Slashdot Mirror
Ken Thompson Receives Kanai Award
JerseyTom
writes "Ken
Thompson, co-creator of UNIX gets the first
Tsutomu Kanai Award which includes a 20-million yen
grant from Hitachi Ltd. " Silly me, I thought Unix
was obsolete. Guess I was wrong *grin*.
← Back to Stories (view on slashdot.org)
I don't believe Ken uses Unix anymore. They all use plan9, or Brazil, actually. One of the reasons I have mixed emotions about the rise of Linux is its old-style design. But it's better than the Steaming Turd from Redmond. While I'm at it, C and C++ aren't that thrilling either. I'm shocked at how slow my industry is progressing.
It doesn't say that they're necessarily software patents. Anyone want to check?
It's funny how Apple dusted off 10-year-old A/UX, now called Apple OS X, and proudly anounced somthing "new": a 30 year-old operating system!
Unix is still useful! Thanks, Ken.
He also invented the Internet and brought computers to every household in America.
The words "UNIX", "Internet", and "America" are trademarks of Microsoft Corporation.
Actually, surprisingly enough, I think Ken Thompson own's the patent on Just in Time compilation, or as old-school programmers like to put it, On the Fly compilation. This was related to a project of his which compiled regular expressions into actual machine code to derive better run time performance. (See The Practice of Programming by Kernighan and Pike)
Well, at least he isn't like Unisys and he obviously hasn't been enforcing the patent.
He did.
To address your concern: :)
I once read somewhere on the net an interview of Linus. He mentioned that he looks at developments in Plan9 and adopt ideas from them as well. So Linux may not be as obsolete as you think.
I am pretty sure my memory serves me well in this case. Why? Because afte reading that interview, it started me off on a frenzy looking for more information about Plan9, Brazil and Inferno. Didn't know anything about them b4 that.
Now that would be funny if Dennis called Gore on his remark that "I invented the Internet" and asked him a few questions about the UNIX TCP/IP stack ;)
You all are talking about how computer science
works slowly "like mathematics" or should not
be "thrown away" just because some stuff is
old. The world is advancing at an alarming rate,
especially in the computer industry. I admit
that applications and developers need a good
stable operating system, but come on, it takes
forever to do anything in Linux, the only reason
I use it is because it only has to have the stuff
I want in it... but if it wasn't for the fact
that some nice cutting edge stuff is being done
in Linux, I would not use it. A good unix-compatible but newer-implemented OS will come
along rather soon (I hope) and kill all others.
Oh.. and you can't compare ANY Windows sytem
or any other system to Linux unless that OS
is a UNIX OS, because that is what Linux is,
although Linus swears it is not, if it was not,
then it would be mmuucchh better.
"I'm shocked at how slow mmy industry is progressing"
i suppose this man wants everything to be updated every other year....
hey why don't we just abandon all the standards and start everything from square one
maybe you should advance to the porn industry, lots of new innovations there!
Beacuse any geek out there better call
his bluff. Or wait till he's in a press conference
http://plan9.bell-labs.com/plan9/
Fredrik Henbjork
http://g204.ryd.student.liu.se
This a good reference as well.
Hello
Did I mention I love Dejanews?
JP
What is it that takes "forever" to do in Linux?
:-) I'm moving my projects to Linux.
Well, my experience is that when you need to do anything system related under Linux and you know your ways around the system, you just do it and that's it! You never click around among unresponsive dialogs guided by crap docs that are not meant to be of any use when you get into *real* trouble, and then reboot and hope your system doesn't blow up, as with certain other OSs...
Programming? Hardly. Unless you're looking for some visual development tools that's more or less meant for non-programmers, Linux or any other UN*X like system is about as clean and easy to code for environment you'll get. Yep, I'm hacking both for Windoze and for Linux, and I've seen some other systems as well. (AmigaOS was a nice one, BTW.
Applications? Well, if you wanna' blame the OS for the lack och some kinds of applications... It will certainly NOT help to introduce yet another, slightly different OS.
What would any other UN*X-compatible OS have to offer that would make it kill all others? If you're thinking about nice configuration tools and that sort of stuff, you should really stick with Linux, 'cause that *is* the system for which people are right now developing that kind of tools!
A UN*X comaptible kernel is a UN*X comaptible kernel... Linux is one of the very best of those on the technical level, and that's all that matters for that part of a system. It has to do what it's meant to do, and it must be rock solid. Very few kernels fulfill those requirements to the degree that Linux does.
The fastest way to create a very good system is to improve the system around it (installs, desktop, config tools...) - not drop the whole thing and start over AGAIN! Get real... Linux and the systems based on it may not be theoretically perfect, but you won't get a better base to build a good OS on in years.
//Harmony
At least he doesn't demand that every UNIX out there has a moniker to him. And he invented the stuff and didn't simply clone it.
It will be first released in Win98 OSR whatever and will allow a win98 box to share its internet connextion among other windows machines.
Gee, I wonder how long until MS invents a true pre-emptive multitasking, multi-user OS? The worls will just have to wait and see.
So I've heard. He is ...
Why not use BEOS.
According to the article he holds 6 software patents. How dare he! He must be ignorant, according to many slashdotters.
No, I'd say your post pretty much defines ignorant in the opinion of most slashdotters
If it's mediocre, we toss it aside. If it good, we incrementally improve it. If it's fantastic, we don't improve it. Unix and C/C++ work well, so we incrementally improve them.
C and C++ are mediocre at best. They are inflexible, lack abstraction, and are conducing to writing buggy programs (often with security holes due). It takes way too many lines of code to do anything useful in C/C++. The languages also have many design flaws.
Do we need a new OS or language every year?
At least once every ten years would be nice.
i think you are misled, like most people, about what computer science is. the commercial sector has unfortunely succeeded all to well in fooling everybody into thinking that software is "technology" and you should upgrade to the latest and greatest platform because all older "technology" is obsolete. this simply is not true. when i argue that unix shouldn't be replaced, i am refering to the api, not the user inteface. indeed one of the brilliant design choices of unix is that it is completely ui independent. why should we change the api? won't our new operating system still need to provide functions like read(2), open(2), and creat(2)? should we forget all we know about searching or sorting and instead implement all new algorithms from scratch? we will end up right back where we started. or perhaps we should give them all new names? easily done, however if you don't want to work with an already existing function i suggest that instead of rewriting something you implement an abstraction layer above
said function. what makes unix a great operation system IMHO is that everything is designed with simplicity in mind. it dictates as little as possible to the programmer. since the ui is not a part of the os, we can build anything we want instead of being forced to use a gui or cli. i think brian kernighan described this philosophy quite well by saying (on why unix uses \n to terminate lines vs CRLF or records), "When a more complicated structure is needed, it can easily be built on top of this; the converse, creating simplicity from complexity, is harder to achieve". all of this is not to say computer science doesn't change. it does but slowly. again take the unix api for example. it certainly isn't the same as it was in 1970. those parts that were unused faded away into obscurity, others slowly changed with time to reflect current ideologies, and yet others were added as some idea proved true and useful. back to your comment that "it takes forever to do anything in Linux", i assume you mean that
whatever ui you are using is less than satisfactory in supplying the functionality you desire. this comes down to the fact that each ui abstracts the machine at different levels. obviously you can't manipulate objects of a certain type without a tool that deals in those objects. for example you can't directly and conveniently access an int or char in scheme and the same relationship is true of c and vector? or complex?. what this boils down to for ui's is that, like c and scheme, the is a trade off between ease of use, munipulatable objects, and performance. while a gui may be "easy" to use it is also a resource hog. however from a different perspective it is much easier to provide a loop and work with many objects at once from a shell. so until your computer can read your mind their won't be an interface that is all things to all people. one last point, i think a great example of what computer science is and how it has changed is The Art of Computer Programming series by Donald Knuth. if you want
see what has changed then observe the differences from edition to edition of these books.
You're scaring me with this whole 20 half-man, half-horse beasts thingie. Reminds me of this bad dream I had were an evil alien scientist had this chaos theory and built, with the help of 20 centaur mathematicians (these weren't like the ones you see on tv that chop girls in half and make elephants disappear, but real bad like ones with fire balls and evil spells), a turing machine that attacked innocent planets 'n stuff.
--crackhead
Kernighan & Pike, _The Practice of Programming_, Markov chain algorithm on the Book of Psalms
C 0.36 sec 150 lines of code
Java 4.9 sec 105 lines of code
Perl 1.8 sec 18 lines of code
(See Chapter 3 for details. CPU/memory requirments not recorded. Numbers above on 250 MHz R10000). Is there no place for C? Of course there is. Is UNIX the best of all possiblie operating systems? Of course not, but it does mostly work. Use what you need depending on the circumstances. (What's Plan 9 written in? And Inferno? Java? Do Lisp and Smalltalk suck *because* they're old?)
That has to be, bar none, the worst pun that I have seen on /. yet...
US $350 per user... OUCH!
Kernighan & Pike, _The Practice of Programming_, Markov chain algorithm on the
Book of Psalms
C 0.36 sec 150 lines of code
Java 4.9 sec 105 lines of code
Perl 1.8 sec 18 lines of code
(See Chapter 3 for details.
That's not the greatest collection of languages. A Dylan version of the program would run nearly as fast as the C version, but would take many less lines of code. Where can I download the benchmark? I'll port it.
Linux or any other UN*X like system is about as clean and easy to code for environment you'll get.
Surely you jest. The unix API is horrible.
The BeOS API is pretty nice.
computer science, much like mathematics, doesn't change drastically
but rather it evolves slowly over time. new advances usually add
to the knowledge base, not replace it. while calculus was an
extremely valueable addition to mathematics it certainly didn't
supersede algebra or geometry just like object-oriented
programming didn't supersede other philosophies like structural or
functional programming but rather enhanced the general tool set of
computer science. as long as computers remain functionally the same
i don't see why a language such as c would no longer be needed. sure you could
design a new language, but if it was at the same abstraction level as c, ultimately it would need to be able to describe the same tasks that c already does, thus just resulting in a new syntax. programming languages really only differ in abstraction level and support for different design philosophies.
How about:
Alan Turing, inventor of Turing machines or Church that created lambda calculus. They were both 20 th centaurians mathematicans. One example of mathematic from our century: Chaos theory, signal analysis, fractals and not forgetting many of the fundamentals that computers are based on.
No, but he did discover some of the software Al-Gore-rithyms used in Unix.
Im not familiar with Plan9, does anyone have some good links (technically oriented) to read? Thanks
Check out the Inferno FAQ. You may have to search for "Plan 9" in the text.
--
--
=8^
What I remember from "Life with UNIX: A guide for everyone" is that they didn't know by the time that software patents were possible, so he registered the patent (of the setUID mechanism) as an analogous hardware device.
Why did this get a -1 score?
--
All Glory To The Hypnotoad!
But Al Gore did invent the Internet!
--
--
I noticed
It's getting about time to leave everywhere
OS X isn't A/UX, it's Rhapsody, which is in turn NeXTSTEP. Of course, I'm no expert on NeXTSTEP and it is entirely possible that it is a derivative of A/UX, but I don't think so because it's all 4.4 BSD-ish and Mach microkernely and stuff.
Plus I severely, *severely* doubt Steve Jobs would have used anything Apple-derived at NeXT given the circumstances under which it was created.
--
Posted with Mozilla
"The award, endowed with a 20-million yen grant from Hitachi Ltd. in 1995, was established in honor of Dr. Kanai, who joined Hitachi as a researcher in 1958 and retires as president this month."
It's nice that he gets honors like this, and the National Medal of Technology that he and Ritchie will recieve from Clinton next month, but it'd be nice if someone gave him a cash award, too. I'd thought it'd finally happened. Alas.
Professional Wild-Eyed Visionary
The famous one is for the SETUID bit mechanism -- first software patent I ever heard of, way back when. Fortunately Bell Labs assigned this patent to the public domain almost immediately, so it never hurt anyone.
Professional Wild-Eyed Visionary
What about Graph theory? There are some modern 20th century advances in graph theory..
What does "old" have to do with obsolete? Any OS out of Redmond is more obsolete than Unix, as long as billborg sets its major design goal to be different from everything else so he can daydream about staying in control.
Technology doesn't become obsolete just by aging.
--
Infuriate left and right
OS X is _not_ based on A/UX. It's based on NeXTSTEP. Having run both A/UX and NeXTSTEP (on black hardware), I can tell you that they have very little in common. (other than both being unices) A/UX is a pile of crap, NeXTSTEP is nifty.
One of my AT&T UNIX books has a reprint of a software patent. Unfortunately it is packed away in a box. I believe it was for the user/group/other file permission scheme. The patent was put into the public domain.
Mea navis aericumbens anguillis abundat
My understanding is that OS/X runs MacOS applications much the same way A/UX did. (That's the "Blue Box", I think.) But that has more to do with hooks built into the MacOS than the Unix side of things.
A/UX was a System V port to Apple's 68K hardware, so it's definately not the same as NextStep/OSX
--
Business. Numbers. Money. People. Computer World.
UCC rules! I get it in my mail everyday so I can tell what the equivalent of US $0.02 is in yen...;)
-lx
According to the article he holds 6 software patents. How dare he! He must be ignorant, according to many slashdotters. The thought that he might want to make a little money off of his good work. He is denying you freedom by creating anything that isn't 'free software', according to RMS. It must be detrimental to society, right?
In all seriousness. For all the talk of innovation in 'free software', I find that it doesn't keep pace with even Bell Labs.
PS: Where is the GPL'd Plan9 equivelent?
Just to let you know... MAC OS 10 will have NextSTEP functionality integrated into it, including remote access stuff. Maybe this will rival NT.... Linux even.
OS X is not NeXTStep--it's more like 4.4BSD with that phun Microkernel stuff from CMU thrown in. If it's NeXTSTEP, where's the incredible graphical environment you got on the NeXTS?
--jon. Postel is dead. May we all mourn his, and our, loss.
20,000,000 JPY = 166,251.04 USD
Courtesy of The Universal Currency Converter
Taral
WARN_(accel)("msg null; should hang here to be win compatible\n");
-- WINE source code
Plan 9 is based on 3 ideas: all resources are files referred to by the same file-name-space; (2) a filesystem can be added to the file name space at an arbitrary point via the mount command; (3) the file name space is customizable on a per-process basis.
The first 2 ideas come straight from Unix. The third idea seems to be a genuine advance, and Linux simply does *not* have it. in fact, to benefit from this third idea, you would probably have to rewrite most of Linux and its apps--it is not just something you can add to Linux-- and it really does seem like a good idea that should be included in all new serious operating systems from now on. especially in a networked environment, it confers a useful kind of flexibility. Plan 9 is not change for changes sake or change for the software-owners sake a la Microsoft and other software vendors. it is an honest attempt to improve the way people use computers.
Unix is still valiantly holding on in it's fight to remain dead.
Cheddar Cheese
This is technology, not art. If it's mediocre, we toss it aside. If it good, we incrementally improve it. If it's fantastic, we don't improve it. Unix and C/C++ work well, so we incrementally improve them. We don't toss them out simply because they are old.
Do we need a new OS or language every year?
A Government Is a Body of People, Usually Notably Ungoverned
Oh, thats okay. UNIX has been obsolete for the last 30 or so years. Not to mention that the press pronounces UNIX dead atleast twice every year!
On the subject of patents, there's an interesting article, Patent Reform Is Not Enough over at the GNU webserver. It's a reprint and it appeared originally in the GNU's Bulletin vol.1 no.13 (June 1992).
This one?
Sure, they went out of their way to make it look more like MacOS 8, but its still largely the same.
Gee, I wonder how long until MS invents a true pre-emptive multitasking, multi-user OS? The worls will just have to wait and see.
What, Windows Terminal Server doesn't count? It's multi-user, sort of.
Plan9 is not even being developed by Bell any more. The most recent FAQ is over a year old, and as for Brazil, the next release of Plan9, they say: "Work on Brazil has been stalled, as the developers have been pulled off the Brazil project to work on Inferno. The system is not in a state that is releasable, so there probably won't be a Brazil release in the foreseeable future." This is the OS of choice for grandaddy uber-geeks? :/
UNIX will still be obsolete thirty years from now. You will still have the requisite press announcements of UNIX's demise.
AT&T ownned it and couldn't kill it. Microsoft can't kill it. IBM/TransMeta will not kill it.
by daywalker: Al-Gore-rithyms -- priceless
To get an idea of how slow, do you know *any* mathematics from the 20th century? Fundamental advances are *hard* and do not happen that often.
>A good unix-compatible but newer-implemented OS will come along rather soon (I hope) and kill all others.
;-)
At the moment, that OS *is* Linux. AT&T has always diligently protected the trademark UNIX. There was a long and bitter battle between Berkeley and AT&T over the independently derived BSD Un*x. Linux is an independent implementation of the Un*x standard. Because of the trademark and copyrights, you have *BSD and Linux, not UNIX. When that better OS comes along, the good parts will find their way into Linux and into *BSD.
From the introduction of FreeBSD. "For copyright reasons, FreeBSD may not be called UNIX. You be the judge of how much difference this makes."
Linux *has* to insist that Linux is not UNIX.
GNU, GNU's Not Unix. Same thing.
Linux *is* UNIX. You just can't call it UNIX(tm).
>The world is advancing at an alarming rate,
especially in the computer industry.
Right. It's called Windows(tm). It is alarming.
And if he didn't, he should have.