Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
Linux Software

MkLinux Not Dead 51

Posted by justin++ from the native-parisc-port-soon! dept.
Ethelred Unraed writes "At Linux am Mac, there is an article written by Larry Kollar, an MkLinux developer, as a rebuttal to the erroneous tidbit at Mac OS Rumors claiming that MkLinux was dead (as Apple had ceased active support). Kollar's article seeks to explain just what IS going on, and where MkLinux is headed. " It's interesting to note that Apple isn't the only one to stop development on MkLinux - as we reported before, The Open Group also has ceased development. The article explains that the user community has been maintaining the port since then.
This discussion has been archived. No new comments can be posted.

MkLinux Not Dead More

MkLinux Not Dead

Comments Filter:
  • There's a Preforma 6116CD here at work that we put mk on, but it's very sad. I'm such a mach-kernel bigot even if it ran 1000 times faster than anything else I'd still have to complain. I wish I I could get sbus support some other way though.
  • I think you hit upon it right there. It's a curious design. It's a neat idea and all, but I haven't ever seen it put to any practical use. I don't need a Mach kernel on my mac all it does it add one more level of complexity to loading a single kernel.
  • MkLinux was wonderful when it initially came out, and it gave my 6100 another good year of life. I even ran MkLinux on my G3/300 for a little while... but as soon as I switched over to monolithic LinuxPPC and saw the perceived speed of my system increase by a significant amount, I never looked back.

    Now that LinuxPPC is out, fast, stable, and up-to-date with the Intel world, why is the Mac still the base development platform for MkLinux? Why not develop it on a system which won't run Linux yet? For that matter, if porting it to a new system is just a matter of porting Mach, why don't we see MkLinux running on a whole lot of other architectures these days?

    Another thing I'm curious about: why isn't Linus Torvalds supporting MkLinux more strongly? I know he's a proponent of portable implementations of Linux, so it seems odd to me that he seems so focused on the Intel port of monolithic Linux.

  • Apple sure as hell thinks it has a practical use. they wont let go of the damn thing and move on (OS X). nuff said.
  • OS X will basically only run on Apple G3 hardware which is only a small part of the overall PPC hardware that is there already. It is in no way POSIX compliant - even apple have said that it will not be made to be so, therefore you can't just use the source luke.

    Reading this, I have to wonder if you have any idea just what POSIX is. All evidence points to no. Therefore, a little explanation is in order:

    POSIX is a standard for operating systems. It has absolutely nothing to do with hardware, unlike what you seem to be implying. When it was written, it was based off of Unix. To be considered "compliant" it must pass a rigid set of hideously expensive tests and get certified and all of that good stuff.

    You'll be interested to know that Linux isn't POSIX-compliant either. It "aims toward" compliance. Guess what: so does OSX. Apple has said they're not going to bother getting it certified, that's true. Last I checked, neither was the Linux crowd.

    What does that leave us with? Something not unlike Mesa. You can't call Mesa OpenGL, since it's never been officially certified, but let's face it: for all intents and purposes, it's OpenGL. Likewise with Linux and OSX: You can't call them POSIX-compliant, because they were never certified as such and likely never will be, but both are striving to (and eventually will) achieve that goal.

    As for that "use the source Luke" bit, I'm not sure exactly where that came from. I sincerely hope you don't mean that it doesn't qualify as Open-Source because it's not POSIX-compliant (and the rest of the way your article reads, this seems highly likely). POSIX has nothing to do with the openness (or lack thereof) of one's source. If you mean compiling POSIX programs out of the box, prepare for another revelation: it's about as easy to do with OSX/Darwin as it is with Linux; I'll be happy to point you to more than a few programs which have already been ported over.

    This means that OS X is set up as a commercial venture .... you have to pay for the software you want to run.

    No one ever pretended otherwise. Not only that, but in case you haven't noticed, Open-Source software can and does exist for proprietary platforms. Also, for the last time, Open-Source does not necessarily mean zero-cost.

    Sure apache compiles out of the box ... now ... only after the apache group ported it accross.

    Actually, last I heard it was Apple who did that port, not the Apache group.

    In my experience it looks good runs well but it is not an open-source solution.

    Why not? Looks pretty Open-Source to me. It seems you're still a little fuzzy on just what Open-Source is.

    The other point about OS X is that it is based on the mach kernel so it inherently has the same performance issues as MK.

    I hate to say it, but you're wrong. I've used both Darwin and the OSX Developer Preview. I see no horrendous speed issues such as you describe. In fact, I've found it quite fast.

    OK, perhaps I'm just misinterpreting what you've written. It doesn't look that way, though. If I have, I apologize.

    Also, the standard disclaimer when I go to rebut a post: This is not intended as a flame or as flamebait; please do not take it as such.

  • I'd love to see a real linux port designed for NuBus Macs, but I am certainly not holding my breath.

    Could it be that they don't want to pave the way for a direct comparison of Mac hardware against x86-based hardware? Could it be that they're worried about MacOS market share?

    It could be those things, but I think it's much more likely that Apple generally just wants to ignore all the NuBus Power Macs. Just like I increasingly ignore mine. I guess they think that if you're the type who'd buy a $3500 7100/66 back in 1995, you're the type who'll buy a fancy new G3 today. (btw, I don't have that model and didn't pay nearly that much, thank goodness.)

    Poor x100 series owners; we can't install a fast, up-to-date, actively supported Linux on our machines to breathe new life into them like the 486 crowd can.

    Oh well, guess it'll just stay turned off...

  • Their are some people on LinuxPPC working on that. Right now the latest BootX booter supports Nubus Macs, but the kernel still panics on load, do to the kernel doesn't yet know about the Nubus-based archutechure in the early pre-1995 PowerMacs.

    Native Linux support is underway, some drivers have been written, but still not enough is done to make it run natively.

    Forently, MkLinux and Monolithic Kernels are pretty compatible with each other; it's not a problem to change from MkLinux to LinuxPPC or vise versa. Hell, I even have both MkLinux and LinuxPPC installed on my machine, via. BootX. Works great.

    That saying, MkLinux is going to become glibc 2.1 pretty some, with a new version of distro being released as soon as the ftp server gets it new hard drive.
  • Yes Linux am Mac is a pretty nice site, I have been reading it since like Febuary, when it was starting to become somewhat English.

    It does have a occational mistake on that, but thats not too bad, since nobody is perfect (execpt the nobody dude on my system)

    Keep up the great work!.
  • I think you may have missed the last 5 years of Linux history or have not read some of his stuff.

    Basically, he values anything that he considers to be true Linux, ported directly from the Linux source, such as Linux/PPC, Linux/Alpha kernels, etc.

    Read about the story of Monolithic Linux/PPC at http://kernel.crashing.org [crashing.org]
  • MkLinux has been a very touchy subject for Linus and Linux and general. In many ways, MkLinux can be considered a fork off of the orginal code (since it is running on a mach kernel and has many changes).

    In February of 1996 Apple and OSF publically announced the MkLinux project at the Free Software Foundation converence held in Boston.

    Linus looked down at MkLinux and declared it *NOT* an offical version of Linux. In the months following, he eventualy starts to back down from that, but it's unlikely he will ever support MkLinux directly.

    However, MkLinux was the reason why Linux on the PowerPC became successful, MkLinux created the first real Linux base on the PowerPC platform. MkLinux ran on cheap Nubus PowerPC hardware, compared to more expensive PCI PowerPC hardware, which often was purchased for new, at a expensive price.

    These are some of the highlights of the development, the full story with the correct address is at:

    http://gate.crashing.org/doc/ppc/doc003.htm

    (Yes it is a .htm extention, even though it's on a Apache box running LinuxPPC R5).
  • Yes, people try to remember that MkLinux DR 3 is oldish distro, being almost a year old.

    Much has changed in a year. A year ago libc5 was standard, KDE 1.0 was a hot new technology (first nicely intergrated desktop, that was easy to use), and the GIMP had just recently turned one. The standard stable Linux kernel of the time was 2.0.32 or something. GNOME was just barely to the .20 released and crashed all the time.

    Much has happened in a year. But as Nester pointed out MkLinux Genric 7 Kernel works pretty good with updated Linux on the PowerPC distros, such LinuxPPC R5, with some little changes (and it's pretty well documented.

    Don't fret, MkLinux Community team is working on releasing a update to the standard MkLinux distro (basically DR 4), including MkLinux Generic 7, glibc 2.1, KDE 1.1, GNOME 1.0, and the lastest of many other great packages.

    MkLinux isn't dead.
  • People will continue to use MkLinux for years, even though something better might come out for the Nubus system, since they already have it installed.

    Since it's open source, development will continue, even if it's only at a slow pace for some things.

    Heck, somebody might in 30 years boot up a copy of MkLinux and be amazed how simple and how powerful it was, even on today's awful slow, low technology stuff.

    If the technology is the best for the machine, people will continue to use it with out any development at all (my Mac Plus with 4 megs of RAM and System 6.0.8, hasn't been updated in 5 years with new software, yet it works perfectly for day to day work!)
  • " I'd love to see a real linux port designed for NuBus Macs, but I am certainly not holding my breath. "

    You will see one of these in the future, since a new version of MkLinux is almost ready for release, and that at least a few of LinuxPPC developers are working on Nubus Macintosh support.

    " Could it be that they're worried about MacOS market share? "

    "Apple doesn't give a flying rat's-you-know-what for what happens to old NuBus Macs."

    That's very true (at least $$$ wise). Software is only a small part of their revenue (like 1% the last time I checked). Almost any profit made from Apple selling software is redirected back to paying software developers. Apple develops most of it programs, just to encourage people to buy Macintosh hardware, and through hardware sales make money (example A: Quicktime). Heck, they used to give away the Mac OS for free, until 1993, and after then, gave away old versions for free (example B: Window 95 costs you now still like $85, while Mac OS 7.5.5 costs nothing, it's a free download. Both released about the same time).

    "OS X won't run on them anyway, so why should they care?"

    Some people at Apple are actually happy to see their Macs continueing to run (some developers at Apple wanted Mac OS X to run on more then 5 different machines).

    "The person who wants to run Linux on a NuBus Mac is the same kind that can't afford a monster G3 with OS X anyway."

    Yes that is true in most cases. Anyways, Mac OS X is just a way to make people buy new macs to run this OS, if you don't have the money to do this, you can continue to run Mac OS 8.7/9.0 or a Linux on your PowerMac with out problems.


    " I don't know the answers to these questions, but I'm sure that Apple stockholders would be strongly against the idea of diluting the already weakened
    profitmaking potential of Apple products. "

    " *sigh* Sounds like flat-out Mac-bashing for little reason (yet again). I'm not going to try and claim that any Mac can flatten any Intel machine, because that's not the point. The point is, I already *have* that hardware and want to get the best out of it. That's the glory of Linux--I can install it on just about anything, and use it for just about anything. You can't do a hell of a lot with an old Pentium without Linux (or *BSD, for that matter), either. With it, you can do plenty.

    That's not true. Windows 3.1 works pretty good on many 486s and Pentium machines out their, which can do quite a bit. Even Windows NT 3.5.1 (assuming you have the RAM), works okay on those machines.

    Of course if you want to make a good webserver out of those machines, Linux (or *BSD) is the way to go.

    Just because it wasn't released yesterday, doesn't mean it's useless. damn computer people ;-P

  • Folks!

    Whenever some journalist complains that linux lacks commercial backing, he gets major flames for being clueless. Even the last AC here states it as the most obvious truth that Open Source projects don't depend on corporate backing, in fact, projects live as long as anybody cares about them.

    And suddenly, as soon as an Open Source project gets dumped by a company, everybody cries that this is the end of the project. I have some bad news for all these pundits: Nobody can predict the future in computing. A guy called Bob Metcalfe had to eat his predictions once, and might have to eat his predictions again in a few years (except that this time he was more careful and didn't guarantee for his rants.)

    I don't know whether MkLinux will be around five years from now, but I also don't know, whether Apple will be around either (please, no stupid flame wars here)! Nobody knows!

    At this point in time I'm happy that MkLinux exists, since I have a PB5300 to write my thesis on, and I wouldn't trust the more common office packages which run under MacOS. In addition to that I didn't have to pay Windows tax (how many got their money back after the "Great Windows Refund Day" ?). LinuxPPC doesn't run on NuBus machines, so performance comparisons are moot. As far as I'm concerned, MkLinux is alive.
  • I have "fond" memories of MkLinux ever since I started with DR2 on my 6100. Playing with changing libraries (static to dynamic), changing disk formats, interesting X servers/clients, installing with -nodeps, sudden crashes, periods of stability... I loved it! Why?

    MkLinux introduced me to Linux. (note: it did not introduced me to unix, the &$#!@ IBM RT did that dirty deed). I was able to work at home writing code that I could then take back to work. I could program naked!:-)

    I now have a G3 and run LinuxPPC but I will not look down on the great MkLinux development folks. At the end, Apple "official" dropping of support for MkLinux really didn't mean much (except perhaps to Gary Thomas). So if I may, Good Luck David Gatwood in your new career. Anybody who could put up with constant questions about how their mouse buttons stopped working must be an angel.

    Speaking of the mouse, my Mouse System 3 button mouse is not officially support under LinuxPPC R5. I got tired of manually patching the kernel source (plus the mouse was bunged up) so I looked for another solution. Since I have a USB PCI card, I wondered if a USB 3 button mouse was supported. What nice about Linux? Go to the kernel source and look at the code to see if there is support. I now have a new logitech usb 3 button mouse that works just fine.
  • .. but mklinux is a pretty crappy distro imnsho.
    Linuxppc is pretty damn stable, while mklinux's latest (way dated) distro crashes if you look at it sideways. It's nice for nostalgia, I guess...
    I just think it pretty much illustrates Apples's true stance on open source or free software or whatever you want to call it.....
  • Just because they won't develop two completely different OS architectures in parallel? Please. Darwin seems luck a much better approach.


    Why?
  • Another thing I'm curious about: why isn't Linus Torvalds supporting MkLinux more strongly? I know he's a proponent of portable implementations of Linux, so it seems odd to me that he seems so focused on the Intel port of monolithic Linux.

    The problem is MkLinux isn't really Linux, it's the Linux kernel running on top of the Mach kernel, which means bloat and twice as many points of failure. Linus' preferring LinuxPPC to MkLinux is just like Linus theoretically preferring Linux-on-Alpha to a Linux subsystem to NT-on-Alpha. You might get more device drivers that way, and you might be able to run Linux apps anyway -- but it still would be NT, not Linux.
  • MkLinux is Mach 3.0.
    Darwin is approximately Mach 2.5

    However, the non-server MacOS X that will be released next year will be based on Mach 3.0. If Darwin is updated by Apple, it will then be easier to move MkLinux code to Darwin. And (virtually) nobody with the $500 for MacOS X Server is going to bother with a NuBus machine, anyway, so the demand for it won't be there until the client is released.

    In short, it actually makes sense to focus on MkLinux now and MkBSD (Darwin) next year...
  • Seeing how Linus originally developed Linux for the x86 (namely, 386) I would assume he wants to support that platform First, and worry about the less important platforms later.

    -- Give him Head? Be a Beacon?

  • MkLinux DR3, if I remember correctly, ran on the Mach 3.0 microkernel, and did support SMP. However, the SMP code used in the Linux kernel server was Daystar property, and therefore not open. To compile an SMP Linux kernel server, you had to go through a few hoops, which involved a dummy compile, then including a pre-compiled .o file. This was for the stock DR3 kernel, and may have changed in subsequent releases.

    JeC
  • I'm curious why you don't like Mach. I think that it's a curious design. Admittedly, you take a speed hit when using it, but are there other reasons you don't like it?

    neutrino
  • Give the guys who designed that website a raise. I like it. So much better than standard linux-related website fare :)
  • Q>WHAT VERSION?

    Q> Why does it matter ?

    A>Well there are several differant issues with diffrent versions as far as I know Mach 3.0 had stuff to do SMP in but before it, no go

    Q>am I correct ?

    oh well teach me slashdotter's

    john
    a poor student @ bournemouth uni in the UK (a deltic so please dont moan about spelling but the content)
  • dr3 is old. you can run mklinux's kernel/server on top of linuxppc.com's r5 (there are some minor 2.2.x vs. 2.0.x issues) or yellowdog, or debian. in fact, i'm running mk on debian on an old nubus box right now.
  • I thought that a netbsd came out recently for ppcs, even nubus ones. any word on this, anyone?

    --Andrew Grossman
    grossdog@dartmouth.edu
  • At the base level OS X is a unix derivative but scratch the surface and you find that there is still a place for linux on the PPC architecture.

    OS X will basically only run on Apple G3 hardware which is only a small part of the overall PPC hardware that is there already. It is in no way POSIX compliant - even apple have said that it will not be made to be so, therefore you can't just use the source luke.

    This means that OS X is set up as a commercial venture .... you have to pay for the software you want to run. Sure apache compiles out of the box ... now ... only after the apache group ported it accross. In my experience it looks good runs well but it is not an open-source solution.

    The other point about OS X is that it is based on the mach kernel so it inherently has the same performance issues as MK.

    Still my intro to Linux was MkLinux and I think that Mac OS will be inherently improved by having a full multitasking kernel.

  • The Open Group also has ceased development.

    That's true in general. There are only a few developers still working for TOG on some old DARPA contracts (mostly doing NT stuff). I'm pleased to see that the user community is keeping things going, though it would probably make more sense to get native Linux running on those systems only supported by mkLinux as a long-term solution.
  • I don't like Mach because it's bloated and buggy.

    I have trouble calling it a microkernel with a straight face, since it's so large.

    I have trouble keeping it running for more than a few hours, since I run real Mach code (not just stuff that talks to the Unix server). I've managed to patch the bugs that caused me the most trouble (in the external memory manager code, among other places), and they're in the mainline Open Group codebase, for all the good that will do anyone.

    We did do a complete re-write of Mach in C++, called mk++. That was a nice clean implementation; a good example of software engineering. Unfortunately, the plug was pulled just a few months before mk++Linux was to be released (as a CD-ROM accompanying a book).
  • I suggest you use google to search on linus (monolithic proponent) and tanenbaum (microkernel proponent). It is a very touchy subject for linus.

  • Yes, NetBSD did just release version 1.4 for several more 68k machines as well as PowerPCs, but does not support the Nubus Macs. More info can be found at NetBSD.org/Ports/macppc [netbsd.org].

    MkLinux was also gave me my first experience with installing and running Linux. I've really enjoyed the effort David Gatwood and others have put into the discussion groups and MkLinux.org [mklinux.org].
  • Let me get this straight. The server port of linux is 2.0.36 and will be abandoned in favour of 2.2.X when it's ported. The mach kernel is falling behind in hardware support and will have to have bits ported from Mac OS X. Isn't this going to become a bit unmanagable?

    And if you're a fan of Mach kernels why not get behind Hurd? To port Linux to Mach, you'll be in a forever catch-up mode with Linux proper. To do it on Mac hardware is to play forever catch-up with whatever Apple is so kind as to bestow on you.

    Hey, if they're having fun hacking on this, good on them. But it will always be slower than real Linux. And it will always be less "pure" and nice and micro-kernel-ish than Hurd. So what's the point?
  • - is Hurd still alive?

    I think so. They are in a quasi-0.3 release judgeing from the mailing list archive. When they release, I have no idea. I assume development may pick up once they do release.

    I am wondering, if anyone knows. Are there any advantages to using the Hurd other than it's microkernal design?

    --

  • Good luck to them, but I'm not expecting much from the MKLinux people, not because of skills, but from a lack of support of Apple and little social stroking from the users.

    I concur. Why don't the MKLinux guys use their considerable skills to get the new Darwin project off the ground? When Darwin 0.3 ships it will make absolutely no sense to support two different versions of the same kernel (Mach 3). Why not just grab the official Apple kernel and take it from there?

  • MkLinux was a great project and I'm sure people learned a lot about Mach and about hosting a Unix environment on top of a microkernel, but when Apple dropped their support in favour of Darwin it became obsolete. You guys are such great engineers that it hurts to see you waste your energies on a dead-end project when you could work on improving the future Apple OS. You would capture a much bigger audience if you incorporated your kernel modifications into Darwin and empowered mere mortals to run Mac OS X on their NuBus machines.
  • And if you're a fan of Mach kernels why not get behind Hurd? To port Linux to Mach, you'll be in a forever catch-up mode with Linux proper. To do it on Mac hardware is to play forever catch-up with whatever Apple is so kind as to bestow on you.

    Au contraire. When Mac OS X will be released, Apple's open source Darwin project (akin to the lower levels of OS X) will have all the hardware support you could ever ask for. And talking about dead technologies - is Hurd still alive? The latest release seems to be from 1997.

  • Because Apple is using Darwin as the core of it's next-generation (or is that NeXT-generation? ;-)) "real" OS (i.e. commercial OS that'll be preinstalled on Macs), so the average Mac user can benefit from the contributions of the OSS community, while the OSS community benefits from Apple's work.

    And also, I ran Mk (back when it was the only Linux implementation on the Mac), and it sucked . Sure Darwin doesn't run on NuBus machines or even some non-G3 PCI machines (it does run on most non-G3 PCI machines however, and so does Mac OS 10 Server*, in spite of Apple's claims), but it's Open Source, so if anybody cares enough that can change.

    *Stop the insanity Apple! Use numbers instead of Roman numerals before you confuse even more people!
  • Give the guys who designed that website a raise. I like it. So much better than standard linux-related website fare :)

    Thanks for the compliment. Especially the use of "guys" (plural)... ;-)

    Too bad it's only my hobby. No cookie for that work. :-/

    cya

    Ye Olde Webdesigner [surf.to]

    "Ma'am, what's yer favorite brand o' bread? In-Bread!"

  • I think Apple is (as it has been for the last couple of years) far too focused on generating hype and keeping the propaganda flowing to it's willingly naive user base.

    Be that as it may, what does that have to do with MkLinux? Just because Apple ended support doesn't mean that individual Apple people aren't involved. Some still are, though not officially. In the end, it makes no difference.

    I'd love to see a real linux port designed for NuBus Macs, but I am certainly not holding my breath.

    As an owner of two NuBus Macs, I heartily concur. But MkLinux has the best chance of quickly getting _something_ of Linux full-tilt on the NuBus platform, and many old Macs are out there crying for Linux. A 6100 with a halfway decent HD makes a great little server and can certainly hold its own with any Intel hardware of the same age.

    Could it be that they don't want to pave the way for a direct comparison of Mac hardware against x86-based hardware?

    Why? Any 6100 is just as good, if not better, than any 486 or early Pentium.

    Could it be that they're worried about MacOS market share?

    Apple doesn't give a flying rat's-you-know-what for what happens to old NuBus Macs. OS X won't run on them anyway, so why should they care? The person who wants to run Linux on a NuBus Mac is the same kind that can't afford a monster G3 with OS X anyway, so it's no skin off of their teeth.

    I don't know the answers to these questions, but I'm sure that Apple stockholders would be strongly against the idea of diluting the already weakened profitmaking potential of Apple products.

    *sigh* Sounds like flat-out Mac-bashing for little reason (yet again). I'm not going to try and claim that any Mac can flatten any Intel machine, because that's not the point. The point is, I already *have* that hardware and want to get the best out of it. That's the glory of Linux--I can install it on just about anything, and use it for just about anything. You can't do a hell of a lot with an old Pentium without Linux (or *BSD, for that matter), either. With it, you can do plenty.

    cya

    Ye Olde Webdesigner [surf.to]

    "Ma'am, what's yer favorite brand o' bread? In-Bread!"


  • I just think it pretty much illustrates Apples's true stance on open source or free software or whatever you want to call it.....

    Just because they won't develop two completely different OS architectures in parallel? Please. Darwin seems luck a much better approach.

    - Scott
    ------
    Scott Stevenson
  • I think people are missing the point. This should really show the world that even when the "official" developers have dropped support, an open source project (ANY open source project, even those that people seem to hate) will continue.

    There will always be people who want to use MkLinux for one reason or another. If at the very least it is to support the older NuBus macs and give them additional life.

    --Mark
    Proud MkLinux & LinuxPPC developer
  • whats next, Whopper NT?
  • I think Apple is (as it has been for the last couple of years) far too focused on generating hype and keeping the propaganda flowing to it's willingly naive user base.

    I'd love to see a real linux port designed for NuBus Macs, but I am certainly not holding my breath.

    Could it be that they don't want to pave the way for a direct comparison of Mac hardware against x86-based hardware? Could it be that they're worried about MacOS market share?

    I don't know the answers to these questions, but I'm sure that Apple stockholders would be strongly against the idea of diluting the already weakened profitmaking potential of Apple products.

Slashdot Top Deals

Today is a good day for information-gathering. Read someone else's mail file.

Close