Slashdot Mirror
Debian FreeBSD Distro?
antigen fiend wrote in to note that Debian Weekly News has a bit about a recent debate about a Debian port to the FreeBSD Kernel. There are several comments relating to licensing, ease of porting, and other relevant topics. Its definitely an interesting project, with a lot of political and technical snags. Any thoughts?
I use both debian and FreeBSD,
Debian for it's excellent desktop/workstation capabilities with their wonderful autoupdating website/dependencies thing and the total hackability of it all. The packages always work, and their dependencies aren't all funny like they often are with SuSE and RedHat,
However I use FreeBSD for my servers, yes it's not the easiest of distros and it's very hard to get anything real working (for a relative newbie like me anyway), but it works, very well in fact. It never ever crashes and it servers my websites just fine. There are very few bugs very few updates, and I can just leave it alone totally.
To combine both? Well - heaven - what else can it be. Add to it KDE 2.0 and you have the complete MS competitor for the desktop/low end server market.
Bring it on!!
If I understand the BSD license correctly, you are allowed to release FreeBSD under the terms of the GPL.
It would give you some enemies for life though..
Sure this is an interesting project, but I'm a bit doubtful concerning the impact this will have. Currently the Linux distros, as different as they are, all use the same kernel. Given that the issue of compatability/fragmentation is a real concern to any potential corporate (Linux) adopters, I'm wondering wether this won't just turn out to be food for MSs propaganda machine ... after all we're taking Linux software and porting it to a different (yet still Unix based) kernel ... technically though, this is probably good news as it will uncover some new bugs and make the software more portable ...
The paragraph in DWN mentions the concern that it would open the doors for unscrupulous 3d parties to take their work and sell closed-source derivatives, but I don't see that as a real problem ... the FreeBSD kernel would have the BSD kernel, but much (most?) of the software that the Debian team puts in the DebianBSD distribution would still be GPL, which means FastBuck Inc. would not be able to take DebianBSD as-is and apply a closed-source license.
And now a question for the License Lawyers.
Christopher A. Bohn
cb
Oooh! What does this button do!?
Getting the libraries ported would make supporting applications on both systems a dream. Also, the FreeBSD group could then take advantage of the ported glibc to increase interoperability.
We talk about freedom, yet all the major Free Software vendors are using the same kernel. If Debian pushed GNU/Linux, GNU/Hurd, and now GNU/FreeBSD, then we will eventually have the day that we can choose which kernel is the best for our purposes.
Few people use Linux for the kernel, they use it for the capabilities. I have a friend who loves his GNU/Solaris machine. Don't laugh, he administered Solaris machines for years, so his home PC has the Solaris x86 stuff on it, but all it runs is Free Software (and StarOffice).
Linux advocacy is silly. Advocating *nix, or even Free Software makes sense, but obsession with a kernel is silly. As Linus says, (I might be paraphrasing) "Linux sucks, it just sucks less than anything else out there."
The Linux kernel is fast, it is stable, but it isn't revolutionary. As I understand it, it is Microkernel-esque, although it still probably has remnants from its days as a monolithic kernel. I mean, when someone comes out with a solid exokernel, are we going to scream and yell about how Linux is still better?
Support free software, but support choice. In a free software world, we could pick our kernel without worrying about our apps breaking. This project has both technical and political merit. Although, if the BSD license allows you to do whatever you want with the code, can Debian release the GNU/FreeBSD system under the GPL? If they made the best FreeBSD distro and put it under the GPL, the license issues would be over. I'm not sure of the specifics of the BSD license, but if you can include it in proprietary licensed systems, I don't see why Debian's system couldn't be GPLed.
First, Debian is also being ported to the HURD, which I am eagerly awaiting.
On a more general note, the more people that do porting projects like this, the more pluggable the whole system will get. Wouldn't it be great if you could select your kernel, libc, and packages independent of each other? This is what UNIX is about... choice.
It would also be great to have glibc ported to FreeBSD. I sure hope that is the route they decide to take. This would allow FreeBSDers to have better access to new applications (although I'm not sure if they care - BSD is usually about stability, not newness). Also, given the user base of Debian, this would give more competition to Linux, and give us a reason to still be competitive.
Finally, this will expose more bugs in all systems, which is always a good thing.
Engineering and the Ultimate
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
1. Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
3. All advertising materials mentioning features or use of this software
must display the following acknowledgement:
This product includes software developed by the University of
California, Berkeley and its contributors.
I guess they retained the original copyright, at least for the 3.3-RELEASE.
Intosi
Intosi
Sure, as a desktop FreeBSD user this would make me happy, as it would boost FreeBSD acceptance. However, I'd rather see HURD ready first (this is one of Debian's projects right now). It seems like a very interesting and worthwhile alternative to current kernels, and might show the world that the OSS movement can produce a true, quality microkernel-based system.
This "project" is currently not official. It is being worked on by a few developers that are doing this without the support of Debian as a whole. By readin ghte lists it is very hard to tell wether it's going to be official or not.
1. dpkg, *loads* of maintainers
2. Varies, but with a bit of work you can port most things
3. Nothing, other than their working methods. Debian's approach is highly parallel.
4. It would be hard. dpkg is GPL'd.
-Yarn - Rio Karma: Excellent
Debian is the best linux dist imho.
FreeBSD is the best x86 os imho.
KDE 2.0 looks _very_ promising.
And them together sounds lika a real threat to MS Win* as the desktop alternative.
The 4 s:
Speed, security, stability and style.
Its my lucky day!
Lost Carrier
Lost Carrier
http://www.geekboys.org
my feeling is that Debian/FreeBSD should base itself on glibc2.1, i.e port glibc to the FreeBSD kernel, and compile and run the whole distribution on that, *then* also include a copy of FreeBSD's libc, to be able to run regular FreeBSD programs (just like RH6 can run libc5 programs too). otherwise, Debian/FreeBSD is nothing more than FreeBSD with a bunch of GNU tools added and the installation procedure re-tooled.
OTOH, I wouldn't commit myself to a project like this for the long-term; it appeals to me more for the hack value, and for the political implications (hopefully the good ones, of showing the various factions how easily software can work together) than for actual usefulness.
I mean seriously - why do we need it? The FreeBSD/OpenBSD/NetBSD systems are just as painstakingly careful with their release-system as the Debian people, so it's not like we'd be getting a more mature development cycle, right?
On the other hand, there really aren't any "distros" to speak of with the BSDs. I wonder why... It would seem that a company willing to put a bit of effort into ease-of-use could become the RedHat of the BSD world. In fact, they could probably go head-to-head with RedHat in terms of OS sales. Doesn't everyone here always say "competition is good"?
-----------
"You can't shake the Devil's hand and say you're only kidding."
A significant merit to adding FreeBSD to the mix is that this makes Debian less and less kernel-dependent. In the long run, that makes it more and more possible for Debian to support more "UNIX variants."
Interesting, in the longer run, would be:
Thus, support for FreeBSD tomorrow may help there to be support for more unusual OS selections a couple years from now.
That seems to me to be a Good Thing.
If you're not part of the solution, you're part of the precipitate.
(For the curious, I am a FreeBSD user)
:).
There are several reasons I do not think DebianBSD is going to be advantageous:
1.) Most people using FreeBSD enjoy the less restrictive, more open BSD license.
2.) The FreeBSD Ports Collection is truly incredible; there little or no need or desire for a Debian package system.
3.) I personally love the fact that FreeBSD is *not* a distribution, it is a full featured, integrated Operating System. I think this allows FreeBSD to be more intelligently organized and to work more smoothly than the various Linux distributions forced to integrate an operating system beyond their control with software largely beyond their control. DebianBSD sacrifies this advantage.
4.) The licensing issues are difficult, if not impossible to resolve.
5.) Any variation on the name GNU/BSD is going to cause an uproar and permenantly doom any Debian based FreeBSD distribution.
6.) Finally, from quickly scanning the Debian mailing lists, it seems as if most of the Debian developers have no respect for FreeBSD. One called it "dying software" and others claimed it offered no advantages over Linux. While everyone is entitled to an opinion, however ill-informed and erroneous, I wonder how dedicated Debian could be to an operating system it does not like and does not respect; after all, part of the allure of working on open source software is being able to code for your own pleasure rather than someone else's.
In short, DebianBSD seems like an abortion from the start, althougth I am prepared to be surprised.
(Note, even if DebianBSD did come into existence I would never switch; Debian and the FSF irritate me with their holier-than-art-thou pronouncements on freedom. Sadly enough for DebianBSD, I have a feeling a large number of FreeBSD users will agree with me.)
This is better than the opposite observation that I was going to point out, which is that there are components of Debian, such as Perl, Python, and XFree86 that already use non-GPL-like licenses.
If you're not part of the solution, you're part of the precipitate.
Interesting... I feel completely the opposite. I found the RH init scripts infinitely more logical and easier to admin than Debian's. In particular, network configuration was miles better in Red Hat with all the relevant details for each interface in a config file in /etc/sysconfig, rather than hardcoded into an rc.d script. Still, that's one of the great things about Linux (and indeed, free software in general). Everyone is free to use what they're most comfortable with.
Disclaimer: I haven't looked at Debian since 1.3, so things may have improved since then. Also, much as I like Red Hat, I freely admit it's not perfect.
"The invisible and the non-existent look very much alike." -- Delos B. McKown
> Debian GNU/NT, anyone? ;-)
There was such a project, basically it was to be a Debian format distribution of Cygwin. It would have been very useful, given that Cygnus doesn't put in much work on making free distributions, but nothing seems to have come from it. Nobody volunteered to do actual work.
There are ports of the X libraries for cygwin, and there are gratis X servers that run under NT.
I haven't bothered with them, as Cygwin 1.0 is close to a perfect fit for my needs. It gives me a nice Unix programming environment under NT for producing win32 executables, and since XEmacs runs natively under win32 (and is included on the CD from Cygnus), I cannot think of any X11 programs I miss.
Debian is one of the last Linux distros which
are not yet share holder driven (watch out for SuSE).
It is IMHO a sad fact that the Debian development
model seems slower than it would be good for Debian.
I think that Debian's receipe contains too much
democracy and not enough (benevolent) dictatorship.
Embracing additional kernels can be a nice thing
if your development model is fast and you are
among those who march in front.
Splitting up forces further when you seem to
be falling behind is IMHO the wrong thing to do.
Debian first should do some streamlining of their
development model before increasing entropy
further.
Don't get me wrong: it might be a big plus 'for
the whole thing', if different exchangable kernels
are competing against each other. But increasing
complexity might also be another nail in the
coffin if your development system is too slow to
handle it.
Personally, I like the BSD license much more as compared to the GPL, but I'm not a programmer. You just need to make a decision of what is more important to you:
1 - Having thousands of people potentially debugging your code for you.
2 - Having complete control over the modification and distribution of your work.
I think that GPL is great for programmers, but I can't see how it can make great business sense, since it allows everyone in the world to potentially be your competitor with no cash or time outlay. Maybe Redhat will prove me wrong... Time will tell, I guess.
Just from this thread, it sounds as if there is enough support for Debian-FreeBSD to make it not DOA.
The idea of it all is very interesting, no matter the pros and cons of it.
FreeBSD and Linux users needs to wake up and learn to respect each other. Most of us know that competition is good, and FreeBSD and Linux are in direct competition in many areas.
If the BSD community does not want GPL, I can respect that. GPL is not some Holy Grail of licensing as many of you make it out to be. I believe things can be TOO open. Why? It is because things become chaotic and fragmented, slowing progression and causing all sorts of problems. You are going to see a lot of this in the Linux community over the next couple of years. (It is going to happen because commercial distros are going to try to get ahead by breaking away from the others. The lag time for others to pick up any advantages to their ideas is also enough to cause problems.)
Besides, for the vast majority of us, the license does not matter anyways. We just install it and use it. We are not modifying the libraries, etc. etc.
Debian-FreeBSD is a wake up call to us all. It's time to start repecting our differences. We are all individuals, not members of cults.
E
EverCode
Sure, there's a BSD distro. It's well supported, solid, comes packaged for ISPs and Servers. The fine people at BSDI sell it.
I'm beginning to think the reason there's no inexpensive, mass-market distro for *BSD is because of the license. You just can't stop the suits from closing off important pieces, and jacking up the price to what the the market will bear because they can .
Look at Cygnus. They started out as a business to offer support for Open Source "products". Now, they sell closed source products as well. Why the change in business plans when, according to Cygnus, they've always enjoyed phenomenal growth and profitability? I think it's because the suits just can't stand the thought of giving everything away.
This is what's really new about running a business distributing GPL'd code. The suits have to start rethinking where they can make the money.
I'm not sure there is a lot of money to be made, long term, in being a distro vendor. A lot of people will just buy the same thing from Cheap Bytes. Ultimately, I would think that the distro vendors would get more involved in support, custom projects and training. I think RedHat knows this and this is why the Cygnus purchase makes sense.
The advantages were that great, right?
Like, um, errr, um, nope I'm stuck already.
Oh wait, I remember, *BSD has more crazy fanatics, so you can feel more at home. And if someone asks "How do I delete a file?" you can call him lame, and tell him to buy a clue. The microkernel dig above from a supposed *BSD fan is hilarious "Our kernel has less features, and is therefore inherently better".
The *BSD kernels each have a lesson or two to teach Linux, but equally there
are lessons they could learn from Linux, and from each other. It's not enough
to be worth fighting over. IMHO today's *BSD userspace is poorer than the
GNU userspace, and Linux is a better platform for bleeding edge kernel
meddling.
I respect the OpenBSD crew for having the guys to actually audit code, and the adventurers doing "First Ports" to platforms like the old 68K Macs, but the trend towards fanatical hate of other Free Software projects leaves a very unpleasant taste, BSD users are pronouncing that a GPL'd branch is "Evil", while a MS proprietary branch is "Proof that BSD is truely free".
This inconsistent value set (and I don't clame it's present inside the holy temple of the core developers) is a rotten stinking thing inside the BSD community, and makes the "Anything but MS" sentiment found in some parts of the Linux camp seem healthy by comparison. While that rotten thing remains in the community I hold no hope of GNU/FreeBSD seeing the light of day.
Nick.
Sure, but let's remember that all BSD's forked from a common code. 'Distrobutions' could eventually end up going the same way as BSD if Linux does eventually fork for any particular reason.
And programs under one BSD aren't promised to work on ALL BSD's..
-- I'm the root of all that's evil, but you can call me cookie..
Where I think there might be an issue is that the BSD world does not have this idea that the operating system is the kernel, but rather the operating system is the whole distribution. A debian distribution based on the FreeBSD kernel is not the Debian distribution of FreeBSD but a completely different branch, which one might call DebianBSD.
I think there are advantages to the BSD way of looking at things: security issues around OSs are not issues of the kernel, but issues about the whole system as deployed. I would rather the BSD vision of the operating system was not buried because it is not the same as the Linux vision.
FreeBSD does have a centralised package management system, Jordan_hubbard's pkg_add system.
I agree with you as far as the BSD license thing goes; I prefer it (and I am a programmer). I like its wide-open nature.
However,
I think that GPL is great for programmers, but I can't see how it can make great business sense
The GPL, IMO, sucks for programmers, because it explicitly limits their right to steal source (note: this is not a criticism of the GPL). The GPL, believe it or not, is actually best for marketers and businesses -- marketers who want to get 'into' the free software movement (good for them!) and businesses who want to get the debugging advantages of open source but who don't want to allow their competitors to use their source to build an exclusive competitive advantage.
What, you ask, about the programmers who like GPL? The answer is simple: we're also marketers. We want to position our product not so that it serves the largest number of needs, but rather so that it serves a long-term goal, the establishment of open software.
A lot of us, BTW, are seriously decieved about the purpose of the GPL -- some even say it's more free than the other licences. It's not, it's merely more practical than the other free licenses, since the other free licenses depend on the community to keep a product free, while the GPL gives the illusory impression that even if the community disappears the software will still remain free.
Foolishness.
-Billy
What I meant about being great for programmers would be better phrased as "for programming". In that, it facilitates programming because everyone can get their hands into the source code. But yes, it definetly seems to me to be extremely restrictive, in that you really don't have any control over your code once it's released. BSD allows you really, it seems, to do whatever you please, so long as you leave the prior copyright notices visible. If you want to distribute a modified work, you have to give credit to the people that originated it, but you don't need to release their source or your changes.
So far as the GPL goes... I think it still is really theoretical in nature. When/If someone violates it and is found to have done so and is required to pay penalties (monetary or otherwise), then it'll actually hold more weight. Until that time, it really seems like an idealism transcribed into a pseudo-software license.
Why don't all programs built for one Linux-based operating system work for another one? For myriads of reasons, including hardware version, kernel versions, operating system versions, adminstrative strategies, and user environment. I have numerous such examples, as do we all.
Debian GNU/BSD or Debian GNU/FreeBSD seems to be the likley name of such a product.
There was alot of worry over FreeBSD branching, but overall there seemed to be a positive reaction.
A good number of people like the idea of the .deb package under FreeBSD. just as alot would like to see the FreeBSD ports tree in a Linux Distro
A quick search of the FreeBSD mailing lists on the topic reveals some .
A few choice articles can bee seen below.
The seemng start of discusion: FreeLinux
A nice answer: RE: FreeLinux (Debian/GNU BSD)
I have recently switched to FreeBSD, since I beleive it is better than Linux. And I personally think that GNU/FreeBSD would be the worst thing possible for FBSD. Here's why:
/usr/ports//
.deb packages would start appearing for fbsd and there wen't good old pkg_add for anything you need... In addition to that, I'm afraid debian would make a sort of linux distribution around the fbsd kernel (actually, I sould say GNU in this case). This means that even at the absolute minimum, the install would fill up your HD with every available piece of free software, no matter how bad it is. In BSD, you get a clean, structured system.
FBSD is a whole lot better than Linux mainly because the distribution is centeralized. Just like only good code makes it into the Linux kernel, only good code makes it into the fbsd distribution - all of it. In fbsd, there are no 5 different libraries for a single task, and you can be sure there is a single good library all programs use. You can be sure all programs would run on your distro, since there is only one distro. In linux, you have to install 43 and half different libs for the same tasks just to be sure everything runs on it.
There is also one single distribution for fbsd, so you can be sure things work on it in any case - you can safley download binaries, unlike linux where you have binaries for redhat6-glibc2 redhat5-glibc2 redhat6-libc5 redhat5-libc5 slackware-libc5 slackware-glibc2 suse6-libc5 suse6-glibc2 and so on, you get the point. In addition to that, sources might not be compatible between distros, since they use different libs and in differend locations. Simply put, you must tweak the program to get it working, unless an archive is available for your distro especially.
When I switched from linux to fbsd, it was like entering heaven. Things are organized, code is better. No more tweaking. If I want a program, it's simply a matter of typing:
cd
make && make install
and presto, a network connection is established, the sources are downloaded, patches are applied, the code is built, and then installed. Also, all the neccessary dependancies are also downloaded and built. Or even simpler, just mount the packages cd and do a
pkg_add
All the binaries, libs, etc. will be installed, everything will be in place and registered for later uninstall, and without an error. This is all because of a centeralized distribution.
Now, if Debian (whom I do respect for their linux distro) were to make a Debian GNU/FreeBSD distro, things wouldn't work this way. First of all, glibc would probably be ported so fbsd would start having seperate binaries for glibc and fbsd libc. And probably seperate bins for debian and the official distro. In addition to that, debian would start using it's own sources for linux programs, which means ports loses a lot of it's value. And
Now, a point that *must* be emphasize: freebsd is an *operating system*, not a kernel. Linux is a kernel, fbsd is an os. Read this sentance ten times: fbsd is an os, linux is a kernel. Making a freebsd debian distro would be like taking window's solitair and building a linux distro around it. fbsd would lose alot since people would start using it just for it's kernel, and not the beutiful system it is.
It's been mentioned that this way in an open unix system you would be able to choose your kernel, libc, etc. But this is a plain wrong view! Have you seen a new linux distro with just libc5 or glibc2 on it? NO! You would need to have ALL the kernels, and ALL the libcs this way. This would also make things worst for the linux community... if libcs become interoperable, linux binaries are opt to show up with the fbsd libc, which means 3 libc's for linux, which means even more mess.
One last point I want to make is that the support gnu, debian, et al get from linus in the linux kernel might not repeat itself in fbsd if debian won't make sure the fbsd community wants this step - and my guess is we don't.
Please try to consider all sides to this before taking such drastic steps. Thanks.
--
Oren Sarig
sarig@bezeqint.net
As for the licensing, I well and truly believe that the lion's share of the daily users of these systems could not possibly care one wee whit less about the distinctions in the licences.
Finally, just some would prefer an operating system with an FSF-derived user environment but a BSD kernel, others would prefer the world the other way around. You read about what progress they're making toward this goal in active threads the BSD newsgroups today.
But please, please consider those threads read-only. Don't flame. Just read. In particular, don't have a coronary when you hear about how like Microsoft's dirty tricks some people find GNU's "embrace-and-extend" and anti-POSIX strategies.
Just let people have what they want to have. Sure, a BSD kernel and FSF non-kernel would end up making one more free Unix operating system than we had before, but likewise would a Linux kernel (I don't believe the FSF owns it yet, right?) combined with a BSD non-kernel.
So what? The Linuxes are so splintered and disorganized now that people would never even notice another one. Sure, you'd be upping the number of BSD-based operating systems by a far bigger jump than you'd be upping the number of Linux-based operating systems.
(4+1)/4 is a a bigger number than (122+1)/122--or whatever--is. Strangely, the complaints about the smaller number involved here dramatically outnumber those about the larger one. Isn't that peculiar? Try to resist.
TRY TO RESIST!
Let's just let everybody have what they want to have, ok? Unix is Unix. It's not Microsoft. Isn't that enough?
Other people I've seen talk about this have used the term "FSF taint" instead. The kernel that Linux-based operating systems use is, for the most part, pretty well respected by other Unix programmers. The same cannot be universally said for the non-kernel clutter that various vendors slap together and call an operating system. So really, I'd try to avoid the term "Linux taint". I don't think it's really what you meant, and it won't get you in as much trouble. Try using the specific operating system, like "Redhat taint" or "Corel taint", or the specific political group, like "FSF taint", because that way you won't seem to be slamming the fine work that Linux Torvalds has done.
Actually, there are other important things, too, but none of them are related to making the moronic masses happy. Sometimes they're about making adminstrators, programmers, engineers, and scientists happy; in other words, people who aren't afraid of thinking and who aren't afraid of learning, people who aren't looking so much for a flashy new toaster without an instruction manual or a redundant TV set to babysit the kiddies, but rather for a complete system that does precisely what it was designed to do, one solidly integrated and tested by time.
Other venues exist for the less technophilic--like hiring a secretary. :-)
Considering how GNU-free BSD already is (modulo in most cases the compiler, but that's largely irrelevant to the user's experience, and doesn't produce infectious output), this always sounds strangely redundantly, so I always have to rescan a few times until the words jumble back into place. :-)
Given that, apparently, the "FreeBSD-based Debian" would use only the kernel and C library from FreeBSD, and would use their own utilities (and, presumably, other libraries) instead of the BSD ones, they wouldn't have to do any more work, over time, to "keep the GNU there" than to keep it in Debian GNU/Linux.
True - but that's not a consequence of the BSD folk using fewer and fewer GNU utilities over time, if the Debian folk aren't using the BSD utilities, except maybe to the extent that ports of GNU utilities to FreeBSD become less actively maintained.
Yes, the fork/exec interface isn't ideally suited to that.
Well, it could spawn it with CreateProcess() (IE 4.0, at least on Solaris, doesn't handle mail or news itself, but punts to whatever mail or news reader you tell it to - which, amusingly enough, makes it arguably more UNIXy than Netscape on UNIX, at least in that regard!), but that interface isn't, as far as I know, any more friendly towards that sort of pluggable text-editor functionality than is fork/exec. A COM/OLE editor interface might be better suited to that.
I have the impression that in some message about a mailer for GNOME that they were thinking of making a Bonobo interface for text-editor widgets, so that different widgets could be plugged in.
If they do so, I'm curious whether it'd be the first desktop environment/toolkit to do so; I have the impression something such as that could be done in Windows, or in KDE, but I don't know whether anybody's actually done so.
(The Andrew toolkit had, as its text widget, a fairly powerful editor, which had, I think, some amount of vi compatibility available atop its more EMACSish base, but if you're used to a particular editor, that still might not be all you'd want.)
It's messy to have to reboot to install tetrinet under leenux, reboot to fbsd and be able to run a good version of X with truetype support...
---
-
ping -f 255.255.255.255 # if only