Slashdot Mirror

Oh, goody! A list of features! Can that list remind me about my wife's birthday? Well, no. It can't do anything. It's not software. It's just hot air.

Do I smell JOS here? (I know that site vanished two or three years ago; that's the point).

Where's the product? I see an announcement, and I see a public discussion about what people might like to do, if by some quirk of fate they were to shut up and start writing code. I see an elaborate "Mission Statement" located on a slick-looking web site. But I don't see any code. I don't see any output at all.

A lot of people are going to jump into this and start arguing endlessly about features, programming patterns, methodology, licenses, and all manner of irrelevant crapola. No functional product will ever emerge, because they're doing it backwards. This is a truism, but people never seem to remember it: If you start with code, you may end up with something. If you start with a flashy web site, a vague 400-word mission statement (any "mission statement" longer than ten words is a death sentence), and a public call for sidewalk superintendents to gum things up, you'll never end up with anything. The latter approach is best described as Announcement Engineering. It's been tried, and it has failed, and it has been tried again, and it has failed again.

Why does it fail? Because if you start a discussion, you'll get people who specialize in discussing things. If you start a slick web site, you'll get people who like slick web sites. Both of those groups are self-selected for parasitism and uselessness. If, on the other hand, you start writing code, you'll get people who like to write code. If writing code is your goal, people who write code are the people you want. Not sidewalk superintendents. Not methodology-obsessed BS artists. Not visionaries, not self-appointed "philosophers", not "online community" addicts, not Open Source rock stars, spokesmodels, public figures, beloved elder statesmen, or opinionated teenagers -- all of which are going to descend on Kapor like a horde of locusts. For programming, you want programmers. But all the programmers are somewhere else, working on projects that actually exist.

Devil's Advocate Dept.: What about the GNU Manifesto? It does superficially resemble Announcement Engineering, but one crucial ingredient of AE is missing: Stallman never asked for anybody's goddamn opinion, and to this day he still hasn't. He doesn't want to discuss anything with anybody. When Stallman wants to know what you think, Stallman will tell you what you think. He never asked anybody for an opinion in that announcement; he just asked for code.

actually it is

EROS. No, it isn't an OS that displays pr0n. It stands for Extremely Reliable Operating System and is used as a test bed for new OS enhancements such as OS persistence and token security. Besides, these guys get a real kick out of showing how they can kick the plug out of the wall and have their machine back up moments after they put the plug back in.

Another OS of interest is JOS, a Java based OS. While I agree with them in principle, they defined too large of a scope initially and ended up drowning in their own specs. Maybe one day we'll see an awesome OS out of them, but not today.

Look at all those OS kernels and low level software. Is there there any way that Linus will throw away his kernel for a Java replacement?

Ok, I hear your arguments. But it is not about choice of language. It's about choice of model.

In C, you're developing in structured programming model. In C++ you're developing for object oriented programming model. C is the best known language in its model. It has easily killed BASIC, PASCAL or any other competitior. And Java is a very strong language in its arena it seems like it will be able to kill C++ in a year (as the article says) and any other competitor (Smalltalk, ObjectPascal) is already out of the way.

But until we change the way this machines operate object oriented model will not dominate. Thus in a few years we may see KDE replaced with Java version. But it's unlikely to see Linux replaced by JOS (http://www.jos.org) in even twice that period.

The closest thing that's currently even feasible is JOS (http://www.jos.org/)

This may sound like a troll and get modded down, but who cares? Since Java is OOP, wouldn't it be easy for Sun to just change the implementation of the JXTA framework to "backdoor" oar peer to peer architecture. Not that Sun would do something like that, only M1cr0$0ft would do something that evil *(and we all know M$ doesn't use OOP since they don't use Java, although they're oracle db is OOP, IIRC). Anyone have an idea how a truly free version of this and Java could help us circumvent this dastardly tactic of the Coprporate Republic?

Actually there is an effort being made to create a native java os called JOS. I don't know how mature it is right now since I am not part of the project, but it sounded promising the last time I checked their website. They also use a very liberal/open source/"free" method of development and online collaboration known as Wiki.

Hey, if you want low-level Java, why don't you check out www.jos.org.

JOS is an effort to produce a full Java operating system for many architectures. I think the point is to load a JVM "kernel" at boot time and off you go. :-)

-----
"People who bite the hand that feeds them usually lick the boot that kicks them"

Can I ask why text files with XML structure are not good enough?

Because it just pushes the burden of structure interpretation up to the application programmer. Ideally, when you're programming in a high-level language, you should be able to handle any objects in the system through a consistent set of primitives; the actual representation of your data on the computer should be implicit and left for the system to decide, unless the programmer himself specifies otherwise.

I see no reason why the Unix security model in particular cannot be extended to support safe, shared control of user applications.

The reason is that the Unix model (like most others today) is based on a "stupid" executive which, essentially, just goes around giving control to any piece of binary data which is tagged as "executable" and has the right headers (a.out or ELF). The web site for ETH Oberon goes into significant detail about an alternative executive model which is much smarter.

I think Sun calls that Java. :-)

Well, all of Java's essential flaws aside (and Ghod knows there are plenty of those :), I dare claim that an OS based on a Java VM would be far superior in all the aspects discussed here to our current crop of C/C++-based ones. (I'm not sure whether the JOS project for a Java-based OS is still going on... they seemed to have the right idea, kind of.)

??? Please elaborate.

You obviously weren't here the last time Slashdot had a "X sucks/X rocks" flamewar :) If you're interested, you can check the archives for the last two weeks or so. (Evidently, I stand on the "X sucks" side :)

After all, you can try all the variations you want, but you always end up with a flat cylinder for the wheel.

But that's already assuming that the software you started out to imitate is comparable to the wheel. Maybe it's not; maybe it's comparable to those squareish rock wheels you see on "The Flintstones" :)

And it's why we still use X11 and the WIMP model. They work well.

Well, "works well" is subject to discussion. IMAO, "works just barely well enough and it'd be a damned shame to start all over" is more fitting. And what you call "pragmatic", other people call "worse is better". (Then again, this kind of philosophy, "if it works, however barely, don't fix it", seems oddly fitting for a man who recently claimed that systems research was over. Again IMAO, that's akin to (Lord Kelvin's?) claim in the 1890s (less than two decades before Special Relativity and Bohr's atom) that essentially all of physics had been figured out, or to Hilbert's call in the 1900s (three decades before Gödel's Incompleteness Theorem) for a mechanic universal theorem decider.)

As for your comments regarding UIs (i.e., the part of the discussion which is actually on-topic :), I really think you should take a look at the unorthodox projects I mentioned in some other post. A look at Tunes' User Interfaces Review page should also prove interesting.

wtf "Orange Book B1" was. I went and found the following defin ition:


labeled security protection
The B1 system class. B1 (and higher) systems support mandatory access controls. The system architecture must more rigorously separate the security-related portions of the system from those that are not security-related. Documentation must include a model of the security policy supported by the system. It need not be a mathematical one, but it must be a clear statement of the rules enforced by the system's security features. Testing is more stringent.

Take a look at JOS. It is an open-source project and if you really understand it, you will know the real power of Java

not to say I dont agree that Metcalfe made a few erroneous mistakes in his 'article', but what he may have been referring to as java might have been the JavaOS project. I've got a friend who's working on it, and according to him, once some kinks get worked out it could quite possibly be the next revolutionary os..

just my $.02

Justin