Yamit is a microkernel, developed under a BSD compatible license. Mentioned capabilities are a thread-aware kernel and multiprocessor support (including both tight SMP and loose – [cc]NUMA architectures). The project is largely based on the Mach microkernel that was being designed for many years by CMU and OSF.However, Yamit is not a Mach-based distribution. Yamit is trying to address various problems that are familiar to developers who have experience developing Mach based operating systems like poor performance shown by earlier Mach-based systems such as the Lites. Other changes have to do with code clean up as the Mach sources have deteriorated over the years and experiements building large Mach based SMP computers met with limited success so far. Many of these predicaments could only be solved through complete re-design of concepts and code. These are the most notable differences between Yamit and other Mach-based efforts like xMach and GNU/Hurd.
That’s what I got (Opera 6.0 and Mozilla 1.0 in Linux):
You have probably reached this page in error.
If you were attempting to reach a site that’s domain ends
in .sourceforge.net, it is likely that your browser
is not HTTP/1.1 compliant. This may be because you’re running
a very old version of MSIE or Netscape.
Please upgrade your browser and try again.
Please Click Here
try http://yamit.sf.net
Somebody reinvented the square wheel — weeeeeeee!
There’s a reason why Lites began and ended as a college project. Those who don’t learn from history doom themselves to repeat it.
you really have a problem with people working on something that you feel is a waste of time. get over it already. if people want to work on a project, they don’t need to ask your permission. I know, and so does everyone else by now, that your not saying they can’t, that you just wish they wouldn’t, or whatever. really, you keep posting the same thign, with different wording, to every article. if you only want one os, only want one program for every need, go to a microsoft site!
p.s. – Eugenia, here’s an OSNews enhancement request: the ability to automagically hide posts from certain users, though, this would require a login
It would be WORTH signing up and logging in to avoid Speed’s whining. He/She is not adding any value and just creates a lot of noise. If you take a look at many of the recent topics, she/he posts just to get attention.
The more power you can give the users to get value from your site the better.
PLEASE ADD AN IGNORE OPTION.
for my own interests this is a cool idea… i will be watching this closely … will be interesting to see the differences between this and xmach
i wish the project well
as far as people’s comments… dont worry about them…
they are free to have their views as we are to have ours
and if they make theirselves look foolish enough we are free to laugh at them ;o)
Doesn’t seem to be moving very quickly (at all?). Only has had 8 developer mail list postings in all of 2002, and that includes an automated posting/month. Too bad, sounded interesting.
What an idea, the ability to not have to see a particular person’s posts, wonderful. I’d PAY to not have to read those damn useless posts of Speed, i mean get a life man! if you dont like news like that, find another news site to read cause believe it or not, even though its not Windows or Linux, its still important.
Following brought to you by rage:
Well, why don’t they work on releasing 4.6 stable!!! AH! They really need to release 4.6 stable, it’s been 4 days since the LATE release date! Thank you for your cooperation.
From what I understand, Hurd is dumping mach entirely for an L4 based microkernel (instead of being stuck with mach 2.5/3 as stated on their page). I’m not up to date with details on microkernel features, though…
http://os.inf.tu-dresden.de/L4/
By all means have an ignore option! It works both ways though. I’m sick of seeing ignorant nitwits wax orgasmic about ideas that are no different from the original ones that came out ten years ago. And I am every bit as entitled to my opinion as each of you are.
There’s a reason why Lites began and ended as a college project. Those who don’t learn from history doom themselves to repeat it.
The only thing Yamit redoes is Mach, and that is its goal. (And besides, the most used desktop UNIX is a Lite, nice college project, no?) That is OS X
We all know that OSX is not UNIX.
There is an ongoing effort to port The Hurd to L4,
that’s right. But so far there is just research going on.
Right now The Hurd is very dependend on Mach’s architecture,
and if you take a look at the Hurd/L4 mailing list, you’ll
see that many thing haven’t resolved yet (btw, there is
just little activity in this mailing list).
Said that, L4 is very likely (one of) the future kernel(s)
of The Hurd. But I really doubt version 1 of The Hurd will
use L4 instead of Mach.
Sorry rajan r,
obviously you misunderstood:
Speed was talking about “Lites” ( http://www.cs.hut.fi/~jvh/lites.html ), a 4.4 BSD Lite Server on top of Mach. It basically failed, since there was no really usable implementation of Lites released. It’s development stopped years ago. But at least other projects took benefit from their work.
One of the main goals of Yamit is to get BSD running on it, so it has something in common with Lites. But on the other hand, Yamit also tries to improve the Mach kernel and add good SMP support.
There is also xMach (based on the Lites code), basically continuing the Lites project. Unfortunately there is hardly any activity (not may developers, CVS server was down a long time, homepage is often not working).
greetings max
“We all know that OSX is not UNIX”
umm yeah that explains this big grey graphic on appple’s website
http://www.apple.com/macosx/technologies/darwin.html
ohh and you better read this one too
http://developer.apple.com/darwin/
umm so what is darwin if its not a unix?
Max, thanks for providing the info. I keep on forgetting that some of the most vocal posters have little to no knowledge of the subject matter!
Yes, of course I meant Lites. One can’t help but think of Lites when one reads the Yamit website. The parallels are palpable.
In case you haven’t followed previous threads, I’ll catch you up. I (and others) have noticed that the microkernel paradign has failed to take the world by storm, and that many companies who proudly boasted of microkernel-based products 10 years ago have quietly abandoned the philosophy. At the present, microkernels are of more academic interest than they are commercially viable.
Naturally this doesn’t sit very well with the Apple sheople, who cannot tolerate the slightest criticism of their religion-product. And so they wage their petty little holy wars…
Well Julian, your troll will be rewarded. My “predictable response” is, as always that allusions and weasel words are a far cry from facts. Apple is not the UNIX authority, the Open Group is. And the Open Group says that Darwin is not UNIX.
Similarly, that guy on the street corner selling “Rolex” watches for $25 is not selling the genuine product.
As for what Darwin is, it really doesn’t matter to this discussion, does it? You’re welcome to satisfy your own curiosity if you want. From what I’ve seen, it’s kind of like looking at what goes into sausage.
As says Ambrose Bierce, this is my opinion and bla bla bla…
This is an interesting project, and I am glad to see them around again, after having disappeared for some months from the web. Glad to see them going on, even at a slow pace.
>>Apple is not the UNIX authority, the Open Group is. And the Open Group says that Darwin is not UNIX.<<
Where do they say that on their website. You have already proved that Apple doesn’t have the UNIX trademark and we believe you, but it doesn’t erase the fact that OS X is UNIX (based)!
Speed got you on a technicality. UNIX is a trademark of the OpenGroup. Darwin/OsX may be Unix like, but can’t be UNIX unless they are certified by the OpenGroup. (Notice the case difference). Really truly, though, there is only one true Unix, the one originally developed by Bell Labs. All of the modern Unixes are “Unix like”, for they all can’t be the same, since, they are all different.
Moreover the certification is a time- and cost-expensive thing, and you have to certify each new version of you OS again …
Let’s just hang a sign around Speed’s neck that says “Please don’t feed the troll”.
It’s obvious this guy is a troll. I mean come on, this is the same guy who spent days trying to argue that no one is using Java.
…nobody is using Java? No, wait, that’s C#…
Yeah I am not really sweating the case anymore… as long as the it still works regardless of the trademark or not. Actually it seems more like a compliment if OS X is not actually true UNIX, but shares its likeness, which I like. It’s not the best implementation of UNIX, of course I think Solaris is probably one of the best, but OS X works for me and I can’t complain!
UNIX has become like the word “freedom”. a lot of people hear these two words (together if youre RMS) and think entirely different things.
some might say that unix is about the tools available to you and what you can do in shell, with the fs, etc… BUT a lot of linux desktops try and hide that, are they still UNIX? macosX very successfully hides and trace of a shell and the fs seems very similar to macos9. is it UNIX?
UNIX is such a nebulous set of concepts now that nobody can really point to one thing and say “that is UNIX and UNIX is that.”
Each story should have three links at the end, connecting to separate lists of comments:
(1) for discussion of licenses;
(2) for discussion of whether the subject of the story has been done before, or is worth doing now;
(3) for everything else.
Mailing list activity is hardly a sign of how fast is the project moving. Projects with many developers would have more posts. Project with two/three developers as in case of Yamit does not need to have heavy traffic. Developers would most likely communicate directly. I tried binaries and it seems to work. I have been communicating with the author or the arcticle and understand they decided to not use source control and keep updating sources only at [pre-]release milestones. If there were more developers it would make sense to spend time on keeping public appearance up-to-date though.
Seems pretty reasonable.
IS.
“Well Julian, your troll will be rewarded. My “predictable response” is, as always that allusions and weasel words are a far cry from facts. Apple is not the UNIX authority, the Open Group is. And the Open Group says that Darwin is not UNIX.”
Yeah…. And The OpenGroup also says that Linux is not UNIX. FreeBSD is not UNIX either. Neither is AIX, or Solaris. You are just being argumentative. Everyone here (with the possible exception of you) knows that UNIX is both a trademark for a particular product, as well as a generic name for a family of operating systems. to pacify your trollness, would you be more happy if we called Darwin “unix” or “Unix” instead of “UNIX”?
Simba: “Yeah…. And The OpenGroup also says that Linux is not UNIX. FreeBSD is not UNIX either. Neither is AIX, or Solaris.”
You are right that neither FreeBSD nor Linux are registered with the Open Group. AIX and Solaris are however, so they would qualify to use the UNIX (all caps) trademark maintained by the open group. OS X is not registered with the Open Group, and therefore can’t be classified as a UNIX system. I prefer “Unix-based”, but I’m not going to go over this discussion for a third time…
What is the difference between a kernel and a microkernel?
“You are right that neither FreeBSD nor Linux are registered with the Open Group. AIX and Solaris are however, so they would qualify to use the UNIX (all caps) trademark maintained by the open group.”
Are you sure Solaris is still registered? The only reason I think they might not be is that when Solaris 8 boots up, you don’t get the familiar “UNIX System V Release IV” message. However, on Solaris 7, you did get that message. I thought maybe it had something to do with Solaris going open source or something.
Take that subject however you want.
Speed, you claim that everyone commercial company taht has touted microkernels has abondoned it. I dont recall any change with QNX that they dropped the microkernel design
And just because monolithic kernels are the norm doesnt mean they are superior in every instance. Everytool has a use. Like using a language for a job, using a kernel design for a job is also important.
If i remember right, I think its Neal is heading over to work on L4/Hurd this summer. I am curious to see how far aong it will get and to see what kind of improvements it can offer.
I went to the Open Group site, and there was a partial listing of registered products. In that list were the following:
* NCR: Solaris 7 and on, on NCR World Mark and ‘S’ series (X86 based systems)
* SUN: Solaris 8 and on, on 32-bit and 64-bit SPARC based systems
* SUN: Solaris 8 and on, on X86 based systems
* SUN: Solaris 7 and on, on 32-bit and 64-bit SPARC based systems
* SUN: Solaris 7 and on, on X86 based systems
I guess Solaris 9 hasn’t been registered yet, but I can’t imagine them dropping the registration at this point. Interesting note on the popup display though 🙂
Speed got you on a technicality. UNIX is a trademark of the OpenGroup. Darwin/OsX may be Unix like, but can’t be UNIX unless they are certified by the OpenGroup. (Notice the case difference). Really truly, though, there is only one true Unix, the one originally developed by Bell Labs. All of the modern Unixes are “Unix like”, for they all can’t be the same, since, they are all different.
It’s hardly a technicality!
Sorry arougthopher, but playing around with case is also a violation of the trademark. This is common sense! Would you plunk down a million bucks for a “Ibm” mainframe? Of course not! You know that you’re not going to get the famous IBM quality and support. It’s no different with UNIX.
AT&T sold the UNIX name to Unix Systems Labs, who later became part of the Open Group. So you’re right that there’s only one true UNIX. You’re just a decade or so behind the news about who has that one true UNIX name.
Another important thing to know is that UNIX is far more than a trademark. To be worthy of the UNIX brand, a candidate must pass a battery of tests to ensure that it meets or exceeds many different standards for operation and interoperation. UNIX systems are trusted to handle massive amounts of money, and data that is literally “life and death”. The strict testing ensures that you don’t lose your shirt, or lose your life!
If you’re looking for a word to represent systems that resemble UNIX in “look and feel” the accepted term is “UN*X”.
There’s nothing mystical or vague about it. Go to the Open Group’s website, and read to your heart’s content. It’s all there.
The problem comes when ignorant people pretend to be authority figures when they aren’t, or when shady businesses commit fraud. For the former, you can read an excellent paper:
http://www.vmyths.com/fas/fas1.cfm
So now you have what you need to find out exactly what UNIX is. No excuses any more, k?
This is a good time to remind everybody that Simba is a liar. Simba’s earlier attempt at character assassination is not only a lie, but fails to disclose the previous lie of Simba’s. Simba made a claim about Java, couldn’t back it up, and became a sore loser.
Now let’s look at the latest lie:
“The OpenGroup also says that Linux is not UNIX. FreeBSD is not UNIX either. Neither is AIX, or Solaris.”
Well, nobody here is claiming that Linux or FreeBSD is UNIX, so the first part is a diversionary tactic. But the last part is a lie. Both AIX and Solaris systems are registered:
http://www.opengroup.org/regproducts/xx.htm
It’s a topic that has been discussed at great length before. Simba has been informed, and chooses to say what is untrue, a lie.
Anonymous, the microkernel concept is the subject of teaching at the college level, so don’t expect to get a full understanding of it here. But to strip it down to basics, the idea is to split up the traditional kernel, having a microkernel operating in ring 0, priveleged, supervisor bit etc. mode. The rest of the kernel is put into server processes that operate in user mode, typically ring 3. The microkernel’s main job is to pass communications between the hardware and the servers. So the main difference is that what traditionally ran as priveleged kernel code now runs as unpriveleged user code.
Remember that this is a very simplistic explanation, and I didn’t even get into the philosophies about why they tried to do this. But if you do any systems programming, the liabilities of the microkernel design are evident.
Gee whiz the vast majority of postings here are SUPPOSED to be about Yamit. Up earlier I posted something about L4. There’s only like 4 posts of all of these that are actually on topic.
Anyways, here’s what i know about micro/monolithic kernel stuff.
It does make sense to break up a system into pieces. OS classes are taught that way, to examine each piece in turn. Microkernels came from trying to make clean breaks between system components by creating different server processes, each process to handle some primitive part of the system. This is supposed to make the system scale nicely to multi cpu systems, and also make it easy to unload/load drivers on the fly, just by stopping a service, reinitializing it and restarting with out touching the rest of the system.
(Btw, Linux dealt with the latter problem when it introduced loadable modules, and in my opinion at the time was a huge victory over the claims of the microkernel people)
While this works nicely on paper, trying to get all the pieces working together isn’t nice. The biggest problem traditionally has been the penalty of both context switching overhead for servers (system with few cpus) and communication between servers.
Apparently L4 claims to have dealt with the IPC issues in the kernel, and I think this may yet again heat up the monolithic/microkernel wars.
Just my uneducated opinions
Brian
But if you do any systems programming, the liabilities of the microkernel design are evident.
really? could we have some evidence, anecdotal or otherwise, of your claims?
This has been covered ad nausaeum in other threads, but the jist of it is that moving OS services out of ring 0 kernel space means context switches. Context switches are expensive CPU-wise, and the microkernel design means that they must be done for just about everything.
If you want evidence, compare the performance of Windows NT 3.1 vs. 3.5 and 4.0, as more and more functions were moved back into ring 0. I hear that it’s a similar story for BeOS, moving from the original BeBox version forward. But I’m far from a BeOS expert, so I defer to anyone with better information. Another useful comparison can be made between MkLinux and LinuxPPC, two Linux distributions, one microkernel-based and the other monolithic, that were concurrent for a while. All things equal, the performance difference between the two was dramatic.
Brian, I still have yet to read a reasonable explanation about why microkernels were supposed to facilitate multiprocessing. Since your post is far more literate than most on the subject, I put it to you — any help?
“””moving OS services out of ring 0 kernel space”””
Layer 0 is always hardware; also, you can’t quite use the layered design approach as a metaphor.
“””…means context switches”””
It’s not the context switches that hurt the most, it’s switching Heavy Weight Processes (HWP) and paging.
“””I still have yet to read a reasonable explanation about why microkernels were supposed to facilitate multiprocessing.”””
Simply put, the system is already completely preemptably multithreaded (including the drivers). The only thing that isn’t particularly pre-emptible is actual hardware access (inside the Mk). The more complex implications are that since IPC is handled for you, each process can be migrated seamless between nodes (be they different instances of an OS on a big NUMA machine, or a cluster, or a heterogenous network ala DOS [distributed os]).
Layer 0 is always hardware
OK, we’ll stop right there. You’re done.
If you don’t even know what ring 0 means, you’re not going to be able to understand any of this, much less make relevant comments.
If you want to learn about computer architechure and operating systems, by all means go to school and study the same. But please don’t waste my time by trying to bluff your way through this.
Some people say that microkernels are slower, but is this true?
Let’s look at facts
1. It’s very hard to write a portable and efficient multiprocessor kernel.
2. Microkernel approach makes it way easier because boundaries are enoforced
3. People have mentioned context switches, but what is the problem there? The only answer I know is the expence of a TLB flush. However, most architectures don’t require that. Even x86 in segmentation mode does not need to flush.
4. Mono design is nice for a small UP comp, but Linux and BSD have been around for years, In fact Linux has biggest developers base and probably more money than any other project (I am not sure, but Windows could probably compete on funds level, but the number of developers). Still, Linux poeple have not yet (probably never) turned it into a decent OS for MP hardware.
Now what make your own conclusion.
Personally I think Speed just likes to argue about stuff he has only little knowledge.
Isaac Stern, let’s look at your post:
1. You made a claim, but never followed up on it.
2. You presented what you called “facts” that were only unsubstantiated opinions.
3. You made a demeaning personal attack.
Personally I think you “just likes to argue about stuff he has only little knowledge”!
I provided evidence above about microkernels’ sloth. But you have nothing to say about that! Hmmm, I wonder why? I see you taking potshots at Linux (although Linux works just fine on MP systems — I’ve been doing it for years), but nary a mention of all the other monolithic MP operating systems, ranging from various UNIX flavors to Windows to OS/2. And the “segmentation mode” reference was truly laughable! LOL What modern OS runs in real mode!
Isaac, I don’t know what you hoped to accomplish by being such a jagoff, but you certainly managed to prove yourself a fool and a jerk.
“This is a good time to remind everybody that Simba is a liar. Simba’s earlier attempt at character assassination is not only a lie, but fails to disclose the previous lie of Simba’s. Simba made a claim about Java, couldn’t back it up, and became a sore loser.”
It’s mot a lie. Anyone who wants to can go read the posts in the thread. You tried to claim no one is using Java. And you are saying I didn’t back up my claim that lots of people were using Java.
By the way… You have ignored my question 4 times now. What do you want me to give you to back it up? Here is what I have given you so far:
1. An Evans Data Corp survey that says over 50% of US programmers are using Java and over 60% of Europian programmers are using Java.
2. An survey that says 78% of IT executives polled stated that J2EE does or will play an important role in their network business strategy.
3. A list of 49 major companies that are using Java in major way–many of the companies on the said list being fortune 1000 companies.
I ask you again: What more do you want? I think these three points do a pretty good job of backing up my claim that Java is being widely deployed.
“(Btw, Linux dealt with the latter problem when it introduced loadable modules, and in my opinion at the time was a huge victory over the claims of the microkernel people)”
Linux was hardly the first operating system to introduce a moduler kernel design like this. SunOS had already been doing this, and I would say that to this day, Sun still does it better than Linux.
But this is not really the benefit of the microkernel anyway. The benefit is, as has been pointed out, that most of it can run as a user process. This allows a great deal more flexibility in de-centralized systems and virtual machines than the monolithic kernel does.
Btw Speed… Why don’t you explain to everyone here why QNX and vxWorks are both using microkernels if they are such an inferior design? As I have pointed out numerous times, vxWorks is only the most popular RTOS in the world.
Simba, you have gone so far afield in misquoting me and burying the issue in irrelevant statistics that I think you forgot what my original point was. Anybody who read that thread knows that I didn’t just waltz in and say “no one is using Java” as you claim.
You certainly don’t have the guts to take on what I actually did say, I notice.
You’re just pissed off because you make poor decisions, and you’re taking it out on me because my comments remind you of those poor decisions. That’s not my fault. I’m not going to be a scapegoat, so as long as you tell unpleasant lies about me, I’ll keep on telling unpleasant truths about you.
If you want to know why QNX and vxWorks are how they are, I suggest that you ask their respective vendors. Any fool can tell that I don’t represent those companies. (What kind of sub-fool that makes you, I don’t know.) It is enough that I showed how the design is inferior. And for all we know, both companies are abandoning the microkernel as I write! Considering your record of dishonesty, I wouldn’t be surprised if they’re already microkernel-free.
It’s bad enough that you’re a liar. Being as lazy as you are makes you a truly pathetic individual. After all, you are the one who keeps on bringing those products up. But I don’t see you producing any evidence that either of the two outperform other monolithic products! Why is that? What is it that you’re hiding this time?
Speed: Others have already pointed out that your single article on microkernels from Linux Journal was a load of crap. So I am not even going to go there.
But as far as the Java thing, and you calling me a liar, your personal attacks are starting to cross the line.
And to prove I am not lying…
After I said that Java more popular than C, Speed responded as such:
“How many people run Java apps outside of a browser? How many run Java apps inside a browser?!? ‘Nuff said… ”
To which I responded that Speed was not looking in the right place for people using Java and that most Java use was internal to corporations. Speed then made the following post:
“As for Java, I notice that you failed to name a single “major corportation [sic]” in support of your claim. No matter, I think you know that no substantial business is going to let me audit their software, even if I was foolish enough to do your legwork for you. So it’s just another empty claim.”
I obliged speed by proving him with a list of 49 major companies that are using Java which seems to be what he was attacking me for not doing in this post. He has since claimed that I gave him irrelevant statistics. This after I gave him what he attacked me for not giving him originally.
Speed then digs himself a deeper hole with the following posts:
“That’s not my problem. I do know that I can don a blindfold and throw a dart, and I’ll hit a C/C++ application. But Java apps are as rare as hen’s teeth! So obviously Java hasn’t taken over the world, just as I said.”
“If Java really is eating C’s lunch, then you’d be able to produce at least one or two for crying out loud!”
“But since you brought it up, I’ll point out that the man on the street would recognize the Oracle name, and would not be able to name a single Java app. So much for your faulty logic!”
This all after I pointed out to him that he was looking in the wrong place for Java apps. He continued to use his faultly logic (which he accused me of using) based on the idea that you can’t go to your local store and buy many Java applications.
Speed: Perhaps your problem is that you don’t have any real world experience in programming? Well let me give you a hint: The vast majortity of software out there are internal applications written for a specific company performing a specific task. It far exceeds the number of boxed software applications. Most programmers don’t work for software companies. They work for insurance companies and banks and airlines, etc.
I am not lying at all. And I am not being lazy. I did your research for you and gave you a list of 49 companies that were using Java and many of those companies are on the Fortune 1000 list.
Gee Simba, you can tell a half-story and quote me out of context all you want. But the part that’s conspicuously absent is any quote that has me saying what you claimed!
Now enough of your selfishness. The very least you could do is to say something that’s on-topic. Your original whine-fest ironically came out of a microkernel discussion. What’s with you about that?
Quit lying. Stay on-topic.
“Gee Simba, you can tell a half-story and quote me out of context all you want. But the part that’s conspicuously absent is any quote that has me saying what you claimed!”
If this quote exists, I didn’t find it and I must have missed it the first time around. I’m not saying it doesn’t exist. Only that I didn’t see it. So if it exists, could you let me know where it is?
“Your original whine-fest ironically came out of a microkernel discussion. What’s with you about that?”
I was only using Java as an example that sometimes speed and efficiency are not always the most important consideration when designing a system.
Java has disadvantages. It is relatively slow and relatively inefficient. Some of the “features” it touts as being the messiah of the programmer are quite expensive to implement. (For example, sure its great to not have to deal with pointers or worry about memory management. But on the other hand, calling every object by reference and keeping track of object references is expensive).
I’m not claiming Java can replace C or C++. It can’t. For all of the problems caused by memory management (mis-management?) and misuse of pointers in C and C++, C and C++ will always be faster than Java because of the fact that C and C++ doesn’t keep track of references to objects, and do automatic garbage collection, etc.
However, sometimes rapid deployment and cross-platform capability are more important than the loss of performance. So Java has advantages as well.
The only reason I brought up Java was as a comparision. Microkernels DO have advantages in some applications. And sometimes those advantages outweigh the disadvantages that you pointed out.
Want rapid deployment and cross-platform capability? Use Java. Need speed and performance? Use C or C++.
The same decisions are made with microkernels vs. monolithic kernels.
>Isaac Stern, let’s look at your post:
>
>1. You made a claim, but never followed up on it.
I am following up now. Since you seem to desire that.
>2. You presented what you called “facts” that were only >unsubstantiated opinions.
Let’s review the my statement:
|1. It’s very hard to write a portable and efficient |multiprocessor kernel.
If rest my “opinion” about Linux is a fact, then Above statement is a fact as well. Otherwise, why could not Linux been fixed yet.
|2. Microkernel approach makes it way easier because |boundaries are enoforced
That is a fact, because that is the follows from very definition of microkernel.
|3. People have mentioned context switches, but what is the |problem there? The only answer I know is the expence of a |TLB flush. However, most architectures don’t require that. |Even x86 in segmentation mode does not need to flush.
I will get back to it a little further.
|4. Mono design is nice for a small UP comp, but Linux and |BSD have been around for years, In fact Linux has biggest |developers base and probably more money than any other |project (I am not sure, but Windows could probably compete |on funds level, but the number of developers). Still, |Linux poeple have not yet (probably never) turned it into |a decent OS for MP hardware.
Linux has more advanced than FreeBSD SMP (Net/Open I undertsand don’t support SMP) has no concept of a kernel thread, but only process. Linux threads are simply processes that share address space. Even though at least until very recently is was imposible to schedule threads on separate CPUs. That’s why all decently perfoming programms use non-blocking pattern instead of threads.
Either way benchmarks for multithreads apps will show who knows his beef.
>3. You made a demeaning personal attack.
That is not correct + look below.
>Personally I think you “just likes to argue about stuff he >has only little knowledge”!
This is a fact, look below why that is so.
>I provided evidence above about microkernels’ sloth. But >you have nothing to say about that! Hmmm, I wonder why?
Well what is your evidence? It was mentioned that QNX is very fast. Thus your evidence is worth as much as everything else you say – zero.
>I see you taking potshots at Linux (although Linux works >just fine on MP systems — I’ve been doing it for years), >but nary a mention of all the other monolithic MP >operating systems, ranging from various UNIX flavors to >Windows to OS/2.
Windows is not really a monolithic system. At least it ws not design as such. On contrary MS hired Rashid who was the Mach designer. Although Windows was optimized and colocated it’s has concept of API subsystem and console subsystem calls still fire off IPC just like in microkernel.
>And the “segmentation mode” reference was truly laughable!
>LOL What modern OS runs in real mode!
You should read IA32 manual, it’s being freely distributed by Intel. Segmentation is a memory model and has nothing to do with real mode. And yes your comment are laughable.
>Isaac, I don’t know what you hoped to accomplish by being >such a jagoff, but you certainly managed to prove yourself >a fool and a jerk.
This is about “a demeaning personal attack.”
I have stated only facts this far. One of them is that you are INCOMPETENT. Your posts prove it one after the other.
I have not used faul language, you have.
Let’s other decide who is the …
Maybe you should select a new nick, old one seems to have very little credibility among readers, myself included.
However, I was not the one (actually more than one) who asked to just hide all posts from you.
Get some books.
Sorry about my English it is not a native language for me, I may have not been correct grammatically, but seems that I did hit the bullseye.
Christ.
Simba, I think that your admission that you don’t even know what you’re attacking me over says it all. You have proved that all you’re doing is being an asshole. Cease and desist now! For the millionth time, the topic is not Java!
Quit lying. Stay on-topic.
Isaac Stern, you’re another little brat who needs to grow up. Your last post is incomprehensible. If you’re going to be so antisocial, then the only reply that you deserve is a reminder that you’re being an asshole. While you certainly do not deserve such notification, I was polite enough to inform you. If you cannot participate in society, then I suggest that you leave.
Posting incorrect facts isn’t trolling. Otherwise even Eugenia could be a troll.
And as for OS X and UNIX, Apple never claimed it is UNIX, just claim it is UNIX based. Which is true.
And as for Java, it is far from “nobody uses it”.
Mac OS X could in fact qualify for the UNIX; but to do so, it must use stuff like X, Motif and CDE, and a host of our stuff which would make OS X as hard as… UNIX. So, OS X is taking the best of both worlds, the stablity of UNIX/BSD and the slooooow graphics engine of NEXT and the ease of use of Apple.
And as for OS X and UNIX, Apple never claimed it is UNIX, just claim it is UNIX based. Which is true.
No, it’s not true — you know that very well! I’ve said it many times, OSX is not UNIX, “based” or otherwise. Weasel words don’t alter anything. OSX contains no UNIX code, and has passed no UNIX certification tests. Therefore, OSX is not UNIX, period. No ifs, ands, buts or baseds.
Mac OS X could in fact qualify for the UNIX;
Coulda, woulda, shoulda. That’s the difference between the men and the boys, son. The men actually do what the boys only talk about. Come on back when you have more than peachfuzz to show.
but to do so, it must use stuff like X, Motif and CDE,
That’s not true. The UNIX 95, UNIX 98 and UNIX 98 Server specifications have no such requirements. Your excuse is made-up. And now that you have no excuse, what are you going to do?
and a host of our stuff which would make OS X as hard as… UNIX.
So you say. I see no evidence of your claim. While it may be hard for a product to become UNIX, blaming a product for being “hard” is nothing more than a cowardly way to say that you’re lazy.
So, OS X is taking the best of both worlds, the stablity of UNIX/BSD and the slooooow graphics engine of NEXT and the ease of use of Apple.
The problem is that OSX has neither UNIX nor BSD to brag about. Darwin borrows heavily from FreeBSD, which is a totally different thing. I’ve told you before that playing games of semantics doesn’t alter what a thing is. You will not find a penis on a woman just because “woman” contains “man” and a man has a penis. So your convoluted argument of compound faulty logic fails miserably.
And while FreeBSD is stable, OSX doesn’t use the FreeBSD kernel! Instead OSX uses some cobbled-together Mach kludge. Who knows if that’s as stable as FreeBSD? You certainly can’t ride FreeBSD’s coattails, because OSX is based on something that’s very different from what FreeBSD brings. Without any proof of alleged stability, you have nothing to brag about.
So in reality, all you’re left with is “the slooooow graphics engine of NEXT”. Nothing there to brag about, if you ask me.
Wouldn’t it be easier for everyone to just read the three other message boards where this discussion has been hashed out in identical fashion already?
“Simba, I think that your admission that you don’t even know what you’re attacking me over says it all.”
I know exactly what I am attacking you over. And the fact that you claimed I made a glaring omissions but when I asked you to point out what I omitted, you failed to do so and only responded with more vulger language and name calling says a great deal about your level of maturity.
Of course, recent posts in more recent threads prove I am not the only victim of your vulger and immature name calling and personal attack.
Grow up, and stop trolling.
Quit lying. Everybody knows that you’re a liar.
“Quit lying. Everybody knows that you’re a liar.”
If I am lying, point out my omission. This is the third time I have asked you to do so, and it is the third time you have failed to cough up anything to back up your claim that I made an omission.
And now you accuse me of lying again? LOL Would you like me to go through the recent threads and post all of the vulgar comments and name calling you have directed at others besides me?