Slashdot Mirror
Will Linux have the same fate as Java?
geophile writes "This Boston Globe article starts out talking about Java's failure to take over the world, and then questions whether Linux will suffer the same fate. " Interesting question, and perhaps I'm being partial, but I feel as though Java promised that it was going to change the world-right then and there. Linux has been building for quite sometime, and continues to develop.
Sure, java is a language, but its not the language that had everyone hot and bothered; it was the VM. What has collapsed is not the language, but the business case for java everywhere, which rested on the VM.
What was the business case?
1) Idependence from Microsoft.
2) Access to non-microsoft desktops.
3) Low deployment costs to pure thin clients.
4) Powerful language has most important semantic capabilities of C++ without the most troublesome features.
What are the realities?
Idependence from Microsoft: Most PHBs aren't going to see dependence on Microsoft as a problem, because it occurs in a intellectual realm orthagonal to their daily experience. Also, if everyone is in the same boat, it's not a _competitive_ issue. Finally, the key MS monopoly is not in OS, but in office applications. Without destroying this, essentially everyone is going to have to have a copy of Windows around (leaving aside people who can happily live on the fringe and don't go blank when they hear words like "XML").
2) Access to non-microsoft desktops. The two significant segments are Mac and Unix. The Java VM isn't going to deliver the Mac users the native user experience they prize, so scratch the Mac.
3) Pure thin clients as a solution to deployment costs. Most internal MIS shops will find the retraining costs orders of magnitude higher than deployment costs. The download bandwidth requirements for non-trivial applications make java impractical in the applications where pure thin clients are the most desirable.
4) Powerful language has most important semantic capabilities of C++ without the most troublesome features. IMO, the strongest argument for java, and alas the least persuasive to any non-technical businessman.
Now, what is the business case for Linux?
1) It is free (beer).
2) It makes surplus hardware useful.
3) Service based business model.
4) Stable and faster on common hardware.
5) Open/free source.
What are the realities?
1) It is free (beer).
The usual response to an acquisition cost differential is to appeal to TCO. The biggest cost is going to be retraining existing staffs (who are often barely competent to begin with). On the other hand, the scalability problems, and increasing complexity of MS offerings may somewhat negate this. For many shops, TCO is going to be lower sticking with the status quo, but if they have a few people who know Unix and have a role in which Linux can play, it's going to be hard to say no except out of pure intransigence. Overall, this argument is a mixed bag.
2) It makes surplus hardware useful.
If you need a intranet discussion and web server, can buy a new NT box with the appropriate license packs etc., or you can press that 486 in the basement into service. This is pretty much a no-brainer, because most of the cost is going to be content related; you can always give MS your bucks later.
3) Service based business model. This is the strongest case to put to management. Microsoft's revenues are primarily from controlling the copying of their software, not service. Third parties have stepped into the breach, but only have the vaguest idea of how things really work internally, and so are limited in effectiveness. Everybody has been burned by this. On the other hand Linux and BSD are freely copyable by all, so that the only feasible source of revenue is to provide service. This is practical, because every internal detail of how things work is public knowledge. This has already produced a situation where the _free_ support of Linux is far superior than almost any commercial support you can buy for Windows, so the bar is high indeed for companies like Red Hat who intend to make money in the Linux service arena.
4) Stable and faster on common hardware.
Strong case for server software, weak for desktop use. The most important factor in desktop use is to provide an adequate suite of end-user tools. Again a place where the MS application suite monopoly is key. Users hate BSOD, but they would hate even more not getting fidelity in printed output or scaling fonts, or not being able to cut and paste between two apps they use. Koffice, where are you?
5) Open/free source.
Again, truly a strong case, but with little or no appeal to management.
Overall, the case for Linux is much stronger than the case for java. While it holds great promise, the most serious problem for Java is the issue of ownership. If java fails to meet the business goals of Sun, or at some future point undermines them, you can be sure that java will be no more. Linux boxes do many useful things well today, and will continue to do them forever no matter the fate or strategy of any one company, so long as even one person has the desire to keep it going.
This last point is critical for developers. Apple developers have been burned by about faces on Lisa Pascal, Open Transport, Open Doc, and Applescript. MS has been incomparably better, but far from perfect. They've done things like restrict the screen resolution of WinCE, which naturally is a result of the difference between their users' interests and their own.
In the free software model, the user's insterest always wins.
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.
Another point that people are missing is this: if linux fails miserably in the corporate market, it doesnt matter (The Rock style inflection). People will still use linux. At least I will. And I'm willing to bet others will as well. The model of opensource is inherently longterm. The only thing to really stop development is if everyone in the world looses intrest in programming.
"Fighting the underpants gnomes since 1998!" "Bruce Schneier knows the state of schroedinger's cat"
Caterpillar. Cargill. Tennessee Valley Authority. American Airlines. Sabre. Tricon Restaurants (Taco Bell, Pizza Hut, KFC). These are not "mom 'n' pop" operations, they are huge corporations collectively dumping hundreds of millions into the Java industry. There are many, many more.
Good point. I'll look for one or maybe take a stab at one.
IBM did no such thing. Not only does IBM still sell and support OS/2, but it's sales are exceeding projections. OS/2 may serve as an example of a market failure, but it also serves as an example of how IBM does not abandon its customers.
What you didn't mention is the companies aligned against Java - Microsoft and HP to name two. Both are pushing HP's Chai.
This statement is really proof that you are way too ignorant to be criticizing Java. Chai can hardly be called anti-Java. Chai is a Java Virtual Machine. Chai supports Java, albeit not necessarily Sun's strict definition. And I did mention HP -- it's another company investing in Java technology (you did read my post, right?).
Beta is dead, and I'm sure Sony can't be too pleased to see DAT relegated to a tiny market of recording enthusiasts.
ROTFLMAO. Anyone who could dismiss ABC, NBC, CBS, Fox, HBO, BBC, CBC, Univision and who knows how many global networks both broadcast and cable as a "tiny market of recording enthusiasts" is irretreivably stupid. I challenge you to find a broadcast production facility (among thousands) in the civilized world that does not feature a Beta SP deck and a DAT.
Java exists to attack Microsoft. Everybody knows this. Java is a top-down marketing strategy to hit 'em where it hurts. In a way, Java is just another closed product that claims to fit needs that Microsoft has missed. Java is essentially fighting fire with fire.
People think that Linux is just Anything But Microsoft, but it's not really. It's an organic, bottom-up movement. Linux grows into what people need and want, in a direction that provides a stable, useful tool.
This is exactly the opposite of the direction that Microsoft products take. Microsoft products grow in a direction that increases your hunger for more Microsoft products and new releases of the old ones.
Linux and GPL in general being so completely different than the MS-wares makes people believe that it's a directed attack on MS, but it's not. Linux users generally just have work to get done and Linux does it for them. I won't deny that some Linux users ALSO enjoy the Anything But Microsoft aspect of Linux, but this isn't the raison d'etre for Linux as it is for Java. Linux is like fighting fire with water.
Linux being completely different than Microsoft products, gives Microsoft little leverage to try and defeat it. The Holloween Memos show that they don't know what to do, and the "Labor Day Memo" pokes fun at the possibility of their using their standard tactics.
It should go without saying that they never support this product, fail to mention it on most of their web pages, remove all references to it from their Knowledge Base, charge exorbitant prices for it (while also distributing it free in computer books) and include the 49.7 day bug.
The obvious problem with L++ is that there would be no community support for it. If MS didn't support it heavily (as you suggest), and the Linux community wouldn't touch it (as I would guess) then it would be stillborn. Anyone who tried to make it work would give up on it almost immediately when they found they couldn't get any support for their problems.
Then, after dividing the market, there's no way Linux could sue them.
Dividing the market just isn't a problem with Linux the way it would be with Java. Java only works at all if it works exactly the same everywhere. Something called Linux works if it works anywhere. People who were interested in getting work done with a stable platform that has rich functionality (love those MS buzzwords!) in Internet applications just use some version of Linux. It doesn't hurt Linux that there are branches. If somebody sees things from two different branches that could productively brought together into one, this can be done fairly easily.
Once you've used Linux somewhere for an application it's just dirt cheap to clone that success N times. The incremental costs are often near 0 as you can use old computers that have long since been amortized down to nothing and just happen to be laying around. As MS knows from their experience of marginalizing Netscape, you just can't compete with free.
Give me a break.
1. Java is alive and well. Script kiddies won't know it, but those in the industry certainly do. Java is rapidly expanding into corporate IS/IT, server-side startup software ventures, and embedded systems. Java talent earns somewhere around 50% more on average than equivalent C/C++ talent.
2. Java doesn't suck. You only think it sucks because either you haven't really tried implementing a significant project in it (since 1.1.x), or you're a script kiddie who can't handle complying with imposed methodology. The class libs in Java are very close to object-based systems like Smalltalk. Very nice, but very strict. It's not PERL, VB, or any other script interpreted thing. It's more on a par with C++, Smalltalk, CLOS, etc.
Crusading script kiddie zealots would be well advized to embrace Java, Delphi, Oracle, Informix, ODI, or anything else enterprise level which legitimizes Linux in the enterprise. I know you hate it, but the enterprise is important to your succeeding beyond "hobbyist" status. (And save your ISP args; witness FreeBSD q.e.d)
--
Am I the only one who sees this as basically a cheap attempt at jumping on the buzzword-bandwagon? In all seriousness, this article doesn't say much, although it speaks volumes about the technical ignorance of the author.
Java was born as hype, remained hype, and fell from grace because it didn't live up to the hype. Among other things, the concepts of Java and Virtual Machines weren't exactly new; in a lot of ways, it was just an interpreted language like Lisp, excepting compiled into bytecode to make it more proprietary.
Regardless of anything Java has and has not done, Java has about jack shit to do with Linux, other than you can run Java under Linux.
Just my $.02.
-[Blaine]- "'Oh dear,' says God, 'I hadn't thought of that,' and promptly vanishes in a puff of logic."
Well, it does lead one to think of the definitive way for Microsoft to kill Linux.
They develop their own version, release it as L++, base it on C libraries that are incompatible with both libc5 and glibc2 and in various other ways make it flashier but impossible to implement next to real Linux, but claim it's the same.
It should go without saying that they never support this product, fail to mention it on most of their web pages, remove all references to it from their Knowledge Base, charge exorbitant prices for it (while also distributing it free in computer books) and include the 49.7 day bug.
Then, after dividing the market, there's no way Linux could sue them. I mean, Sun has hundreds of lawyers. The best we could do is throw a couple FSF volunteers at them.
In a matter of six months, MS could have Linux deader than OS/2 and be free to force Win2000 upon everyone (because we know Linux people aren't going to bite the bullet and use *BSD).
Conquering The World 101. They've done it before, they'll try it again. Unfortunately.
-Chris
the obvious problem with this thinking is: if they did release an "L++", why would anyone use it? The obvious answer is that it would supposedly have some kind of interesting features that no other linux flavor has.
But then you kind of run up against the fact that if it's released under the GPL, then it is trivial for, say, Redhat, to pull out all the nice features and incorporate it into redhat. Or for someone to hack together a compatible version of L++ and release it.
And if it isn't GPL, then, well, it doesn't have a gnat's chance in hell. Never mind for a second what exactly it is that wouldn't be GPLed; it sure isn't the kernel, cuz they'd have to rewrite it from the ground up to be free of the GPL virus. And if it's a replacement for GNU GCC/Make, well, it has to be compatible enough to still compile the kernel perfectly.
Whatever a non-GPLed L++ would hypothetically be, it would get no support from the open source community. Meanwhile the closed source community would consider it a waste of time to develop for, since they can reach a wider audience by writing for NT anyway.
Irritable, left-wing and possibly humorous bumper stickers and t-shirts
With a couple of straight lies: 'And Microsoft's Edwards says that Linux lacks many advanced
capabilities, such as the ability to run on computers with multiple
processor chips.' Linux runs on 4 different SMP platforms: Intel, Alpha, Sparc and PowerPC. On Sparc64 Linux has been demonstrated to boot/run with 64 CPUs. NT runs on one SMP platform (Intel). The biggest SMP box NT ever ran on is 12 CPUs. Draw your own conclusions about SMP capabilities and hardware/vendor independence.
--Coke
I can't believe I'm seeing yet another article that says linux won't run on multiprocessor machines (YAATSLWROMM). The direct quote is "And Microsoft's Edwards says that Linux lacks many advanced capabilities, such as the ability to run on computers with multiple processor chips." Obviously, linux can run on multiprocessor systems... the issue is that at present it isn't very efficient when handling more than a few processors.
/. have no credibility to back them up, and that there is no code to research and fact check stories. But when dealing with tech especially, old media is as bad or worse at spreading rumor and misinformation as the new media. Sheesh.
In net media vs. old school media arugments, the old school guys are always arguing that the net media and forums like
He said, "You'll be able to tell your grandchildren that you helped assemble the first NT supercomputer," and I cringed.
This is something I've been saying for a while: Linux is great, and I've used it for a while, but it is *not* necessarily going to achieve world domination. Technological superiority doesn't always win. In the end, it's a business decision, and denying that will get us nowhere.
The possible key difference is open-sourcing. If all bugs really are shallow to a sufficient number of eyeballs, then we should continue to see far more robust software. That can impact the bottom line for lots of corporations, and thus gain substantial extra market/mind share. But as far as the individual desktop goes, Linux (while always a viable alternative) won't succeed purely on the basis of being more robust. Users need out-of-the-box hardware support and the ability to run the software they want to run, from productivity apps to home applications (read: Quicken) to, yes, games.
Java hasn't won primarily because Sun seems to have forgotten that while technologically superior doesn't always win, it is important. I carpool with a Java programmer, and whenever I make the point that Java is slow, he tells me to run a performance VM. There's the flaw: a "performance" anything should be for excellent results, not acceptable results.
Linux needs to build on its technological achievements, always improving, to win corporate share, and on its fundamental usability (HW support, apps) for individual use. That's the way to avoid the Java trap.
"You can never have too many elephants on your team."
Everyone is missing one vital point.
Linux has its place
Java was and is a great idea. Sun saw it for what it is. They marketed it, they put it out there, they tried to take over the world with it.
They got it an image problem when they stuck it into buggy browsers on slow machines, but that's a discussion for another time.
Linux didn't start out like that. Linux wasn't some germ of an idea that needed incubation and careful marketing. Linus didn't make deals with $large_software_company to include it with every copy of $popular_application. He coded it and stuck it out there.
And it stuck.
Guys, Linux has nothing to prove. The mindset is already there. The people who need to trust it most, trust it. We know when it can be used. We know when it can't. We can prove it.
Once we have that, we can work on the PHBs, and no dodgy benchmarks are going to change that.
This is not some ivory-tower dismissive. It's a reminder: in an industry awash with false promises and vapourware, Linux delivered long ago. It will remain as long as it is useful.
Dave
--
First, Java was doomed as soon as Sun decided on a closed-source format that invited companies like HP and Microsoft to 'modify' it without incorporating the changes into the 'official' distribution. Sun should have followed the appropriate standards committee, and forfeited the ownership of Java to avoid the fragmentation of Java.
Java has yet to deliever the promise of Write Once, Run Anyway. The poor Java support of systems beyond Solaris and WinXX, the existence of inconsistant versions with the same major version number (1.0 and 1.2) are just some examples.
Finally, and most importantly, Java is still bug ridden. While the rest of their developement is closed, Sun fixed bugs in an open-source sort of way with the Bug Parade. The problem is, no release ever clears out all the bugs in the bug parade. Does that seem wrong to anyone else (and reminisent of a Gates statement a while back?)
Thus, Java is currently difficult to work with outside of an intranet environment. Sure, I think it's a great language, if the above problems were non-existant, and it won't die in any short time, but some of the hype on Java is long-dead. It is NOT the next Windows killer.
Now look at Linux. It's open sourced, no one owns it, and the key element, the kernel, is controlled by one man in the open source model. Sure, you could go off and make your own modifications, but that's got to be a lot of work to do those. It could conviecibly happen but I doubt anyone wants to undertake it.
The kernel works for nearly every system, and with the modularity of it, can be easily met to fit the needs of any system. And save for the libc5 to glibc2 switch, nearly every problem works under every version of the kernel. And this is true for multiple architectures as well.
And of course, Linux may have bugs, but once found, they are generally squashed and patched within a week, if not shorter. No other company on the planet can promise that.
"Pinky, you've left the lens cap of your mind on again." - P&TB
"I can see my house from here!" - ST:
Besides, Java was touted as a key tool in another heavily hyped movement toward ''network computers.'' Instead of keeping software on each machine, people would use small, cheap computers that would get all their software from a central server. Java is well-suited to this approach.
Java is well suited, and Sun came up with a good idea here. But surprise! Microsoft stole the idea, and their name is very similar to Sun's Name.
Charles Fitzgerald, director of business development in Microsoft's software development unit, says the Linux hype has already peaked.
Yep, that may be true. And the Windows hype peaked WHEN? Ohhhhh, about 1992.
Fitzgerald points to recent benchmark tests by the research firm Mindcraft Inc., which found Windows NT performs a variety of tasks faster than Linux.
I'm pretty sure I don't even need to comment here. Fitzgerald points to the Mindcraft bullshit because he's grasping for anything he can.
And Microsoft's Edwards says that Linux lacks many advanced capabilities, such as the ability to run on computers with multiple processor chips.
THAT pissed me off a lot. That part is very misleading. Linux contains Multi-Processor support (and has for a long time). It may not be the best it could be, but it's being worked on. (also, I think Linux has more advanced capabilites then Windows* ever did.)
But Fitgerald says Linux still isn't ready for the heavy-duty tasks that Windows 2000 is designed to perform.
Like Windows2000test.com, I suppose.
-- Give him Head? Be a Beacon?
-- Give him Head? Be a Beacon? :P)
(If you can't figure out how to E-Mail me, Don't.
- Java is a huge success -- just not where it was originally hyped. People don't see Java applets on every server they visted et viola that must mean Java failed. Well, it didn't.
- Java is alive and well in corporate enterprise computing. In fact, it's probably the hottest thing in client/server programming going right now.
- Java isn't any slower than Perl or Visual Basic or any other wildly popular not-compiled-directly-to-native-code language. Sure, it was pretty slow two years ago, but that was half it's commercial lifetime ago.
- Sun is not the only major corporation pushing Java. There are IBM and HP, as well as open-source Kaffe. This is not Sun's "OS/2."
- Remember Sony's "failed" Beta technology? Remember the "failed" DAT format? Well, both are staples of professional production facilities.
In a way, the comparison of Linux and Java is fair. Linux may never overcome the drool factor needed to beat Windows, just as Java has yet to overcome the drool factor needed to beat Visual Basic (or PerlJava was a language for programming garage doors and refrigerators. Somewhere along the line, Sun figured out that it could be used to make dancing pink elephants appear in a browser. Turns out that the demand for pink elephants in browsers isn't that great and can be handled by animated gifs. At any rate, Java got pushed as a one size fits all solution for problems it wasn't ready to solve. As a result implementations were buggy and slow and a lot of people (myself included) stopped caring. Linux could face the same fate. If you push people to use it when they aren't ready, they'll dislike it. Linux got where it is by being a good tool for certain tasks. The users extended it and it got to be a better tool - one that could be used for more tasks. Perl developed a following the same way. Both are strong while java is floundering and requiring massive efforts on the part of Sun. The lesson here is that pushing a technology on people who aren't ready for it is a good way to damage it. Next time you're hyping linux as the perfect solution, pause, take a deep breath and ask yourself if the people you're trying to convince have the time, energy, motivation and know-how to learn to use linux effectively. If they don't you're hurting the movement instead of helping. Also ask yourself if Linux is ready to handle the job you're suggesting it for. If it isn't, back off, get out your compiler and work on it.
--Shoeboy
As others have pointed out, the differences between Java and Linux are so large that a comparison of the succes of these two technologies is essentially useless. If you need to compare the impact of Linux to something roughly comparable, OS/2 would have been a far more logical candidate, and the gist of the article might have ended up much the same.
But Linux has characteristics that lift it into a category of its own. The most important one (IMHO) is that it started as a low key project and that it had a chance to surprise people in a positive way. By the time that the name Linux started to become widely known, thousands of people were already using it seriously for all sorts of purposes, and because it's essentially a reimplementation of Unix, there were a lot of experienced developers for it from day 0.
Java on the other hand was positioned as the second coming that is going to fix all your problems with portability and reusability and sex, and in retrospect how could it have done anything but fail to deliver on those promises? I'm not claiming that it is dead, mind you, but so far it definitely has not become what is was once made out to be.
But now I have customers asking me 'Do you really think this Linux thing is going to make it?' and by make it they mean 'beat Windows'. I simply respond that Linux doesn't need to beat Windows, that it can live in perfect harmony in a Windows network, that it has nothing left to prove for the purposes these customers are looking for, that I've been doing useful work with it for years and that for them to have a reliable and inexpensive server it doesn't make the slightest difference whether or not common people will decide to use it on their desktop as well. Sure, the current Linux 'hype' makes this an easier sell, but essentially the difference is that Linux is not just a promise, it's a working, proven technology.
I hope that Linux learns from Java's (lack of) experience and will not suffer from Java's versioning nightmare.
Sun's doing a fine job of killing Java all by itself - no help required from Microsoft. I've got 5 versions of Java source code for my product to run on 5 different versions of Java Virtual Machines used by my customers - from 1.02 through to 1.3 beta. And no, for practical reasons Sun and Java Lobbiers cannot appreciate - they cannot all simply upgrade to the latest version because they have Java software version dependencies of their own. Frankly, this "write once, run everywhere" thing just ain't what it's cracked up to be.
IMHO, software should be written _primarily_ with "solving the problem" in mind, with readibility/maintainability/reusability also as the foremost concerns. "Optimizing for performance" should be pretty far down on the list. That's why Java's relative slowness doesn't bother me at all.
As far as I'm concerned, the primary place one should look to improve the speed of an app is hardware. We always joke that if you want speed, don't write in Java, use C! Then some other yuckster says, "Be a real hacker, only write Assembly!"
Certainly, an experienced coder can avoid most of the common pitfalls associated with writing an app in a lower-level language. However, experienced coders are a scarce resource, and even they aren't perfect. And when you choose a language because you are optimizing _for_ performance in software, that programming style is more likely to result in buggy systems than optimizing for maintainability.
Higher-level languages generally come about to address deficiencies in the lower-level languages, and almost without exception, they carry with them a performance penalty. But, in Java's case, provided that you don't have a buggy JVM, you can be guaranteed that an app written in Java is immune to such problems as buffer overflows that are prolific in the C-application world.
So, when people tell me, "Don't use Java, it's slow," my immediate response usually starts with, "So What?"
"UNIX" is never having to say you're sorry.