounting and name services to work is at least as much of a pain as on other UNIX systems, and there is a lot of non-standard stuff you have to do on Mac.
And then there is the GUI. It's slow. It's non-standard, in look-and-feel as well as in its APIs. It has a theme that you either love or hate, but you can't change it very much (at least not out of the box).
Coming from Linux, where all the UNIX and GUI tools are integrated, self-updating, and everything just works, and where there is a wide choice of toolkits and programming environments, and at least three different free office suites, the Macintosh was a big disappointment and a money pit.
So, if you are asking when you should switch, my answer is: not yet, not until Apple fixes some really fundamental problems with the Macintosh operating system and GUI, until they actually get serious about making it a competitive UNIX workstation (which means, among other things, decent X11 support), until the hardware is up to modern standards, and until there is more decent application software available for it. If you do switch now, I expect you'll come back to Linux again. I did, as did several other people I know.
Read carefully: Sun isn't opening their patent portfolio to all FOSS projects, only to their own pet FOSS projects. It looks like, for example, Linux does not fall under that grant and Sun can claim patent infringement by Linux at any time.
Altogether, this is another underhanded attempt by Sun to drum up support for their failing kernel and OS efforts and represents, if anything, a threat to Linux.
Don't trust Sun: these guys are desparate and hence dangerous. If they release stuff under an approved OSI license, you can use it (eg OOo). Anything else from Sun is a Trojan horse and a ticking time bomb (eg Java), both for FOSS and for commercial customers.
Slashdot Mirror
There is nothing wrong with working help desk in general; it's an important job. But there is something wrong with working help desk in your 30's from your own point of view if you got into the field with visions of making billions instantly with only minimal knowledge of computers. The elitism here is the elitism of the people who chose a career in computers without the right preparation.
Linus is talking about open source software development as a career path, not system administration or help desks or other standard IT functions. Open source development is special and different from proprietary software development. Administration or support are not (there is just a little less of it).
The difficulties with installing new software using a linux-based operating system arise when you want to install software from elsewhere
That may or may not be difficult. Even if you want to download applications, in many distributions, all you do is download an RPM file and double click on it.
But, importantly, you don't have to do that in Linux at all: if you run a decent distro, all software should be installed through the package manager. That's where you search for applications and tell it what you want. That's the easiest way of installing and maintaining software: you don't have to download anything at all, and you never have to worry about it; all you do is tell the computer what you want.
In fact, most people don't even have to do that: when you install one of the major Linux distributions, you pretty much get all the software you would ever want or need preinstalled: office suites, desktops, PIMs, mail readers, web browsers, scientific software, desktop publishing, image editing, etc.
The report is also wrongly optimistic about software installation on Windows or Mac OS X. On Windows, most software comes with installers. Macintosh, on the other hand, is far messier: not only do you have to download the software and extract it, after that's done and made a mess on the desktop, you have to figure out how to proceed: maybe you open a folder or a virtual drive and drag something to your application folder, maybe you run an installer, and if you run an installer, it asks questions about "what disk volume" you want to install the software on. Macintosh software installations are the most confusing of the bunch. And neither Windows nor Macintosh will automatically update software that wasn't shipped by the OS vendor.
If the main complaint about Linux is that software is hard to install, then Linux is at this point perfect for the desktop. And that's, actually, what I find with my "mom and dad": they have trouble dealing with Macintosh and I have ended up having to install software for them, but with Linux, I just put the machine there and it works, every time, all the time.
Um, no. They were making a product for the market existed, and even then, it *barely* existed.
It was a market that was going to exist whether or not Apple grabbed the lead by putting out a hyped up, heavily marketed, pretty-but-inferior product. In parallel with Macintosh development, there were several other companies that were developing inexpensive GUI-based personal computers. Barely a year after the Macintosh, Amiga started selling a machine with pretty much the same specs and price as the Macintosh, but with hardware graphics acceleration, a preemptively multitasking 32bit operating system, and a really well-designed microkernel. Workstation efforts were in full swing, which yielded graphical workstations that were no more expensive and far more powerful than a Macintosh shortly thereafter.
I suppose Apple employees genuinely believed the marketing hype of their own companies that they were the salvation for the "rest of us" and that the world came down to Microsoft and Macintosh. But the world only came down to Microsoft and Macintosh after Apple had driven all the other, innovative personal computers and workstation manufacturers out of business.
Okay, jackass, implement those "solid technical foundations" on an 8Mhz 68000 with 128K of RAM and a 400K floppy, yet still have room left for applications.
I don't have to, someone else already did: a preemptively multitasking 32 bit operating system with a microkernel architecture, programmable file system devices, and built-in scripting, running on pretty much the same hardware as the Mac (at a considerably lower price, too).
In the early 80's? On what GUI, pray tell? Apple was innovative for having a GUI, and being the first personal computer with one.
The first commercial personal computer with a GUI probably was the Xerox 8010, released in 1981. The Smalltalk GUI came into existence in the late 1970's, and the Lisp machine window system in the early 1980's, both powerful, well-architected, object-oriented systems. X came out in 1984 and rapidly evolved into X10 in 1985. The Blit was published in 1985, although it also had been in development for several years.
Everybody was developing GUIs in the early 1980's; Apple was just following an industry trend, and they cut enough corners to get to market a little earlier than the others.
NOTHING used object orientation yet outside of research; certainly no OS development - it was way too slow for the processors of the era.
Wrong. OOP doesn't require powerful processors, and there were several efficient object-oriented languages around already. Simula-67 came out in 1967 (!).
A bare 68000 did not have the hardware to support a modern operating system.
First of all, Apple had complete freedom in choosing their hardware; if you want to argue that the 68000 was a bad choice, Apple only has themselves to blame for it. They could have used one of the many alternative processors, or they could have added a trivial MMU outside the processor, like everybody else did.
But, more importantly, Apple wasn't building a multiuser system with a hard disk, they were building a single-user system with a floppy. Such a system doesn't need an MMU or VM or UNIX compatibility. But even without those capabilities, you don't have to throw any considerations of architecture out the window. A plain 68000 was a powerful machine at the time, and you can put a well-architected multi-tasking operating system on it.
Amiga, for example, demonstrated a high-quality, 32-bit preemptively multitasking operating system with a microkernel architecture on the plain 68000. Unlike MacOS, AmigaOS was a joy to program. Having the two machines side-by-side, two things were crystal clear right away: (1) AmigaOS was lightyears ahead of MacOS technologically, and (2) Apple was going to kill Amiga because Apple had better marketing. You don't have to take my word for it: you can still run the original Amiga and Macintosh as emulators today and see for yourself.
Whichever way you look at it, Apple cut lots of corners to beat others to market, and we all had to suffer the consequences.
AFAIK, Apple doesn't have much manufacturing anymore, they are outsourcing manufacturing to contract manufacturers in Taiwan and other places, the same companies that also make desktops and laptops for PC manufacturers.
Does anybody know whether there are still any Apple manufacturing facilities left at all? Do they still actually produce any hardware themselves?
The Mac team did their best with what was available at a reasonable cost. I'm not going to blame them for decisions that were suboptimal on processors that would not exist for many years.
You are saying exactly the same thing as I'm saying: Apple cut corners in order to get a cheap product out to market quickly. As a result, they created de-facto standards that the world had to live with for decades to come. You seem to think that it's a good thing that Apple managed to beat everybody to market by a year or two with this strategy, I don't. I think the world would have been better off to wait another year or two for a GUI-based machine with an operating system that was properly architected.
The original Mac ran on a 68000. A slow 16/32-bit processor with no MMU or support for VM. It also had limited memory.
Both Smalltalk and UNIX ran fine on even slower, less powerful machines at the time. You could get BSD UNIX on 16bit machines (and I was using that for several years). Both UNIX and Smalltalk were designed so that they scaled up without a hitch as machines became more powerful. AmigaOS ran on similar hardware and was a well-designed, efficient, multi-tasking operating system (although its design was more hardware dependent).
So, even within the limitations of the hardware they had, the Macintosh team could have done a better job at creating an architecture that would last, instead of a collection of pretty-looking but haphazardly thrown together libraries, which is what they actually released.
And Henry Ford was a jerk for not putting gas turbine engines in the Model T.
I don't know a lot about the history of automotive technology, so I can't judge that. I do know a lot about the history of computing, and in this case experienced it first hand (we had some of the first Lisas and Macintoshes), and that's why I call a spade a spade when it comes to Apple.
Show me a single computer company (hell, any company) that's been around for 25 years or so and hasn't made any mistakes. To succeed, you have to play the game, and playing is a risk.
... inadequate... response to someone/thing doing well.
.NET, Gnome, Swing, and KDE outside Apple.
But the thing to remember is that some technologies were right on even 25 years ago: foremost UNIX and Smalltalk. Apple and Microsoft, effectively prevented them from catching on by dumping cheap visual imitations into the market prematurely without giving a hoot about the underlying technologies that made those kinds of environments easy to create and program.
At least Apple has admitted their mistake, at least in their actions if not in their marketing, by throwing out the mess they created and moving to a UNIX workstation architecture with a stripped down Smalltalk-like system on top of it (Objective C) a few years ago.
I just think it's a bit sad to concentrate on someone's failures. It seems such an
Apple isn't "someone", it's a money-making enterprise. And it's good to mull over their failures and our part in them as customers (because Apple only exists because people buy their machines), and their effect on the industry.
However, it isn't worth dwelling too much on them: in the 1980's, Apple was enormously influential in every aspect of personal computing. Today, they have a few percent market share, and while people love to copy the appearance of Apple software and hardware (Apple has good designers), no mainstream desktop platform is making an effort to copy their software architecture anymore (the closest you get is GNUStep). In terms of languages, the industry is moving to Java, C#, VB.NET, Python, Ruby, and in terms of frameworks and platforms, it's
Although it built the modern Apple, Apple's biggest failure was MacOS 1-9. Why, do you say? Because the company ran it into the ground. Apple had to throw it out and buy a new operating system (NeXTStep) from another company a few years back.
And the reasons are pretty clear: MacOS didn't start out with a solid technical foundations: no multi-tasking, no memory protection, no object-oriented architecture, just lots and lots of essentially procedural code, largely written in assembly language.
What people rant and rave about in OS X today was already available in the early 1980's, in more complete and better form: dynamic object-oriented programming, GUI designers, UNIX workstation technology, etc. But Apple chose not to copy that, they did a half-assed job at implementing stuff, creating something that looked nice but didn't have a solid foundation. Of course, they did that so that they could bring a low-priced unit to market a couple of years earlier than everybody else. As a result, Apple grabbed lots of market share early on, got lots of praise for how "innovative" they were, and we were stuck with poor software archtitectures for decades to come (Microsoft copied Apple's mistakes).
Unlike MacOS, the software that inspired the Macintosh in the 1970's and 1980's, like Smalltalk, has stood the test of time. It is still at the cutting edge of what is possible in software, GUIs, and still sets the standard for easy, powerful, and flexible software development. And even with all the stuff Apple is shipping now (Objective C, Interface Builder, etc.), they still aren't getting close to that.
Name one commercial application written in Java or pure managed .NET
.NET is increasingly being used for the same purpose. There are commercial desktop apps, like IDEs, for example, plus lots of custom apps (what NeXT used to be used for).
Oh, come on, you're joking, right? Java is being used for a huge number of server-side applications by companies, and
The reasons for this are performance issues and the enormous footprint of JVM (or any other "sandbox" type of architecture).
Java is pretty much as fast as C at this point, and C# is catching up quickly. In terms of footprint, neither C++ nor ObjC apps are any better if you account properly for all the memory they actually use.
It is possible though to attack issues of security and reliability by improving existing architecture. Perhaps, by introducing new features into the compiler (such as the Boehm-Demers-Weiser garbage collection), or making the underlying OS less prone to buffer overflows.
Look, we tried this for 20 years, it doesn't work. You can fix Pascal, but anything with C at its core is unfixable, no matter what you do to it. That's why people finally have moved to Java and C#.
You say yes, and then you argue something about X11 and C/C++. Is the answer yes, or no? Would GNUStep have failed for the same reasons as NextStep, or would it have failed because of Objective C and Postscript?
A GUI that relies on Objective-C and/or Postscript for its development is obscure and academic (NeWS as well). That's what killed NeXT and that's what would have killed GNUStep on Linux even if it had made it out of the gate.
I don't think anyone's programmed X11 directly in C since the mid-eighties. In almost all cases, a toolkit is used.
But those toolkits are written in C and C++, not in Objective-C and they don't require DPS.
Unless you actually show a way, I can't help but feel that your "dismissal" is simply an attempt to avoid admitting you're wrong by pretending there's some criteria here that hasn't been mentioned.
It's just not worth debating you on all those points in detail: even if you were right on every single one of them, even if GNUStep were the best GUI in the world, it wouldn't have been a viable choice as an alternative to Gnome and KDE: it was based on technologies that just weren't going to make it and it wasn't even anywhere near ready when it mattered.
And today, something like Mono/Gnome/X11 is just a technically better platform than Objective-C/GNUStep/DPS in my opinion.
In that parallel universe, we'd not have twm, not have Motif/CDE, and not have people's pet desktop environment projects?
I have no idea what you are getting at. twm, Motif/CDE, Gnome, and KDE are all choices for different people, and that's a good thing. That kind of flexibility and modularity is why X11 is still around, while both Microsoft and Apple had to start over, basically from scratch.
You mean GNUStep would have made the GNU/Linux desktop absurdly expensive and virtually unknown outside of academia?
Roughly, yes. X11, C, and C++ were the standard for UNIX workstations and toolkits. Anything even remotely involving Objective-C or Postscript was DOA.
OpenStep was network transparent and GNUStep certainly is [...]
I stand by my statements. You may think that NeXTStep, OpenStep, and/or GNUStep fulfill these and other criteria, I don't.
Both KDE and GNOME upon principles designed for an operating system that doesn't resemble Unix in any real way.
From the point of view of a long-time UNIX user, KDE and Gnome do feel clunky in some areas. But they are decent, usable modern desktops for "the rest of them". They are also highly configurable. And I don't think NeXTStep would have been much better.
What we ended up with is GNU/Linux becoming a kind of Frankenstein OS.
No, what we have ended up with is a wide range of choices for desktop environments, from classic (tvm) to commercial UNIX workstation (Motif/CDE) to mainstream (Gnome, KDE). And, unlike what Apple and Microsoft had to do, Linux+X11 will support the next generation of desktop paradigm, whatever it will be, as well, without having to throw away everything. And that's altogether a really good thing.
It is this mentality, I believe, that prevented the interesting stuff like WindowMaker and GnuStep from gaining any traction at a time when it would have mattered.
What time would that have been? WindowMaker and GNUStep would have done for the Linux desktop what NeXT did a decade earlier: they would have made the Linux desktop fail, and pretty much for the same reasons. And what "interesting stuff" do you think they offered?
Even if GNUStep would have been technically better, it simply wasn't anywhere near usable when Gnome and KDE started getting traction (I still have the old GNUStep CDs that I got around that time).
It is the quality of this standard and the applications' ability to adhere to it that defines the quality of the system. It has taken the X11 community a painfully long time to figure this out.
X11 desktops have had interface standards for nearly two decades, plus a toolkit and desktop environment that implemented them (Motif and CDE).
I think many people would disagree with you, but I'd love to hear some examples of this cutting edge technology and functionality.
Compared to OS X, right of the top of my head, Gnome has XML-based GUI specifications, a network transparent window system, theming, language neutrality (so you can write GUIs in modern OOLs like Python and C#--possible but a lot harder on OS X), and a consistent look-and-feel (as opposed to the Carbon/Cocoa Metal/Glass mess on Macintosh). There is no technology in OS X/Cocoa that I can think of that doesn't have a comparable or better equivalent in Gnome, but if you have examples, please share them; if they are valid, we can add the functionality to Gnome.
He has really made a differnce by thinking different. Jobs is the Che Guevarra of the computer industry except Steve didn't get killed. If you don't him, don't click the links.
The thing that is annoying about the Jobs cult is that he is getting credit for a lot of things neither he nor his companies actually invented. The guy has flair and a knack for packaging and selling technology, but the actual technology and vision comes from elsewhere.
And, flashy as his products are, there are usually better choices. For example, NeXT was targeting rapid, custom software development (banking apps, travel agencies, etc.), but there were plenty of tools like that for UNIX and Windows workstations around that didn't require you to throw away everything and start on a new platform. The situation is not unlike today with Mac vs. everything else.
It's honestly amazing. I'm serious. Can anyone remember Windows 3.11? That's what was state of the art when this came out.
No, it was not the state of the art. The state of the art was defined by systems like the Xerox office workstations, Smalltalk, and the work done at several other companies and research labs.
NeXT, like Windows and Macintosh, was an attempt to bring some of those ideas to market. NeXT incorporated more of the technology, but they also weren't as successful, which shows you again that packing more technology into a product often is not a good idea if you want to win in the market.
Longhorn sounds like they're ripping off a ton of OS X technology, like a new display technology, hardware-accelerated window drawing, and so on.
.NET code just to fit in.
In order to "rip off" OS X technology, that technology would have to belong to OS X. But the people who created OS X did not invent the display model they are using, they did not invent hardware accelerated drawing, etc. Apple and NeXT, like Microsoft, took existing ideas from lots of other products and research labs and put them in their product.
Adobe, Macromedia, id Software, and so on aren't going to rewrite their apps in unmanaged C++
No, but that doesn't matter: if you need to develop a new application, you can do it with a managed API, which will save you time and effort. If Adobe finds it simpler to struggle with creating and debugging unmanaged applications, that's their thing.
Note, incidentally, that Objective-C is "unmanaged" in the C#/CLR sense: you have to add code manually to do storage management, and you have to worry about the possibility bad pointers in Objective-C code.
And, as hardware improved, there could have been an advanced desktop built on top of Linux that would have been a very compelling alternative to Win9x, if not the leading edge of desktop innovation. Instead, we got a start menu, a task bar, and a dump truck full of skins.
A start menu and a task bar is pretty much what OS X uses (Apple menu, dock), together with a bunch of quick-launch buttons. Despite all the hoopla, the OS X GUI is not all that different from any other GUI: separate apps, file storage of documents, file system browsers, icons, desktop, etc.
The real issue is what the underlying technology is. Objective-C is a better language for building GUIs than plain C or C++. NeXT made the right choice in language for when the OS was developed.
Today, however, systems like Gnome are often programmed in Python or C#, and those are even nicer and more modern object-oriented languages than Objective-C. Furthermore, the idea of having a separate persistent and manipulable representation of GUI layouts has caught on and in Gnome, you can use XML-based representations to do that.
Software has evolved and become more standardized. Desktops like Gnome are on the cutting edge of what is done in the real world, ahead both functionally and technologically of both Windows and OS X.
The designs, ideas, and concepts were all there in the 90s waiting to implemented.
Not only were many of the designs, ideas, and concepts around before the 90s, they were already implemented, in systems like Smalltalk; they didn't originate with NeXTStep, although NeXT did a good job packaging them in a workstation system (albeit, commercially with comparatively little success).
I'm not sure what's supposed to be so impressive about the GNUstep GUI builder. Compare it with Eclipse or any other modern IDE. Not only do they give you full visual design of your GUI app, they also provide tons of other functionality, including refactoring, version control, etc. And because they compile into languages with garbage collection, runtime safety, and other features, big projects are a lot easier to create and debug as well than in GNUstep.
And, for historical interest, you might want to compare it with some of the Smalltalk IDEs from the 1980's and 1990's, where a lot of the functionality and ideas of modern IDEs were originally developed.
NextStep in 1989 was an endless series of brilliant concepts and ideas that are just now coming into mainstream operating systems.
.NET.
Yes, they were brilliant concepts, but they weren't invented by NeXT. Objective-C, for example, came from Stepstone and was an attempt to bring at least some Smalltalk concepts to a C dominated world. In the transition from Smalltalk to Objective-C, a lot of important functionality actually got lost.
The organizations who developed those brilliant concepts were places like Xerox, AT&T, IBM, MIT, CMU, Stanford, SRI, etc. NeXT was a well-engineered attempt to bring those technologies into the mainstream, mostly by scaling them down and leaving out a lot of stuff. That's something Jobs is good at: picking good technologies and putting them into his company's products.
As someone else mentioned, the foundations of OS X are a lot more mature than people realize. Cocoa is truly a fantastic way to develop apps.
The NeXTStep foundations of OS X are a reasonable compromise for the late 1980's, when you have to use stripped down OO languages like Objective-C because nothing else will fit well and because C compatibility is a must. But hardware these days ships with hundreds of megabytes of memory and GHz processors. We don't have to leave out important features like safety, garbage collection, full reflection, and dynamic code generation. Furthermore, other window systems offer the same graphics primitives now as OS X (translucency, anti-aliasing, etc.).
In different words, among commercially available systems, NeXTStep was a great system compared to SunOS. It's not such a great system anymore compared to Java, Mono, or
This video was made in 1992!!! It is amazing how advanced NeXT was at that time.
Not really. Most of the technology in NeXT was invented elsewhere: Smalltalk, UNIX, Simula, Lisp, etc., at places like AT&T, Stepstone, Xerox, Adobe, Sun, and IBM. NeXT's object system, programming language, and GUI were generally more limited versions of the systems it copied so liberally from.
NeXT was primarily an attempt to bring brilliant new technologies invented elsewhere to a UNIX workstation environment. It succeeded pretty well at that and was a pretty well-engineered system for the time.
A lot of the technologies that we take for granted in MacOS X were already around at the time,
And they were around before NeXT as well.
Windows and Linux (and every other platform on the planet for that matter) have perfect OSs and GUIs
I didn't say any of that. In fact, Windows and Linux have lots of problems. But Macintosh doesn't fix most of them and it creates lots of problems of its own, so there is no point switching to Macintosh. In different words, I have been on the Macintosh side, and the grass is no greener. That's why trying Macintosh was a disappointment.
I suppose or by your logic you should never buy anything until it's perfect.
No, my logic is much simpler: if you are a Linux user, don't bother with Macintosh; Linux is at least as good, comes with tons more software, and will cost you less.
Seriously, sounds like you played around with OS X once and decided that because it wasn't what you were used to it has "fundamental OS and GUI problems."
Actually, I have used it much more than that. But even if I had "just played around with it", Apple's claims for it are such that that should have been enough to convince me how intuitive it is. In reality, it's just another GUI that does things slightly differently from everybody else.
Well, you yourself just said:
So, the Macintosh is not a UNIX or Linux workstation, it is a pain to get the FOSS to run on it that Linux users are used to, and Macintosh users need to buy expensive commercial apps for a lot of basic functionality. None of the other things you say matter in the context of this question (wrong as many of them are) because you yourself agree on the core issue: the Macintosh is not a substitute for a Linux machine.
I found the Macintosh laptops to be a big disappointment. First of all, there were hardware problems: one died within a few weeks, the replacement didn't play DVDs properly. The processor is pretty slow, too: even in its heyday, a G4 wasn't all it was cracked up to be, and today it is really not competitive anymore. And it's not true that the thing never crashes; it's not bad, but the GUI will hang on occasion, and I have had it crash, too.
Then, your only real choice for an office suite is Microsoft Office. If you want a laptop just to run Microsoft Office, I suppose it is better than a Windows laptop. OpenOffice has too many limitations on the Mac (among other things, forget about using it for presentations) and it requires you to fiddle with X11, which isn't well integrated (and also needs to be installed). iWork isn't a serious academic or business tool either: no spreadsheet, no math, limited drawing.
The iLife applications are useless toys: iPhoto doesn't let you fix even gross problems with images, iMovie has limitations on what you can important and export (looks like they are deliberate). You probably need to upgrade to expensive commercial packages if you want anything that's more than a toy.
I thought there were going to be a bunch of nice outliners and brainstorming tools for Macintosh--lots of them are advertised with great fanfare and colorful ads, but they were pretty much a disappointment, too: proprietary formats, complex UIs, and limited functionality. There are better open source tools available than that.
Fink is supposed to be the way to install more of a real UNIX/Linux environment on Macintosh, but I had no end of trouble. Worse, for many packages, there are two versions of it: the Fink version and a non-Fink version. Some Mac applications assume one, some the other, and if you install both, you run into conflicts. Cygwin on Windows runs more reliable than Fink on Macintosh.
Macintosh network configuration is supposed to be really well done, but it's pretty cumbersome, no better than Windows. Yes, you get a pull down menu of wireless basestations in your surroundings right on the menu bar. But if you use anything other than the default network settings, you still need to dig into the network configuration dialog, which matches Windows in its obscurity. Software like Switcher-X shows that you can do better--why can't Apple at least ship decent network configuration tools with the Mac?
Macintosh also promises to integrate well into Windows and UNIX networks, but that's an unfulfilled promise, too. Yes, it sort of speaks SMB and NFS, but actually getting automounting and name services to work is at least as much of a pain as on other UNIX systems, and there is a lot of non-standard stuff you have to do on Mac.
And then there is the GUI. It's slow. It's non-standard, in look-and-feel as well as in its APIs. It has a theme that you either love or hate, but you can't change it very much (at least not out of the box).
Coming from Linux, where all the UNIX and GUI tools are integrated, self-updating, and everything just works, and where there is a wide choice of toolkits and programming environments, and at least three different free office suites, the Macintosh was a big disappointment and a money pit.
So, if you are asking when you should switch, my answer is: not yet, not until Apple fixes some really fundamental problems with the Macintosh operating system and GUI, until they actually get serious about making it a competitive UNIX workstation (which means, among other things, decent X11 support), until the hardware is up to modern standards, and until there is more decent application software available for it. If you do switch now, I expect you'll come back to Linux again. I did, as did several other people I know.
Read carefully: Sun isn't opening their patent portfolio to all FOSS projects, only to their own pet FOSS projects. It looks like, for example, Linux does not fall under that grant and Sun can claim patent infringement by Linux at any time.
Altogether, this is another underhanded attempt by Sun to drum up support for their failing kernel and OS efforts and represents, if anything, a threat to Linux.
Don't trust Sun: these guys are desparate and hence dangerous. If they release stuff under an approved OSI license, you can use it (eg OOo). Anything else from Sun is a Trojan horse and a ticking time bomb (eg Java), both for FOSS and for commercial customers.