Linked by Eugenia Loli on Fri 8th Apr 2005 20:26 UTC, submitted by Robert Millan
Debian and its clones Debian just added the kernel of FreeBSD to the unstable branch of its GNU/Linux distribution. The package contains pristine 5.3 kernel source together with patchset for conformance to Debian's requirements. This is a first step towards inclussion of the Debian GNU/kFreeBSD port as a candidate for future Debian releases.
Order by: Score:
RE:Debian adds kernel of FreeBSD
by Mark Turner on Fri 8th Apr 2005 20:29 UTC

Did I miss somthing.....whats the point?

RE:Debian adds kernel of FreeBSD
by Rude Turnip on Fri 8th Apr 2005 20:33 UTC

Debian is more than a Linux distro...it's a distro that can be adapted to different kernels. Debian HURD is the example I know of.

RE:Debian adds kernel of FreeBSD
by Anonymous on Fri 8th Apr 2005 20:33 UTC
v Awesome~!
by Chris on Fri 8th Apr 2005 20:38 UTC
Whats wrong with FreeBSD
by Vikram Sharma on Fri 8th Apr 2005 20:38 UTC

WHy does the Linux community hate the idea of FreeBSD. UFS 2 is far superior to ext3, Debian is the best linux flavor out there, nothing wrong in Debian incorporating FeeBSD. Kudos to the guys at debian

bsd propoganda
by Chris on Fri 8th Apr 2005 20:40 UTC

Their propoganda talks about the upcoming release of linux 2.4... It's VERY outdated. At this point (2.6.11) I think Linux has pretty well caught up to BSD in most things...

RE: FreeBSD propaganda
by Larry on Fri 8th Apr 2005 20:41 UTC

How about old propaganda considering that comparison sheet talks about the 2.4 release of the linux kernel "will" ... and also how RedHat used to turn on way too many services. Which was true about 5 years ago and was why RedHat out of the box installs would get owned as fast as a windows box.

v re: what's wrong with FreeBSD
by Larry on Fri 8th Apr 2005 20:42 UTC
v Re: re: what's wrong with FreeBSD
by John on Fri 8th Apr 2005 20:47 UTC
apt-get freebsd?
by Nathan O. on Fri 8th Apr 2005 20:49 UTC

So... wait a minute... is this trying to say I could have a working Linux system and just apt-get myself to FreeBSD?

I know Debian's been incorporating other kernels for a while now. I imagine not all of them are binary-compatible, though. I suppose BSD's have Linux emulation, but since *BSD has its own core-utils type stuff, wouldn't that mean the GNU packages...? I think I'm just confused by the wording.

v Re: what's wrong with FreeBSD
by Anonymous on Fri 8th Apr 2005 20:50 UTC
re: BSD license is more free
by Dave on Fri 8th Apr 2005 20:51 UTC

you should chose your words more thoughtfully. free is a bit too subjective. i think you should say "less restrictive". the difference is what free is quantifying. anyhow, i think this is a great idea. at least a good exercise.

RE: bsd propoganda
by Duffman on Fri 8th Apr 2005 20:51 UTC

"Their propoganda talks about the upcoming release of linux 2.4... It's VERY outdated. At this point (2.6.11) I think Linux has pretty well caught up to BSD in most things..."

I don't think so. As Linus says, the linux kernel 2.6.x is slower than 2.4.x
http://news.com.com/Torvalds+Put+Linux+to+the+test/2100-7344_3-5646...

BTW, why the debian project is searching another kernel and making a new unstable branch, if linux is the best ?

@ Anonymous
by Manik on Fri 8th Apr 2005 20:58 UTC

You're certainly not a lawyer ! Except that it's good for Debian to have many kernels, I didn't understand anything you wrote !

RMS was right!
by Anonymous on Fri 8th Apr 2005 21:02 UTC

Suddenly calling it GNU/* makes sense. This is a *GNU* distro *without* Linux.

BSDL more free?
by Anonymous on Fri 8th Apr 2005 21:02 UTC

If by free you mean allowing 3rd parties to steal and profit from the hard work of volunteer developers without giving anything back, then yes, by all means, the BSDL is far more free than the GPL.

I'll chose fairness over freedom anyday. Apparently from SF/Freshmeat license stats most people agree with me.

RE: bsd propoganda
by Josel on Fri 8th Apr 2005 21:08 UTC

Hmmm Guess i'm lucky since im running on 2.6.10.

Debian is not searching for a kernel they are just adding another choice. But its a dead idea.. Probably very few users will be interested in BSD.. at least i dont know any that are.

Re: Duffman
by Felipe B. on Fri 8th Apr 2005 21:08 UTC

BTW, why the debian project is searching another kernel and making a new unstable branch, if linux is the best ?

They are not in searching of another kernel... Debian already run on other kernels like Hurd and NetBSD far time ago, they started the use of freebsd kernel just because someone wanted, without any technical reason.

@Duffman (IP: ---.fbx.proxad.net)
by r_a_trip on Fri 8th Apr 2005 21:16 UTC

BTW, why the debian project is searching another kernel and making a new unstable branch, if linux is the best ?

They are not searching another kernel, they have added the FreeBSD core, like they are also maintaining a GNU/Hurd branch.

Debian provides a multitude of choice, which is a Good Thing (tm).

@Anonymous
by rycamor on Fri 8th Apr 2005 21:22 UTC

If by free you mean allowing 3rd parties to steal and profit from the hard work of volunteer developers without giving anything back, then yes, by all means, the BSDL is far more free than the GPL.

Look... You guys just have to get over this. Those who contribute to BSD-licensed projects *know* how their work can be used. So, if they choose to volunteer, that's because they are willing to have their work used in that way. Thus it's not stealing, and there is nothing unfair about it. But it is unfair of you to judge based on presumptions that don't apply to the situation.

I'll bet that Stallman is quite happy
by Joe Buck on Fri 8th Apr 2005 21:22 UTC

I suspect that the FSF is quite happy with the idea of Debian GNU/kFreeBSD, because they've long felt that Linus gets too much credit. A system with a BSD kernel and GNU userland that casual users would find indistinguishable from a Debian GNU/Linux system would provide an interesting demonstration.

Also, back around '91 the GNU people were starting to look at extending the Networking-2 partial releases coming out of Berkeley to full "GNU/BSD" systems, when two things intervened: the release of Linux and the AT&T/BSD lawsuit.

So the Debian folks are building the system that might have been.

v Re: Duffman
by nona on Fri 8th Apr 2005 21:23 UTC
Re: Duffman (bsd propaganda)
by ylai on Fri 8th Apr 2005 21:43 UTC

I don't think so. As Linus says, the linux kernel 2.6.x is slower than 2.4.x
http://news.com.com/Torvalds+Put+Linux+to+the+test/2100-7344_3-5646...

Do you know what the Red Hat Enterprise Linux 3 kernel is? It is a massively patched 2.4.21 with almost all significant features in 2.6, from scheduling to VM. It is not a wonder that this kernel has a performance like kernel 2.6. Read this first next time:

http://www.redhat.com/software/rhel/3features/kernel/

BTW, why the debian project is searching another kernel and making a new unstable branch, if linux is the best ?

It is a shame that you haven't even learned that open source is about choice.

Re: Nona
by Jon on Fri 8th Apr 2005 21:46 UTC

Actually, in some situations 2.4 IS faster then 2.6.

Copy left
by CaptainN on Fri 8th Apr 2005 21:48 UTC

If Debian releases a version of Free BSD with GNU userland as GNU/FreeBSD, wouldn't that make the kernal in that release covered under the GNU license? Isn't it viral? In this way the Debian folks would be "stealing" (it's not stealing all right, get over it) the kernal from FreeBSD, and would continue to "use" the work by applying bug fixes from the FreeBSD coders.

Btw, they should be able to add any changes or fixes they come up with for the FreeBSD kernel back to FreeBSD. All they would have to do is give the code back to FreeBSD before they release it (and therefor change the new GNU license), so that it is covered under the FreeBSD license and then take it back for their needs.

Re: Copy left
by zerblat on Fri 8th Apr 2005 21:58 UTC

If Debian releases a version of Free BSD with GNU userland as GNU/FreeBSD, wouldn't that make the kernal [sic] in that release covered under the GNU license?

No, it wouldn't. The GPL doesn't put any requirements on the license of the OS on which you run GPL'd programs.

debian is lacking docs
by tom on Fri 8th Apr 2005 22:00 UTC

The biggest drawback on Debian seems, that the devs don't seem to like user documentation.

This is a major advantage of the BSD systems, they are all well documented, only few Linux distros are to such extend.

"bsd propaganda" reference
by ylai on Fri 8th Apr 2005 22:05 UTC

The actual discussion in LKML is at: http://kerneltrap.org/node/4940

RE:Whats wrong with FreeBSD
by Uno Engborg on Fri 8th Apr 2005 22:07 UTC


WHy does the Linux community hate the idea of FreeBSD. UFS 2 is far superior to ext3, Debian is the best linux flavor out there, nothing wrong in Debian incorporating FeeBSD.


I'm curious, what makes UFS 2 bettr than ext3?
How would UFS 2 measure up against other popular file systems like XFS, and ReiserFS?

GNU/Whatever
by David Johnson on Fri 8th Apr 2005 22:07 UTC

This is a *GNU* distro *without* Linux.

Except for the fact that GNU didn't make that distro. The only reason it's called "GNU/kFreeBSD" is because Debian chose to call it that.

GPL userland on non GPL kernel
by Darren Moffat on Fri 8th Apr 2005 22:12 UTC

If running GPL licensed GNU userland bits on a non GPL kernel was an issue the GNU project would never have gotten off the ground. SunOS 4.x not Linux for a long time was the primary GNU userland development platform for many people.

I see this as very interesting, it gives great choice. It also shows how mature and portal the GNU userland needs to be for this to work.

I suspect that one day we might also see a Debian GNU/OpenSolaris done in a very similar way to the Debian GNU/*BSD releases. Briging a nice (for me anyway) historical twist to the GNU project.

Gentoo
by Marcos on Fri 8th Apr 2005 22:19 UTC

There is a project to use the BSD kernel as core of Gentoo: http://www.gentoo.org/proj/en/gentoo-alt/bsd/

It's only few months old.

What about Linux kernel together with BSD software?
by Claudio on Fri 8th Apr 2005 22:20 UTC

What about going on the other direction, I mean, joining the Linux kernel with BSD (non-GPL) stuff? So we wouldn't have to (according to Stallman) call it 'GNU/Linux' anymore...

RE: @ Uno Engborg
by Vikram Sharma on Fri 8th Apr 2005 22:33 UTC

I have used ReiserFS, ext3 and UFS 2. Compared to UFS ext3 felt slow, also I have had ext3 crash on me much more than UFS, could also be because I have used FreeBSD in my server for only a month. ReiserFS is pretty fast, I would personally choose REiserFS over ext3 but there aren't many disrubtuions of Linux that allow ReiserFS install the ones I know are Ubuntu and XandrOS. Ubuntu was freaking fast on Reiserfs compared with Fedora using ext3.

RE:RE: FreeBSD propaganda
by Wrawrat on Fri 8th Apr 2005 22:36 UTC

According to Firefox, the last modification date for this page was January 5, 2002 19:32:27.

The propaganda was accurate... The propagandist should use more up-to-date information, though. ;)

@Vikram Sharma - ReiserFS
by Anonymous on Fri 8th Apr 2005 22:39 UTC

Add slackware to your list. they use ReiserFS by default. at least last time i check slackware 10 was using ReiserFS as its first option.

Licenses
by John Nilsson on Fri 8th Apr 2005 22:42 UTC

There is no point comparing BSDL and GPL they don't have that much in common.

BSDL (for reasons that elude me) is just a small restriction short of being in the public domain.

GPL is an FSF tool to spread "freedom".

What a waste of thought
by Marc J. Driftmeyer on Fri 8th Apr 2005 22:43 UTC

How is it a thread about FreeBSD being added to the Debian Family turns into a war between licensing philosophies?

Where are all the threads on building this kernel and bootstrapping it to another partition and bootloader leaving one the ability to choose between the various kernel options underneath the Debian Hierarchy?

@Vikram Sharma - ReiserFS
by John Nilsson on Fri 8th Apr 2005 22:44 UTC

You can add Gentoo to the list. I even think there is a livecd for Reiser4 installs somewhere (not official though).

@Vikram Sharma
by finalzone on Fri 8th Apr 2005 22:44 UTC

Ubuntu was freaking fast on Reiserfs compared with Fedora using ext3.

Very bad comparison because you compare two different distros with different format systems. It would make sense if you were compare Ubuntu with reiser and Ubuntu with default fs(ext3).
FYI, you can install ReiserFS on Fedora on the boot screen : linux reiser selinux=0.

Why GNU/BSD?
by John Nilsson on Fri 8th Apr 2005 22:46 UTC

I was under the impression that the thing about BSD was the development model which leads to an integrated OS with a clear direction.

I fail to see how copying their kernel could bring that to Debian...

re: re: what's wrong with FreeBSD
by yawn on Fri 8th Apr 2005 22:53 UTC

FSF/GNU extremists (some of them) don't like FreeBSD because the BSD license is more free than the GPL.

Or maybe they like FreeBSD just fine. But get tired of the BSD extremists missing the point of the GPL. We hear you talking about free this and free that, but you never define this free thing, which leads me to believe you don't really understand what you're talking about.

If it gives freedom to take away freedom, it results in less total freedom than one that takes away that freedom from you, requiring to preserve freedom.

*head explodes*

v lol
by Anonymous on Fri 8th Apr 2005 22:58 UTC
RE: Vikram Sharma - ReiserFS
by Hassy on Fri 8th Apr 2005 23:00 UTC

Mandrake (the late) has had reiserFS as far back as 9.x. Maybe not at it's 1st choice but it was there.

v hippies
by yawn on Fri 8th Apr 2005 23:03 UTC
RE: "portal" GNU
by Kors on Fri 8th Apr 2005 23:06 UTC

>It also shows how mature and portal the GNU userland needs >to be for this to work.
That's why the initial idea of debian team - use bsd's libc + kernel with gnu software failed misreably. Some parts of gnu userland turned out to be unportable, due to lack of development culture, and loads of linuxisms and glibcisms in the sources... And anyone who ever tried to compile auto*, configure, etc. oriented programs on platform with non-gnu userland/kernel/compiler/includes/libraries knows that it's going to be a major PITA next time too... Because one needs "right" awk, "right" cc, "right" ld and "right" everything. Where "right" is NOT standards (I mean posix/sus) compatible, but is GNU. The destruction of a single positive result of unix wars (standards) is in progress, due to gnu/linux crowd with their "mature" and "portal" software.

Re: Vikram Sharma - ReiserFS
by ylai on Fri 8th Apr 2005 23:10 UTC

Mandrake (the late) has had reiserFS as far back as 9.x. Maybe not at it's 1st choice but it was there.

Also SuSE used ReiserFS as its first choice for a long time. But at that time, it was actually a bad choice. ReiserFS caused huge problems with RAID and SMP configurations, because everyone was testing it using IDE and on simple personal computers.

Re: "portal" GNU
by ylai on Fri 8th Apr 2005 23:17 UTC

Because one needs "right" awk, "right" cc, "right" ld and "right" everything. Where "right" is NOT standards (I mean posix/sus) compatible

You forgot to mention: Neither is xBSD fully POSIX and e.g. UNIX98 complaint.

re: Re: "portal" GNU
by Kors on Fri 8th Apr 2005 23:22 UTC

Did I tell anything good about BSDs? ;)
BTW I use several linux distros (including debian) + fbsd + openbsd + solaris + windows, without special preference (though, windows tends to cause more troubles ;) ). I'm not pro or against any system in particular, just against zealous praising of smth. based on false assumptions.

RE: Duffman and others
by Adam on Fri 8th Apr 2005 23:30 UTC


I don't think so. As Linus says, the linux kernel 2.6.x is slower than 2.4.x


I think the benchmark you are referring to is comparing Red Hat's heavily modfied 2.4 kernel, which contains 2.6 features.

Regardless 2.4 is still faster than Free/Net/Dragonfly/Open BSD.

ext3 has worked fine for me, and if you don't like it there is jfs, xfs, reiserfs, and reiser4 which are all better than ufs2.

I have nothing against BSD, but it is not GNU/Linux; it is far behind in terms of important functionality, stability, performance, and software support.

v Why?
by Raven on Fri 8th Apr 2005 23:30 UTC
This never cease to amaze me
by bleyz on Fri 8th Apr 2005 23:31 UTC

So osnews readers are interested in OSes, and yet they don't know that Debian is a F/OSS system that just happens to rely on L***x for a kernel, but likes the idea of being able to use FBSD, NetBSD, the HRD or whatever.

Re: RE: Duffman and others
by Kors on Fri 8th Apr 2005 23:43 UTC

>Regardless 2.4 is still faster than Free/Net/Dragonfly/Open >BSD.
Why do the ghosts have short haircuts? (the question makes more sense than your statement). Faster doing what? On which platforms?

>ext3 has worked fine for me, and if you don't like it there >is jfs, xfs, reiserfs, and reiser4 which are all better >than ufs2.
Better with regards to what?

>I have nothing against BSD, but it is not GNU/Linux;
And you are a GNU/Linux zealot ;)

> it is far behind in terms of important functionality,
> stability, performance, and software support.
I've allready commented on perfomance.
I could (possibly) agree about software support and scalability. But stability? important functionality? Sounds very much like Torvald's "Solaris has no advantages against linux" raving....

RE:Raven
by Adam on Fri 8th Apr 2005 23:43 UTC


Why would i want to use the BSD kernel instead of the Linux kernel? are there big difference?


I assume you mean FreeBSD's kernel, because various BSD's use different kernels and differ in features.

They are both UNIX-like kernels. Particulary Linux strives for POSIX conformance, atleast as much as it can. Linux's ability for threading and support for a large number of processors has much more maturity and ability than the FreeBSD's. There are many other differences, that is the main one.

There is very little reason to use the FreeBSD kernel in my opinion.

BSDL and GPL are both equally free...
by a nun, he moos on Fri 8th Apr 2005 23:53 UTC

...they simply express their freedom in different ways.

The BSDL and the GPL are both good licenses, they are both compatible, and both FreeBSD and Linux distributions include programs covered by each license.

Both licenses have their own strengths and weaknesses (though the GPL seems more popular, at least according to freshmeat). They are both equally, if differently, free. Anyone trying to argue that one is better than the other is not worth listening to. They are simply trying to start another sterile flamewar. Ignore them. Sometimes I suspect they really are MS astroturfers simply trying to sow disunity among F/OSS enthusiasts.

I like both licenses, just as I like both the idea of Debian Linux and Debian BSD. I really wish some people with anti-GPL agendas would stop trolling, and some pro-GPL posters would stop biting...

RE:Kors
by Adam on Fri 8th Apr 2005 23:55 UTC


Why do the ghosts have short haircuts? (the question makes more sense than your statement). Faster doing what? On which platforms?


Generally GNU/Linux is faster. All the recent benchmarks I have seen point to this. All the benchmarks I have seen, usually bechmarking networking performance or a particular service, confirm Linux being faster on the x86 architecuture. However we will not know if FreeBSD is faster on pp64 or S/390 because it does not run on it.


And you are a GNU/Linux zealot ;)


Not really I ran FreeBSD for years and I currently have NetBSD installed and is to inevitably be a personal webserver. I am just pointing out the truth.

Linux does have important functionality such as the ability to run on quad processor servers reliabily and efficiently.

Stability is a matter of opinion, I would say.

re: RE:Raven
by Kors on Fri 8th Apr 2005 23:58 UTC

>There are many other differences, that is the main one.

Why is that difference main? what about NFS perfomance and stability, for example? Code maturity (i.e. amount of childish ugly hacks and in kernel code)? Security? how many smp-boxes do you have on your desk? How many processors do you need on the router or firewall? There are systems with scalability that is by orders of magniture ahead of linux's kernel (I mean solaris). So, does that make solaris a silver bullet?

What's the obsession with speed?
by Calroth on Sat 9th Apr 2005 00:03 UTC

If there is a 20% difference in kernel speed, I say, so what?

If kernel speed is so important to you that you need that final 20% boost, for example if you are doing ganing or high-performance computing, then you might look at the choices available. For everyone else, I would say that functionality, stability and reliability are more important than that last 20% speed boost.

By "everyone else", I mean 95% of OSNews readers and posters to this forums. Nobody will be playing games on FreeBSD and few on Linux, so that leaves the high-performance computing crowd. Who's in that?

Re: re: RE:Raven
by ylai on Sat 9th Apr 2005 00:09 UTC

There are systems with scalability that is by orders of magniture ahead of linux's kernel (I mean solaris). So, does that make solaris a silver bullet?

I think you are quite mistaken. Scalability is mainly a platform issue. Linux scales on IA-64 upto 256 CPU and beyond per NUMA node. That IA-32 has very poor scalability is no secret fact. Otherwise, fell free to test Solaris 10 x86 on 32-way SMP.

RE: Kors
by Adam on Sat 9th Apr 2005 00:17 UTC

I do not know any childish ugly hacks in Linux, if that is what you are implying, not present in different forms in other open-source systems. And when Solaris 10 source code is released to the masses I am sure there will be some there also.


There are systems with scalability that is by orders of magniture ahead of linux's kernel


Right, that is why Linux is chosen to run on the majority of super-computers.

Re: re: RE:Raven
by ylai on Sat 9th Apr 2005 00:18 UTC

Security?

By your experience, I guesst you do not need somebody else telling you that Red Hat Enterprise Linux 4 also supports mandatory access control as like Trusted Solaris.

re: ylai
by Kors on Sat 9th Apr 2005 00:19 UTC

>I think you are quite mistaken. Scalability is mainly a >platform issue. Linux scales on IA-64 upto 256 CPU and >beyond per NUMA node. That IA-32 has very poor scalability >is no secret fact. Otherwise, fell free to test Solaris 10 >x86 on 32-way SMP.
I'm speaking about sparc n-head hardware. Where linux does not really run well ;) As for x86 - linux 2.6 and solaris 10 perfomance on SMP (tested myself with 2 and 4 processors) is on par. BUT Solaris does have special facilities that are unavailable or have inferior rivals on linux even on x86 and x86-64. Zones, dtrace, and (I am desperately waiting for it ;) ) zfs.

Scalability and ReiserFS
by Matt on Sat 9th Apr 2005 00:20 UTC

So Solaris has better scalibility than linux? That's news to me, and several benchmarks I've seen that show the two are pretty much even.

ReiserFS should be fine with any distribution. The ability to use it is built into the kernel and has been there for quite some time. The only question is whether your distro gives you the choice when you are installing it. All except the newbie distros should give you a choice of several types.

Oh, and Linspire is a big backer of Reiser, so add them to the list which defaults to ReiserFS. (version 4, in this case)

re: all
by Kors on Sat 9th Apr 2005 00:25 UTC

>Right, that is why Linux is chosen to run on the majority of >super-computers.
:) So, windows is the superior platform, just for reason of being run on majority of desktops? ;)

>By your experience, I guesst you do not need somebody else >telling you that Red Hat Enterprise Linux 4 also supports >mandatory access control as like Trusted Solaris.
I was thinking about openbsd and their understanding of security as the endless process of code auditing, not MAC/Trusted* ;) I think we both know that security of the system is not determined only by facilities present.

Re: Kors
by ylai on Sat 9th Apr 2005 00:27 UTC

I'm speaking about sparc n-head hardware. Where linux does not really run well ;)

Then specify that explicitly in your critism. SPARC makes a extremely small fraction of Linux hardware installation base. If you need SMP scalability, you definitively should try to run Linux on IA-64 or POWER5. And both platforms are definitively not inferior compared to SPARC.

BUT Solaris does have special facilities that are unavailable or have inferior rivals on linux even on x86 and x86-64. Zones, dtrace, and (I am desperately waiting for it ;) ) zfs.

Dtrace is the typical example always popping up in the discussion - there seems to be no other "real" argument for Solaris.

Concerning zones/containers, why don't you try to run a NetBSD kernel in it? This is one field where Xen is definitively more superior in term of technology.

Can you then give me a reason why AFS and GFS cannot serve your porpose?

re: ylai
by Kors on Sat 9th Apr 2005 00:43 UTC

>Dtrace is the typical example always popping up in the >discussion - there seems to be no other "real" argument for >Solaris.
Well, it is an argument ;)

>Concerning zones/containers, why don't you try to run a >NetBSD kernel in it? This is one field where Xen is >definitively more superior in term of technology.
True. Though, I can't imagine a real-life situation with the need to run netbsd under solaris. W2k3 - probably. NetBSD - why?

>Can you then give me a reason why AFS and GFS cannot serve >your porpose?
Builtin volume management on filesystem level, copy-on-write, fine-grained I/O priority/scheduling, transactions (rollbacks!!!). AFS and GFS are network-distributed centric, I'm interested in a huge fs on a single server. If zfs is really as good as sun tells, I can see no rival for it. Though - I agree, it's vapourware for now.

Just take the FreeBSD kernel and re-release as GPL.
by Han Solo on Sat 9th Apr 2005 01:01 UTC


You do know that you can do this right?

You can take the FreeBSD kernal and re-release it under the GPL.. its perfectly legit. Thats part of the power the BSD license gives you. You just can go the other direction.


I was thinking just the other day of taking some orginal BSD source code and turning it into a GPL application. So, I did quite a bit of research on this to double check it.



Re: Kors
by ylai on Sat 9th Apr 2005 01:10 UTC

True. Though, I can't imagine a real-life situation with the need to run netbsd under solaris. W2k3 - probably. NetBSD - why?

This is just an example. There are however many good reasons why people are interested to run a strictly separated virtualization rather than just zones/jails type "soft" spearation, for security reasons, to begin with.

That you can run an entirely different kernel is just a demonstration of the strict virtualization.

AFS and GFS are network-distributed centric, I'm interested in a huge fs on a single server.

AFS and GFS are just two examples. There are also GPFS, SANFS, and many others. If you look around in the fields with extreme I/O demands, these are the typical FS used. I do not see the point why anyone should have a single server centric solution, or emphasize this.

This is so good.
by itomato on Sat 9th Apr 2005 01:11 UTC

The stability & hardiness of FreeBSD with the beauty that is apt..

Ahh.

FreeBSD is great, just not as a "distro". As an Operating System - yes. A sweet-n-easy joyride it is not.

A DebianBSD would be!!!

re: hippies
by h on Sat 9th Apr 2005 01:17 UTC

"I really think that 90% of GNU Hippies dont' care about the licenses, they just dont' want people making money off of their code (such as Microsoft). And yes Microsoft using BSDL code is a win-win situation for a billion reasons.

I think that 90% of BSD Hippies wish they could get a handful of those billion reasons Microsoft put in the bank"

You've missed the point, BSD is about excellence. What BSD devs care about is writing good code and have the community use it. There is sactisfaction in the fact that your work is thanked and appreciated by millions of people.. hope one day you can realize it's not all about money.

re: Duffman and others
by h on Sat 9th Apr 2005 01:26 UTC

"Regardless 2.4 is still faster than Free/Net/Dragonfly/Open BSD."

> "Linux > BSD!!"

ext3 has worked fine for me, and if you don't like it there is jfs, xfs, reiserfs, and reiser4 which are all better than ufs2.

> "Linux > BSD!!"

I have nothing against BSD, but it is not GNU/Linux; it is far behind in terms of important functionality, stability, performance, and software support.

> "Linux > BSD!!"

Come on, where is any of the BSDs beyond linux in stability? Try to stress-test linux and then a FreeBSD 4.10 or a OpenBSD box and then tell me something.. You're talking about linux, the OS that still has problems with fork() bombing.. FreeBSD won't crash with a fully loaded desktop (GNOME) and fork()ing it up to 8000 processes.. linux crashed on my as soon as I suspended the fork bomb and then executed it again. Just an example out of dozens I could think off.. Kernel security is important too, try to search for linux kernel exploits vs *bsd kernel exploits, then come back saying linux is more stable.

Performance ? Not that big of a difference, and I think you'll be amazed when the work on FreeBSD's 5.X branch is over.. And it's not about performance only, what good is performance if the system isn't stable ?

Software support ? Ports system + the ability to run 99% of Linux software pretty much even the situation, wouldn't you agree ?

About filesystems.. I've had ext3 dying after an unexpected power loss.. As for UFS, I've never, ever lost a filesystem due to ANY REASON in 6 years of FreeBSD. Be it UFS 1 or 2. Can't comment on the other filesystems as I've never tried them, but UFS seems pretty stable and mature to me - as it should be, it's many, many years old.

re: re: kors
by h on Sat 9th Apr 2005 01:30 UTC

"I do not know any childish ugly hacks in Linux, if that is what you are implying, not present in different forms in other open-source systems. And when Solaris 10 source code is released to the masses I am sure there will be some there also.


There are systems with scalability that is by orders of magniture ahead of linux's kernel


Right, that is why Linux is chosen to run on the majority of super-computers."

lol, and you deny you're a linux zealot.

Last time I checked, it was Solaris 8, HP-UX and AIX running on super computers, not Linux.

RE: h
by uF0 on Sat 9th Apr 2005 01:31 UTC

"What BSD devs care about is writing good code..."
oh please if you were talking about NetBSD's code then i would agree but if you say that FreeBSD devs give a fuck about clean code and not implementing hacks then i have completely different opinion.

RE: h
by uF0 on Sat 9th Apr 2005 01:33 UTC

maybe you have something better to do and not only troll around here ?

Isn't FreeBSD a registered trademark, now owned by The FreeBSD Foundation? I don't see why The FreeBSD Foundation should have given permission to use the trademark in a project that is going to be in competition with FreeBSD, since FreeBSD markets the integration of the entire system and not just the kernel as one of its strengths.

They create the monster!!!
by Carlos on Sat 9th Apr 2005 01:45 UTC

Yeh, they create the operating system's Frankenstein
Thanks guys!



p.s.This guys have the "monster", but in the other hand don't ever had gnome2.10 in the packages!!!
LOL

OMG
by Jim on Sat 9th Apr 2005 02:06 UTC

wang burglar sausage!

Linux
by Anonymous on Sat 9th Apr 2005 02:08 UTC

"
Last time I checked, it was Solaris 8, HP-UX and AIX running on super computers, not Linux.
"

From the Top500 supercomputers list

NUMBER 1
IBM's BlueGene/L Supercomputer -- runs Linux

NUMBER 2
SGI's Columbia Supercomputer -- runs Linux

NUMBER 3
Earth Simulator -- runs Enhanced NEC Unix

NUMBER 4
IBM's MareNostrum -- runs Linux

NUMBER 5
IBM's (I believe) Thunder -- runs Linux

So, out of the top 5 spots, Linux holds 4...but wait, there's more. It was recently estimated that Linux holds 300 spots out of the top 500 for Supercomputers. So, if you combine AIX, Solaris and HP-UX, then they almost compare to Linux, but not quite.

See how much more believable you sound when you do a little research instead of saying "last time I checked".

RE: h
by Matt on Sat 9th Apr 2005 02:16 UTC

From http://www.forbes.com/home/enterprisetech/2005/03/15/cz_dl_0315linu...

Meuer reckons Linux powers 301 of the 500 top machines, compared to 189 on Unix, two on FreeBSD, a Unix variant, and one on Microsoft's (nasdaq: MSFT - news - people ) Windows. (Seven machines are categorized as "other.")

Moreover, Seager says Linux outguns popular Unix operating systems like AIX and Solaris from Sun Microsystems (nasdaq: SUNW - news - people ) because those systems contain features that make them great for commercial users but add a lot of system overhead that ends up limiting overall performance. One example: a "virtualization" feature in AIX lets many applications share the same processor but "just hammers performance," Seager says.

*BSD and GNU tools
by Anonymous on Sat 9th Apr 2005 02:30 UTC

Well, guess what:

Some people like:
1) Linux Kernel
2) FreeBSD Kernel
3) FreeBSD Utilities
4) GNU Utilities

I can honelsty say, I like the FreeBSD kernel and do like the GNU Utilities. The FreeBSD kernel is easy to update, via CVS and easy to rebuild, by simply editing a text file.

PS: I like the run levels in *BSD vs the Sys V. or in Linux. And yes, there are some Linux Distro's that use the *BSD run levels. I am wondering how the run levels will work in the Debian Release.

This does scratch an itch for some folks. Forget the all the talking of smack. Use what you like and prefer.

Enjoy

The depenguinator lives in Debian!
by Anonymous on Sat 9th Apr 2005 02:48 UTC

Long live the depenguinator.

Re: h
by ylai on Sat 9th Apr 2005 02:57 UTC

Last time I checked, it was Solaris 8, HP-UX and AIX running on super computers, not Linux.

The blatant mismatch with the reality is already commented by other. But what surprises and wonders me is actually that - sorry - some of your claim appears just ridiculous:

- For AIX, you should know that IBM is slowly transiting from AIX to Linux, especially in term of HPC. IBM pSeries HPC solutions running all AIX was mainly on POWER4+. You should update your information.

- The high of Solaris based HPC solutions were 2000 (5 years ago!), and they still ranked typically below top 40. With CPU like POWER5 or Itanium 2 emerging in recent years, Solaris based HPC is just not competitive anymore.

- HP/UX? How well do you actually know Hewlett-Packard in the first place? PA-RISC and HP/UX was never the HPC platform of choice for HP. The AlphaServers raking at place 2 and 3 on Top 500 list in 2002/2003 was all running Tru64.

kernels of corn
by javajazz on Sat 9th Apr 2005 03:25 UTC

it is nice to have a variety of seed. the farmer then has a better chance of adapting to changes. we all know this. does this not apply to distros. we need variety for who knows what will be required in the future. we need to keep the airwaves free!

FreeBSD + apt-get
by dimosd on Sat 9th Apr 2005 03:34 UTC

That's what I 'd like to see, apt-get handling binary packages on FreeBSD (kernel+userland). Ports are nice and all, but binaries aren't their strength.

FreeBSD kernel + GNU... not that interesting IMHO

FreeBSD + apt-get
by emagius on Sat 9th Apr 2005 03:40 UTC

What's wrong with packages? Alternatively, what's better about apt-get?

On the issue of License being "more free"
by Gent on Sat 9th Apr 2005 03:51 UTC

I think this is dependent on how you look at the term "free software." The way I see it, it's not meant to imply that the user is free to do what they will with the software, but that the software itself is free. As if it were a person -- the software is emancipated so to speak, no one can "control" it in any way.

This is a lot different than saying anyone is free to do what they want with the software. By looking at the software itself as having been emancipated, you can see why the term "free software" adheres more properly to GPL than the BSD Licesne. If corporations are free to grab hold of software and exploit it so to speak, then the software itself has lost it's free nature.

This will make a lot more sense with AI programs and the idea of conscious data -- to the future!

And before you all ask... yes, I'm a communist.

focus
by ed on Sat 9th Apr 2005 03:52 UTC

No wonder Debian Linux 3.1 is taking so long to get release
the Debian team are doing other things.

@h
by a nun, he moos on Sat 9th Apr 2005 04:03 UTC

Last time I checked, it was Solaris 8, HP-UX and AIX running on super computers, not Linux.

Sorry, but you're wrong. Linux dominates the supercomputer field. Blue Gene, the SGI supercomputer and other top entries all run on Linux.

http://news.com.com/Blue+Gene,+Linux+top+supercomputing+list/2100-7...

You're talking about linux, the OS that still has problems with fork() bombing.. FreeBSD won't crash with a fully loaded desktop (GNOME) and fork()ing it up to 8000 processes..

This is true for some distros' default settings, however it's trivial to fix this by limiting processes. On some distros (Debian, I think) the default is quite safe and the system is immune to such fork bombs.

What BSD devs care about is writing good code and have the community use it. There is sactisfaction in the fact that your work is thanked and appreciated by millions of people.. hope one day you can realize it's not all about money.

That's pretty much the same motivation that's behind Linux devs as well. And the GPL license is certainly not all about money! Can't we just stop these pissing contests and agree that both OSes are good (and work very well together, might I add)?

Yes, but...
by itomato on Sat 9th Apr 2005 04:25 UTC

"No wonder Debian Linux 3.1 is taking so long to get release
the Debian team are doing other things."

Buh-dum-bum..

But they just dropped SPARC, ARM, MIPS, PA-RISC and S/390, and I think maybe M68K from stable.


...
by itomato on Sat 9th Apr 2005 04:29 UTC

I just thought of something...

This HURD, FreeBSD, $KERNEL release thing is the equivalent of maintaining multiple architectures in this world of approaching singularity...

It only makes sense to sharpen the 'architecture' mentality once the 'kernel' on 'narrower assortment of architectures' evolution kicks in..

Ahem!
by CRCampbell on Sat 9th Apr 2005 04:34 UTC

>>>Direct comment link re: what's wrong with FreeBSD
By Larry (IP: ---.de) - Posted on 2005-04-08 20:42:41
FSF/GNU extremists (some of them) don't like FreeBSD because the BSD license is more free than the GPL.<<<

Sounds like another LINSPIRE fanboy. Hrm... "GNU extremists". Freedom is such a terrible thing.

Re: Ahem!
by Gent on Sat 9th Apr 2005 04:38 UTC

>>>Direct comment link re: what's wrong with FreeBSD
By Larry (IP: ---.de) - Posted on 2005-04-08 20:42:41
FSF/GNU extremists (some of them) don't like FreeBSD because the BSD license is more free than the GPL.<<<

Sounds like another LINSPIRE fanboy. Hrm... "GNU extremists". Freedom is such a terrible thing.


It's like this:

BSDBush: These people are extremists. They're jealous of our freedom.

About Freedom...
by CRCampbell on Sat 9th Apr 2005 04:38 UTC

Free is free. Free is me being able to download a distro and apps for free, and not have to bend over for a licensing agreement. If you can't understand what freedom is, then you've been duped in this corporate culture of software patents and licenses.

Does this mean I'm not going to give anyone any money for their hardwork? By no means! I'm going to invest in the Free Software Foundation and the distro of my choice. Screw the companies that want to act like the rest of the proprietary herd.

@Gent
by CRCampbell on Sat 9th Apr 2005 04:40 UTC

Quit bringing Shrub into this. A total strawman on your part. ;)

Re: On the issue of License being "more free"
by Deletomn on Sat 9th Apr 2005 04:44 UTC

Gent: I think this is dependent on how you look at the term "free software." The way I see it, it's not meant to imply that the user is free to do what they will with the software, but that the software itself is free. As if it were a person -- the software is emancipated so to speak, no one can "control" it in any way.

Problem... The GPL only really affects the distribution of the software.

For example: I could "create my own version" of gcc and use it as I will and never return so much as a single line of source code to the community. If I gave the software to someone else however, I'm obligated to provide source code. However, the community still doesn't have it. As a result the "freedom" from the GPL really isn't that much "greater" than BSD.

v Re: @Gent
by Gent on Sat 9th Apr 2005 04:44 UTC
Addendum to a nun, he moos post
by Finalzone on Sat 9th Apr 2005 04:50 UTC


You're talking about linux, the OS that still has problems with fork() bombing.. FreeBSD won't crash with a fully loaded desktop (GNOME) and fork()ing it up to 8000 processes.


That argument was debunked by Alan Cox, Red Hat kernel maintainer on this post:
http://fcp.homelinux.org/modules/newbb/viewtopic.php?topic_id=12794...

Re: @h
by ylai on Sat 9th Apr 2005 04:54 UTC

h: You're talking about linux, the OS that still has problems with fork() bombing.. FreeBSD won't crash with a fully loaded desktop (GNOME) and fork()ing it up to 8000 processes..

This depends significantly on the kind of fork bomb. And maybe you should shed a light on the type you are talking about. A typical bash process, for example, has a 300 kB large (non-shared) memory footprint. 8000 processes means a bit larger than 2.3 GB. Even today, this is not kind of typical hardware configuration you can find. And requiring a system with less physical RAM to handle 8000 bash processes well is just a physical impossibility.

This is true for some distros' default settings, however it's trivial to fix this by limiting processes. On some distros (Debian, I think) the default is quite safe and the system is immune to such fork bombs.

The default is always a trouble since the ulimit/limits.conf depends significantly on the hardware and application it should run.

For example: I could "create my own version" of gcc and use it as I will and never return so much as a single line of source code to the community. If I gave the software to someone else however, I'm obligated to provide source code. However, the community still doesn't have it. As a result the "freedom" from the GPL really isn't that much "greater" than BSD.

I understand. Buth again, I think you're looking at it from an aspect of what the user can do. The GPL is written in a way to protect the software itself from control. It effectively gives the software "rights."

You can acquire a copy of that software, do what you will, use it as you will etc, but there is no way for you, under the GPL to exploit the software in and of itself for the generation of capital. At least not without giving back. This effectively abolishes the aspect of private property in the form of expropriated labor within the realm of GPLed "free software."

I've heard a lot of people argue how free software is not a communist ideal, and how the ability to charge proves it. But the software itself is actually quite fitting to what I'm talking about, as communism does not particularly seek to end simple money transactions (although it does), but more importantly seeks to change a social condition.

The GPL upholds this changed social condition where the BSD License does not. The software is "free software" in the same way we look at "free laborers/free workers."

This is what I'm saying. It's the difference between the outlook of capitalism, which is a freedom and right to private property, and the outlook of communism, which realizes the infringement on the freedom of the worker DUE to private property.

In this case you replace the worker with the software itself, so that the capital it generates may very well benefit a company or a person materially speaking, however, the labor itself is not exploited as in order to do this they will have to return that value to the community at large.

I don't want to threadjack this into a political discussion (as hard as that may be to believe). So let me just say plainly, that this is my outlook, most people who I have met willing to argue the similarities between the free software movement and communist ideals are very uneducated on the issue itself. So if you don't understand where I'm coming from, it could just be because of our completely differing points of view. If you do understand where I'm coming from, but simply disagree... it's your right to do so.

@ylai
by a nun, he moos on Sat 9th Apr 2005 05:41 UTC

The default is always a trouble since the ulimit/limits.conf depends significantly on the hardware and application it should run.

True, which is the argument given as a reason to not limit processes by some distros. Personally, I think a simple question in the installation process (i.e. is this going to be a Desktop/Workstation) could be used to provide better defaults depending on intended use.

RE: nona (IP: ---.w82-122.abo.wanadoo.fr)
by Duffman on Sat 9th Apr 2005 07:28 UTC

So, you are telling me that you better know the linux kernel than Linus ... OK

I think you must stop to be a linux zealot and accept that the linux kernel has some problems.

RE: By Adam (IP: ---.nap.wideopenwest.com)
by Duffman on Sat 9th Apr 2005 07:41 UTC

"Regardless 2.4 is still faster than Free/Net/Dragonfly/Open BSD.

I have nothing against BSD, but it is not GNU/Linux; it is far behind in terms of important functionality, stability, performance, and software support."

I think you haven't seen (or forgotten) the benchmark between BSD and linux done by the linux user fefe:
http://bulk.fefe.de/scalability/

In fact, FreeBSD and NetBSD scale better than linux. BTW, why at the beginning of this bench, when linux was winning, every linux users point to that site. But now that BSD community teach fefe how to configure BSD systems and that now BSD scales better, every linux users have forgotten this link ...

In fact, you have something against BSD, and telling that you are not won't make you an "open minded" linux user ...

linux is making the same that windows done in 80/90, it is good enough to be used and less expensive than UNIX (but it is changing), that's all ...

@All who are talking about super computer on linux
by Duffman on Sat 9th Apr 2005 07:54 UTC

Just tell me which distribution on linux this all computers are running. I am curious.
You haven't in mind that the linux running on this super computer has some proprietary patches that you will never get the hand on ?

make my day!
by Tobaccofarm on Sat 9th Apr 2005 08:16 UTC

Cool something to keep an eye on.

Re:@All who are talking about super computer on linux
by daijo on Sat 9th Apr 2005 09:04 UTC

"Just tell me which distribution on linux this all computers are running. I am curious.
You haven't in mind that the linux running on this super computer has some proprietary patches that you will never get the hand on ?"

Not now but soon...SOON...Muwuahahahaha!!!!

http://linux.slashdot.org/article.pl?sid=05/04/08/2148207&tid=98&ti...

RE: @ finalzone
by Vikram Sharma on Sat 9th Apr 2005 09:19 UTC

I forgot to mention that I had tried ext3 and reiserf son Ubuntu and found ReiserFS to be more faster, could be because Ubuntu has it that way. I am not a know it all in this matter but one of my collegues was arguing that ReiserFS is nothing but ext3, had to show it to him. People still prefer ext3, is more acceptable for some reason.

RE: Re:@All who are talking about super computer on linux
by Duffman on Sat 9th Apr 2005 09:35 UTC

You know, enterprise doesn't care about your freedom, but they care about their money and the control of their IT. If the GPL 3.0 is coming soon, they will:

1) fork linux and use a version that come with GPL 2.0 and proprietary patch
and/or
2) choose another system (BSD system for example or invest in their own unix system if they own so)

"You know, enterprise doesn't care about your freedom, but they care about their money and the control of their IT. If the GPL 3.0 is coming soon, they will:

1) fork linux and use a version that come with GPL 2.0 and proprietary patch
and/or
2) choose another system (BSD system for example or invest in their own unix system if they own so)"

And you Sir obviously don´t care much for having fun:) At least your nick suggests you have a sense of humor. Perhaps ComicStoreGuy would be more appropriate. "Worst post ever!"

What the...
by William on Sat 9th Apr 2005 10:34 UTC

I once had an idea of various kernels serving the same set of apps (of course the kernels can only come in one at a time) but they seem to have beaten me to it... ;)

Debian...
by debilian on Sat 9th Apr 2005 10:45 UTC

is more than just next linux distro!!!
Pls just accept that fact!

v RE: Duffman (IP: ---.fbx.proxad.net)
by Anonymous on Sat 9th Apr 2005 10:49 UTC
v I just came to see all the little trolls.
by Mike on Sat 9th Apr 2005 11:10 UTC
Re: I just came to see all the little trolls.
by Olli on Sat 9th Apr 2005 11:18 UTC

Well the news is that it has now officially entered the unstable branch, whereas it was more of side project before. This means that it is for developers usuable to the degree that they can built packages for it, and start optimizing the system. In Debian terms this will most probably mean that a stable release is still a few years away.

BTW it's depressing to see how many people seem to have a problem with BSD as a whole. For me this is real exciting news. As where the BSD usually come with BSD utils Debian GNU/kFreeBSD will come with the full gnu userland. I grew up with Linux and I found the gnu tools compared to the BSD tools more featureful and well, I'm used to them. Next to that I like the binary package system of Debian. I run a quite old server and building from ports simply takes too long and the binary package management in OpenBSD and FreeBSD does not really have priority there (which is understandabgle).

All in all good going team!!

RE: anonymous
by Duffman on Sat 9th Apr 2005 11:44 UTC

"You are so stupid it is not funny. Why would you bother posting such a blathering load of crap?

You have really no idea about Linux, licensing, or the GPL, do you? "

Have you seen any insults in my post ?
You know, those who are aggressive are always those who are wrong.

Actually I think GPL is good for avoiding enterprise to steal code from open source, but it is my opinion and when i see my boss, I can't tell him to take this because of an obscure licensing that he doesn't care about, that's the fact.

Also, I have beginning to learn UNIX with linux, so yes, I think I have ideas about linux and GPL, perhaps more than the majority of linux users who said linux instead of GNU/Linux.

RE: By daijo (IP: ---.bredband.comhem.se)
by Duffman on Sat 9th Apr 2005 11:47 UTC

Have a sit and take a Duff ;-)

GPL the whole freebsd OS.
by Han Solo on Sat 9th Apr 2005 11:47 UTC


I saw take a copy of FreeBSD and re-release the whole thing as GPL and we all start working on it from there on as GPL software.


Then we have another strong GPL'ed OS too.

This is cool
by CdBee on Sat 9th Apr 2005 12:29 UTC

I played with the FreeSBIE BSD liveCD when it was issued and loved it, but the affair went no further due to the relative ease of finding Linux apps for my 2nd OS (I'm still primarily a Win32 user due to certain favourite apps)

The thought of using Synaptic on FreeBSD to install applications is a very tempting one.

I guess it'll be an install-time option though as Apt-get install FreeBSD would require all supporting libraries to be replaced with ones compiled for BSD instead of Linux.. and probably changes in the folder structure as well.

RE: Daijo
by Adam on Sat 9th Apr 2005 12:30 UTC


Just tell me which distribution on linux this all computers are running. I am curious.


I know IBM Blue Gene/L is running SUSE on its controlling nodes.


In fact, FreeBSD and NetBSD scale better than linux. BTW, why at the beginning of this bench, when linux was winning, every linux users point to that site


No they don't. Linux still beats out *BSD on the majority of tests. This does not really test scalability because it is testing one processor on a 256mb machine versus a quad processor or something with more than 5 gigs of ram.

Linux is a test kernel anyway, it is not even a release.

re: RE: h
by h on Sat 9th Apr 2005 12:32 UTC

Trolling? People defend linux and they are ok, people defend *BSD and they are trolls ?

re: Linux
by h on Sat 9th Apr 2005 12:36 UTC

""
Last time I checked, it was Solaris 8, HP-UX and AIX running on super computers, not Linux.
"

From the Top500 supercomputers list

NUMBER 1
IBM's BlueGene/L Supercomputer -- runs Linux

NUMBER 2
SGI's Columbia Supercomputer -- runs Linux

NUMBER 3
Earth Simulator -- runs Enhanced NEC Unix

NUMBER 4
IBM's MareNostrum -- runs Linux

NUMBER 5
IBM's (I believe) Thunder -- runs Linux

So, out of the top 5 spots, Linux holds 4...but wait, there's more. It was recently estimated that Linux holds 300 spots out of the top 500 for Supercomputers. So, if you combine AIX, Solaris and HP-UX, then they almost compare to Linux, but not quite.

See how much more believable you sound when you do a little research instead of saying "last time I checked"."



The official Top 500 list, available at www.top500.org, does not contain operating system statistics. But the scientists who maintain the list privately keep track of what operating systems are running on the computers, says Hans Werner Meuer, a computer science professor at the University of Mannheim, in Germany, who has been publishing the Top 500 list twice a year since 1993.


See how much unbelievable you sound when you state the list of supercomputers is mostly powered by linux, when there isn't an accurate list of the operating systems?

v RE: Duffman (IP: ---.fbx.proxad.net)
by Anonymous on Sat 9th Apr 2005 13:08 UTC
v RE: Duffman (IP: ---.fbx.proxad.net)
by Anonymous on Sat 9th Apr 2005 13:12 UTC
RE: Duffman
by h on Sat 9th Apr 2005 13:32 UTC

"It is from a top FreeBSD kernel developer. He is trying to get this 12CPU system to scale as much as possible on a parallel compile benchmark.

Now you probably don't know this, but parallel compiles are one of the most trivial workloads possible with regard to scalability.

FreeBSD wasn't quite able to make use of the equivalent of 4 CPUs doing real work on this test. It achieved about 32% scalability, which is pretty horrible."

No one said FreeBSD is SMP-ready yet. It isn't, and the more CPUs you add the more catastrophic it becomes.

Linux clearly wins in that field and it would be foolish to deny that.

But as I've said in another post, wait some more time until the hard work on the RELENG_5 is done, and you'll be surprised.

RE: h (IP: ---.netcabo.net)
by Anonymous on Sat 9th Apr 2005 13:32 UTC

Trolling? People defend linux and they are ok, people defend *BSD and they are trolls ?

No.

People blatantly lie about *BSD and Linux in inflammatory ways and they are trolls.

RE: h (IP: ---.netcabo.pt)
by Anonymous on Sat 9th Apr 2005 13:36 UTC

No one said FreeBSD is SMP-ready yet. It isn't, and the more CPUs you add the more catastrophic it becomes.

Yes, actually someone did. Duffman said FreeBSD scales better than Linux. Linux *is* SMP-ready. The logical conclusion is that FreeBSD is SMP-ready. But he is a troll.

Linux clearly wins in that field and it would be foolish to deny that.

Yep.

But as I've said in another post, wait some more time until the hard work on the RELENG_5 is done, and you'll be surprised.

Yes I have been waiting.

RE: By Anonymous (IP: ---.dyn.iinet.net.au)
by Duffman on Sat 9th Apr 2005 13:49 UTC

"Yes, actually someone did. Duffman said FreeBSD scales better than Linux. Linux *is* SMP-ready. The logical conclusion is that FreeBSD is SMP-ready. But he is a troll."

you suppose that alone.. If you have seen the link that have done, the test was done on one CPU system.

RE: By Anonymous (IP: ---.dyn.iinet.net.au)
by Duffman on Sat 9th Apr 2005 13:50 UTC

"Yes. It is insulting to the intelligence of anyone who reads it. "

In fact, I think it is insulting for linux zealots ...

RE: By Adam (IP: ---.nap.wideopenwest.com)
by Duffman on Sat 9th Apr 2005 13:54 UTC

"I know IBM Blue Gene/L is running SUSE on its controlling nodes."
How do you know? (link ?)
what is running on all of the computers outside the controller nodes ?

v RE: Dufftroll
by Anonymous on Sat 9th Apr 2005 14:15 UTC
RE: Duffman (IP: ---.fbx.proxad.net)
by Anonymous on Sat 9th Apr 2005 14:25 UTC

"I know IBM Blue Gene/L is running SUSE on its controlling nodes."
How do you know? (link ?)


How old are you? 13? Do you know how to use the internet?

No? OK, I'll give you some help. Go to www.google.com and type this:
"blue gene" suse

Then click "I'm Feeling Lucky".

what is running on all of the computers outside the controller nodes ?

A custom OS from IBM.

v RE: Duffman (IP: ---.fbx.proxad.net)
by Anonymous on Sat 9th Apr 2005 14:34 UTC
RE: By Anonymous (IP: ---.dyn.iinet.net.au)
by Duffman on Sat 9th Apr 2005 15:04 UTC

"A custom OS from IBM. "

OK that's all.

"What? What link? Both the ones I posted were most definitely run on SMP systems. You are remarkably stupid, even for a BSD troll. "

The link I post ...

I also note that any linux user is not able to speak without insulting the other one...

In fact, you have also assuming that I was a BSD user, but i am not. I run a lots of differents UNIX systems (not FreeBSD as you think).

I think that you must just turn off you computer and have a rest, because you seems to be very angry ...

v RE: Duffman (IP: ---.fbx.proxad.net)
by Anonymous on Sat 9th Apr 2005 15:36 UTC
jeez..
by helf on Sat 9th Apr 2005 15:43 UTC

you people are pathetic. Turning and article on debian adding an unstable branch that uses the freebsd kernel in to flamewars on gpl vs bsd, solaris vs linux, *bsd vs linux, ufs vs ext3 vs reiser etc etc..

...bunch of hypocritical losers.

RE: By Anonymous (IP: ---.dyn.iinet.net.au)
by Duffman on Sat 9th Apr 2005 15:49 UTC

It is funny to see that when someone talk about linux, it seems to hurt you personnally. Are you a part of the linux kernel ?

@Kors
by Nathan Hawkins on Sat 9th Apr 2005 19:12 UTC

">It also shows how mature and portal the GNU userland needs >to be for this to work.
That's why the initial idea of debian team - use bsd's libc + kernel with gnu software failed misreably. Some parts of gnu userland turned out to be unportable, due to lack of development culture, and loads of linuxisms and glibcisms in the sources... And anyone who ever tried to compile auto*, configure, etc. oriented programs on platform with non-gnu userland/kernel/compiler/includes/libraries knows that it's going to be a major PITA next time too... Because one needs "right" awk, "right" cc, "right" ld and "right" everything. Where "right" is NOT standards (I mean posix/sus) compatible, but is GNU. The destruction of a single positive result of unix wars (standards) is in progress, due to gnu/linux crowd with their "mature" and "portal" software."


How about a few facts.

Actually, the initial effort to port Debian to FreeBSD didn't fail for technical reasons. I know, I did it. The problem was that personal difficulties forced me off the net. I still believe the move to glibc was a mistake -- I had tried it first, and decided against it, before Robert Millan started porting to it. I don't believe glibc on FreeBSD will ever become as complete and stable as glibc on Linux or the native libc on FreeBSD. There simply aren't enough people working on it. For a couple years, no-one was.

The GNU programs didn't have serious portability problems. The major headaches were with certain critical Debian packages that depend on Linux or glibc. util-linux, debianutils, dpkg, sysvinit -- packages like that. None of those are GNU, and all of them were doing very system-specific things.

3 year old debs + a BSD kernel is great for stability !
by Anonymous on Sat 9th Apr 2005 20:48 UTC

As we all know, the FreeBSD project and the Debian project have different approaches to managing developers. In FreeBSD world, just because you maintain a port, it doesn't mean you're the boss. We have all seen where the model of a steering committee of looong time Unix developers plus a comitter status leads to: an up-to-date Unix system both in terms of userland sotware and in terms of kernel and networking technologies.

In Debian, however, if you've withstood 1+ year of being tortured with questions, you achieve "developer" status. This means you don't even have to know what **p means, but you get to vote on the project! You become a mighty Debian developer! It's a democracy! It means you get to stall a project! It means you get to argue on equal terms with real programmers! You package debs and argue endlessly, and you don't have to know chicken shit about protocols or any other hardcore Unix stuff. Doesn't it sound great ?! You get to walk around with cool red-spiral T-shirts, too!

What is great about this project is that we now can have Debian maintainers keep the packages late 2+ years behind schedule! This means all the advantages of and up-to-date stable Unix distro like FreeBSD are gone! Now we can have outdated packages software with a BSD kernel! Great for stability!

Debian has shown us the way to stall an open source project, and I think this is fundamental development in the open source world. No more ports update! 3 year old debs for the people!

@duffman
by a nun, he moos on Sat 9th Apr 2005 20:49 UTC

You haven't in mind that the linux running on this super computer has some proprietary patches that you will never get the hand on ?

And what's the relevance of this hypothesis? I don't see any.

1) fork linux and use a version that come with GPL 2.0 and proprietary patch

That wouldn't be legal, at least if they re-release it.

You don't seem to understand the GPL very well...maybe you should lay off the Duff a little.

@duffman
by a nun, he moos on Sat 9th Apr 2005 20:57 UTC

I also note that any linux user is not able to speak without insulting the other one...

You shouldn't made statements such as these that can so easily be proven wrong. I am a Linux user. I speak (well, write, actually) without insulting those I don't agree with them. If all they do is troll and flamebaite, I will point it out, and if I'm insulted myself I might take a jab in return, but as much as I can I try to keep my posts clean.

These facts considered, your assertion that "any linux user is not able to speak without insulting the other one" can therefore easily proven to be false.

algorithmic and parallel scalability
by Yoke on Sat 9th Apr 2005 23:37 UTC

As always in discussions regarding scalability on OSnews, a lot of people conflate algorithmic and parallel scalability.

The frequently referred to fefe benchmark tests algorithmic scalability.

Algorithmic scalability is the easy part. It usually involves code that is localized to one particular subsystem, and the changes made does not have system-wide ramifications. It's just doing the same old thing more efficiently.

An excellent illustration of this fact is that when the fefe benchmarks was first published, NetBSD didn't do too well. They rectified that in a matter of weeks.


Parallel scalability is hard and requires a long-term system-wide effort, and some of the time it's hard to know if you're going in the right direction.

When the FreeBSD project branched 5.x over four (or five?) years ago, they started working on improving SMP (the SMPng project) and multithreading (the KSE project), and those two projects are the main causes for why a production ready 5.x has been so long in coming. And even today, linuxthreads, a FreeBSD port of the old Linux userland threading library, is still the best alternative in FreeBSD for certain applications.


The Linux camp started working on SMP much earlier than any of the BSDs, and has also had to some major corporations (most notably SGI) putting their effort into it and as a result is way ahead of its open source competitors in that area and is considered a viable alternative to Solaris.


So again, algorithmic scalability is low hanging fruit (esp. when you can look at open source systems that has already done the right thing), and is only really interesting when it's done poorly. Parallel scalability is hard, and requires constant vigilance to ensure that you're not moving in the wrong direction, something the Linux folks recently discovered when they found there had been regressions both from 2.4 and within the 2.6 series running a particular benchmark on SMP machines (the regression in question was benchmarked on a four-way Intel Xeon).

RE: Yoke (IP: ---.adsl.no)
by Anonymous on Sun 10th Apr 2005 03:09 UTC

I agree with you mostly, but just a couple of small points.

The Linux camp started working on SMP much earlier than any of the BSDs,

Actually FreeBSD devs used to boast about having a working SMP implementation before Linux.

and has also had to some major corporations (most notably SGI)

In Linux 2.5/6, the biggest scalability pieces came from SUSE and RedHat (per-CPU scheduler, per-device block layer, lightweight rmap VM). IBM was another big contender, with a lot of RCU work. I forget exactly who did the per-inode radix tree pagecache. I think Andrew Morton was the man behind per-CPU page lists, per-zone LRU lists (and he would have been probably at Digeo at that time). Not sure who did the other pieces of the NUMA memory allocator work. Probably much of it would have been IBM.

SGI has contributed *relatively* little, apart from smaller fixes to odd things that blow up at 512 CPUs but nobody else has ever seen... although it looks like they are gearing up - their main development target is now the stock 2.6 kernel rather than their own propack.

Parallel scalability is hard, and requires constant vigilance to ensure that you're not moving in the wrong direction, something the Linux folks recently discovered when they found there had been regressions both from 2.4 and within the 2.6 series running a particular benchmark on SMP machines (the regression in question was benchmarked on a four-way Intel Xeon).

4 way Itanium 2...

This was actually RedHat's enterprise 2.4 kernel, which looks a lot more like 2.6 (trust me, stock 2.4 could *not* drive hundreds of disks efficiently in SMP).

The regression wasn't a scalability problem per-se, just that a single tuning parameter for the multiprocessor scheduler had not been set correctly for the Itanium's huge caches. Fixing this solved the problem.

But you are right that it takes work to keep regressions away.

RE: Just take the FreeBSD kernel and re-release as GPL.
by Perez-Gilaberte on Sun 10th Apr 2005 11:16 UTC

No, apparently you know nothing about how Copyright works. You cannot relicense code you didn't write! If you write a derivative work, then you can license it under whatever license you want as long as the original license permits that, but the original code is still licensed under the original license.

About this project, I honestly think is a total waste of time. They only have a couple of guys working on it. When they encounter a problem and the kernel panics nobody is going to help them because their running this abomination. One of FreeBSD's strengths is that it's a deterministic system, there are no random combinations of this kernel and that userland revisions. It's funny how the BSDs remove more and more GNU userland and replace it with BSD code and these Debian guys do the opposite ;)

RE: FreeBSD propaganda
by N.N. on Sun 10th Apr 2005 11:17 UTC

The point about Windows support is not true. There are free newsgroups that deal with windows issues too. Saying that you can only get support for Windows by calling Microsoft is like saying you can only get support for Linux by calling Red Hat.

@By a nun, he moos (IP: ---.53-203-24.mc.videotron.ca)
by Russian Guy on Sun 10th Apr 2005 14:11 UTC

"I am a Linux user. I speak (well, write, actually) without insulting those I don't agree with them."

You shouldn't made statements such as these that can so easily be proven wrong. Should I repeat your insults directed at me?

"These facts considered, your assertion that "any linux user is not able to speak without insulting the other one" can therefore easily proven to be false."

Not that easily. You are an example of a Linux user insulting people like me, who disagrees with you. Should I repeat your insluts directed at me?

Also, I occasionally use Linux, too, so that makes me a Linux user. Try to claim that I insluted you first, and you'll prove the statement that "any linux user is not able to speak without insulting the other one.":)

RE: Anonymous (IP: ---.dyn.iinet.net.au)
by Yoke on Sun 10th Apr 2005 16:45 UTC

Thanks for your corrections. Reassuring to hear that the supposed scalability regression was easily solved.

@Russian Guy
by a nun, he moos on Sun 10th Apr 2005 17:28 UTC

You shouldn't made statements such as these that can so easily be proven wrong. Should I repeat your insults directed at me?

Oh, please. I did claim you were trolling and flaimbaiting a couple of times, but that's because, well, you were. That wasn't an insult, it was simply stating the truth. Just like stating that you seem to be a very thin-skinned person to pop up and bring this up again weeks after the fact is not an insult.

Not that easily. You are an example of a Linux user insulting people like me, who disagrees with you. Should I repeat your insluts directed at me?

Yes, please, provide us with a link to the last time I "insulted" you, so we can all put it back in context. But I warn you, I won't let you twist the words in order to prove your point (that's not an insult, btw - I suggest you read up on the dictionary definition of the word).

Also, I occasionally use Linux, too, so that makes me a Linux user. Try to claim that I insluted you first, and you'll prove the statement that "any linux user is not able to speak without insulting the other one.":)

No I won't, because "any linux user" means "all linux users". Even if you did insult me first (which you probably did), that would prove nothing. You can't make a rule based on such a small sampling, any basic course in Statistics will tell you that.

Also, when duffman wrote "linux user", he clearly meant "linux enthusiast", not casual once-in-a-while users. That much can easily be deducted from his posts, unless one is only interested in taking things out of context and playing on words to prove a point. You may technically be a "linux user", but you're definitely not a linux advocate, and those were the people targeted by duffman's criticism.

Meanwhile, I guess I'm a Windows user since I use Windows every work day...

Many of you have it backwards
by Anonymous on Mon 11th Apr 2005 14:00 UTC

Why would a FreeBSD user want to use the GNU userland? Ugh!