| |||||||||||||
HehHehPosted Nov 17, 2011 21:57 UTC (Thu) by khim (subscriber, #9252)In reply to: Interview with Andrew Tanenbaum (LinuxFr.org) by welinder Parent article: Interview with Andrew Tanenbaum (LinuxFr.org)
On what planet does he spend most of his time? Ivory tower^Wplanet, apparently. You see: AT has this insanely great answer (it's not 42 but MINIX for a change). Now he just need to find the question. Some strange suggestions: There are a lot of applications in the world that would love to run 24/7/365 and never go down, not even to upgrade to a new version of the operating system. This is valid concern but all these application are legacy code which was written in time when hardware was large, expensive and reliable. Today you achieve the same thing with redundancy and distributed computing. And legacy code can not use MINIX anyway so... what's the point? What most ordinary users want is that it ALWAYS works. In engineering terms I would say this could be expressed as mean time to failure of 50 years. Well, this is big problem, true. When activity manager in Android or Compiz on desktop decide to wreak havoc - this is big issue. Ah, you mean you are trying to fix problems in kernel which brings about 1% of problems to "ordinary users"? Well... this is novel idea, I'll grant you that - but how wise it is? If you want to change the memory manager, only one module is affected. Changing it in Linux is far more complicated because it is all spaghetti down there. And this is really, really, REALLY important. People want to change memory manager every five minutes. But they never plug USB handsets, never play 3D games, never run heavy SMP databases on these 24/7/365 systems, you know. These things are trivial in comparison. L4 is very successful. Gernot Heiser in Australia managed to sell it to Qualcomm, so it is now running on 300 million phones in the world. Not bad for a microkernel. The biggest difference is that we run the entire operating system as a collection of servers in userland. They mostly run Linux as a whole unit in userland. This means that a single bug in Linux can crash Linux. In other words: we've created enormous buzz around the microkernels and managed to shove useless piece of software in millions of mobile phones and desktops (see MacOS X). Now we have this great success which combines worst qualities of microkernel and monolithic kernel. One thing I would like is that when microkernels take over the world, which I fully expect, is that we at least get a footnote. Yeah, because it sure will be a pity to admit that we spent last 20 years on dead end. Being ahead of your time is never good. I published a paper in 1978 on something very close to the Java Virtual Machine, but we never got much credit for it although we were years ahead of Sun. I like this one the best: when AT tries to explain why his current ideas will take over he uses another, earlier idea, which failed to take the world over as well. Just like L4 in example above JVM was oversold and pushed around where it was not really needed till Sun just simply run out of money.
(Log in to post comments)
He focused on the wrong lawsuit Posted Nov 17, 2011 22:31 UTC (Thu) by JoeBuck (subscriber, #2330) [Link] It certainly is true that the lawsuit crippled BSD at a critical time, but the part of the suit that attacked BSDI over the phone number was inconsequential. What mattered was the lawsuit against the University of California over their release of BSD source code. At the time, the BSD code was a lot more mature and complete than Linux was, and had the free software community built on top of that instead of on top of Torvalds' kernel, the world might have turned out differently (but not terribly differently; we might still see a split between a pure BSD world and a GNU/BSD world with userland based on glibc and the GNU tools, which already existed in the critical 1993-4 timeframe).The lawsuit wasn't the only factor; Linus welcomed outside participation while the core BSD people spurned it. That might have been more important, but on the other hand a more bazaar-oriented team might have forked BSD instead. But the fact that Linux was clearly not derivative of Unix made it a legally safer choice for many.
He focused on the wrong lawsuit Posted Nov 17, 2011 23:37 UTC (Thu) by job (guest, #670) [Link]
BSD would still have forked, and Linus would still have realized the distributed development process that kept forks in line.
Most companies would still prefer to contribute to copyleft projects when it matters to avoid playing a losing game against competitors.
So my guess is that it wouldn't make much of a difference at all for Linux world domination...
He focused on the wrong lawsuit Posted Nov 18, 2011 15:17 UTC (Fri) by pboddie (subscriber, #50784) [Link]
There was still plenty of time for BSD to become dominant, but it just didn't happen, and although some people still insist that BSD is the platform of choice for companies (even namedropping organisations like Juniper Networks and IronPort), it's clear that a combination of licensing and collaborative factors have helped Linux to thrive where various BSDs have withered away from the opportunities.
And with regard to the timing, back in around 1994, I remember people explicitly choosing one of the BSDs - NetBSD, in fact - as the focus of a replacement operating system project for the Acorn Risc PC, which shipped with Acorn's proprietary microcomputer operating system, RISC OS, that many people were increasingly fed up with (Acorn having given up on UNIX). For reasons of maturity and portability, NetBSD was preferred over Linux, and sure enough the project did indeed deliver "RiscBSD". Meanwhile, Russell King managed to port Linux to ARM.
I'm sure there are some readers who are more familiar with this topic than I am, but although you can still get NetBSD for a range of ARM devices (including various Acorn machines), it's some variant of Linux that is dominant on ARM. Although the RiscBSD people tried to get the likes of Oracle to use their work (and IBM was apparently shipping BSD on top of Workplace OS), all that is ancient history now.
So, no, I don't buy the argument that a lawsuit ruined things for the BSDs and in a narrow window of time in the early 1990s Linux explosively filled the vacuum completely. There's a good part of a decade to account for, as well as the situation today.
He focused on the wrong lawsuit Posted Nov 18, 2011 20:46 UTC (Fri) by man_ls (subscriber, #15091) [Link]
Especially since in 1995 (after all the lawsuits) Linux was not invincible nor established. In fact NeXT used *BSD code in their kernel. The real battle started after all the lawsuits and the *BSDs lost, plain and simple, in every market. Funny how fictional history can be rewritten so easily.
He focused on the wrong lawsuit Posted Nov 19, 2011 15:27 UTC (Sat) by vonbrand (subscriber, #4458) [Link] The whole point is that if it hadn't be for the pre-95 lawsuits, there might have been no Linux contender afterwards. And I shudder thinking of what would have happened then with the BSDs fading away as they did, and something else taking over...
Linux vs BSD Posted Nov 19, 2011 15:41 UTC (Sat) by man_ls (subscriber, #15091) [Link] That is not Tanenbaum's argument: he saysI think Linux succeeded against BSD, which was a stable mature system at the time simply because BSDI got stuck in a lawsuit and was effectively stopped for several years.However the wikipedia article states that the lawsuit was settled in 1993. The BSDs might have overtaken GNU/Linux easily. Why would have Linux disappeared anyway? Linus started his kernel in 1991 before the lawsuit, and his inspiration was rather the Hurd than the BSDs. What fundamental change might have happened anyway, but for the lawsuits? It is not credible. In fact it's only now that I have read the interview itself, and it is true that the guy misses the point so widely that Microsoft would be envious.
Linux vs BSD Posted Nov 20, 2011 19:30 UTC (Sun) by Wol (guest, #4433) [Link]
I think you miss the point. Linux wouldn't have disappeared, it would never have appeared.
Linus couldn't get a cheap Unix-like (or even just a cheap) OS for his new 386. If BSD had been easily available (thanks to the lawsuit it wasn't) then he might never have started Linux.
Cheers,
Linux vs BSD Posted Nov 20, 2011 19:39 UTC (Sun) by mjg59 (subscriber, #23239) [Link]
USL vs BSDi wasn't filed until after the first release of Linux. The lawsuit may have delayed the development of BSD for home machines, but it's not the reason that Linus wrote his own kernel.
Linux vs BSD Posted Nov 20, 2011 20:23 UTC (Sun) by vonbrand (subscriber, #4458) [Link] Linus himself said in 1993 that if 386BSD had come out before (0.0 came out in 1992) Linux would never have happened. I did compare roughly contemporary MCC and 386BSD versions from the 1992-1993 timeframe, and 386BSD sure was a much more polished system. I was using 4.1 BSD during my PhD studies in 19984 to 1987, it wasn't exactly a long shot imagining porting most of that to a PC in the starting 1990s. But the lawsuits and concurrent assorted wild claims of infringement did make people somewhat nervous about ending up stranded with BSD, and so looked around for alternatives.
Counterpoint Posted Nov 20, 2011 20:26 UTC (Sun) by man_ls (subscriber, #15091) [Link] Yes. Notice that Linus in his seminal message spoke about the Hurd as the competition, not about BSD. And even that "big and professional" kernel did not stop him from starting his pet project. So your point is not too credible, that is my point.
Linux vs BSD Posted Nov 21, 2011 14:23 UTC (Mon) by spaetz (subscriber, #32870) [Link]
> Linus couldn't get a cheap Unix-like (or even just a cheap) OS for his new 386.
This motivation is far from certain. Linux started out as a terminal emulator and increasingly got extended until it was a full fledged kernel. It essentially was a research/hobby project, and it is far from certained that it wouldn't have happened if somehand had handed Linux a free Minix CD at that time.
Too much speculation :-)
Linux vs BSD Posted Nov 22, 2011 21:25 UTC (Tue) by JanC_ (guest, #34940) [Link]
Linus was actually using Minix while developing that terminal client & later the kernel now known as Linux... (The Minix filesystem was also the first Linux filesystem, as Linus needed to be able to read files from his development system's disk.)
He focused on the wrong lawsuit Posted Nov 18, 2011 15:58 UTC (Fri) by ccchips (guest, #3222) [Link]
This reminds me of the difference between Peter and Paul. Paul wanted everyone to be Christian, and Peter wanted the Jews to become Christian. When the knife came down from Rome, the more universal branch survived.
Heh Posted Nov 18, 2011 8:32 UTC (Fri) by oldtomas (guest, #72579) [Link]
What gets me most (actually makes me furious) is this constant reference to the JVM as the first "virtual machine" and "we were first 1978".
The idea is much older. There were working implementations in 1971 (not just one!)
<https://secure.wikimedia.org/wikipedia/en/wiki/P-Code_mac...>
<https://secure.wikimedia.org/wikipedia/en/wiki/Smalltalk>
The idea itself was kicking around since mid-sixties or even earlier.
I mean: Tannenbaum is an estremely smart guy and all. But I think he's being (academcally) dishonest.
Heh Posted Nov 18, 2011 9:48 UTC (Fri) by Wol (guest, #4433) [Link]
Add another one - Pick is a VM. Okay, it sat straight on the hardware (but didn't all of them?) and seeing as they controlled the hardware I think they micro-coded it so it was part-VM, part real cpu, but the design was 100% virtual.
And Pick (in its original incarnation of GIRLS) is older than Unix!
Cheers,
Heh Posted Nov 18, 2011 17:46 UTC (Fri) by andreasb (subscriber, #80258) [Link] And lest anyone think that virtual machines pior to Java were just academic exercises used only by a handful of students and professors: All of Infocom's text adventures ran on a virtual machine created in 1979 (the Z-Machine).
Heh Posted Nov 19, 2011 5:53 UTC (Sat) by JoeF (guest, #4486) [Link]
All the stuff you talk about is about right now. Of course, right now Linux has taken over the world.
But back in 1992, there was no USB, cell phones were big monstrous things. At that time, it certainly would have been possible for BSD to gain critical mass. The phone number stuff, while it was in the tech news back then, wasn't all that important. The AT&T v. University of California lawsuit was the important thing that held back BSD.
Have you read the interview? Posted Nov 19, 2011 9:26 UTC (Sat) by khim (subscriber, #9252) [Link] All the stuff you talk about is about right now. I talk about stuff which is relevant now because AT pushes MINIX as production system now. At that time, it certainly would have been possible for BSD to gain critical mass. May be, but this is just another monolithic kernel. But back in 1992, there was no USB, cell phones were big monstrous things. Sure. But 1992 already had lots and lots of operation systems with [macro]kernels and memory protection. Yet Tannebaum wrote this back then: Microkernels have won. The only real argument for monolithic systems was performance, and there is now enough evidence showing that microkernel systems can be just as fast as monolithic systems (e.g., Rick Rashid has published papers comparing Mach 3.0 to monolithic systems) that it is now all over but the shoutin`. AT (and many, many, many others) just can not accept the fact of life: software development is not engineering and it's not like math. Programs evolve like a breeds. There are both blog posts and articles on subject. Note that this is rare case where, as time goes on, the analogy becomes not less applicable, but more applicable. In selection we learned how to transplant genes from one unrelated species to another. At the same time in programming complexity and diversity grew so much that such transplantation often is quite problematic. When programs were measured in thousands of lines it was easy to just rewrite them for different platform even if languages differed - today we invent complicated tools and even then rejection is not uncommon. The underlying reason for why selection and programming is similar is trivial: the same dichotomy exist. If you want to have more of the same cells - you only need to add sugar to Petri dish and larger organisms can be replicated without much difficulty. And similarly with programs: you can duplicate them in billions for very small price. But to purposefully change them... that's different story. This is why I despise these "novel inventions" (like MINIX and/or Microsoft Synergy) so much: their creators ignore this important evolution process completely and totally. They live in the past when it was much less important (as this anecdote shows) and instead of trying to think about a way to plant their ideas in the existing system they preach the virtues of new system and try to perfect it. This does not work. Both in biology and computers the first "good enough" specimen which is brought to the new niche wins - be it rabbits in Australia or Android in smartphones. Timing is critical, perfection is not. When AT talks about embedded systems and says that We are going to port it to the ARM and do that starting in January it just shows such a lack of understanding that's not even funny. Linux is pushing all other systems in embedded space! Yesterday and today! Not tomorrow or 10 years after today! Your window of opportunity is shrinking - fast. Yet MINIX developers are complacent and talk about virtues of their system instead. Instead of showing working prototypes (which will be kinda "yes, we know we are slow to the party but there are still small hope") they sya that they will start porting in January? Gosh. Just what they are thinking?
Have you read the interview? Posted Nov 20, 2011 9:04 UTC (Sun) by JoeF (guest, #4486) [Link]
Of course, AT trying to push Minix now is a complete waste. I only commented on your post because you seemed to argue about AT's failure of getting Minix to the forefront back then by referring to what Linux has now.
If AT had pushed Minix in 1992 much more, he may have succeeded. Today, there is no question that it is an exercise in futility. I actually read Rick Rashid's microkernel papers back in the early 90ies. From an academically standpoint, they had a lot going for them. Academics don't have to care much about performance in the real world, though. I have encountered that again and again, e.g., with Ted Nelson's Xanadu system being a compelling system in theory, but failing at scaleability. There is a reason why http succeeded (thank you TBL and Roy Fielding) and Xanadu is a foot-note in history (having met Ted Nelson, his influence of course was very important, but Xanadu was never going to work.) Coming back to AT, he is an academic, and it is obvious that he doesn't understand the embedded systems space. I don't expect Minix to get any foothold in the embedded space. On the other hand, I am sure all the embedded vendors using Linux who "neglect" to provide the sources as required by the GPL (or who would like to restrict access, see the recent AVM case in Germany, covered by LWN here: https://lwn.net/Articles/466710/) would love to have an alternative as cheap (i.e., free) as Linux. There just isn't any...
|
|||||||||||||