Debian NetBSD

bXTr writes "Interesting project over at SourceForge. Quoting from the website, 'Debian NetBSD is a port of the Debian Operating System to the NetBSD kernel. It is currently in an early stage of development and cannot currently be installed from scratch. Instead, a tarball of the current envionment is available and can be extracted into a handy directory on a NetBSD system.' Check out the reasons why they're doing it and some interesting commentary at DailyDaemonNews on this."

Re:kaboom

[erlenic]

To me this kind of thing really shows the strength of the open source community. I'm sure we've all seen the flame wars that start here over the difference in the licenses, but in the end, we all have a common goal, share the source!

Cooperation

[awgy]

To me, this is promising. I like to see cooperation between the Linux world and the *BSD world. Both have their advantages, and it'd be great if both would learn from each other more often. Perhaps this is an instance where some exchange of ideas could come about? Those responsible deserve a pat on the back.

Re:kaboom

[Anonymous Coward]

It's not so much a common goal, as it is compatable goals.

Re:I would prefer the other way around

[aussersterne]

What is wrong with the FreeBSD kernel?

FreeBSD doesn't have as many drivers for esoteric hardware. Not a technical advantage for Linux so much as a side effect of the greater market penetration.

But that's the way it is, nevertheless.

Re:the kernel? my god man

[CheeseMunkie]

" Theres a core group in charge of what goes and what stays."
Actually, in Linux it's the same (f.e. Torvalds, Cox, Tosatti).

This is true of the kernel, but the kernel is not the whole deal. One of the major problems with Linux is *that* it's every yahoo for himself -- Cox and Torvalds and a few others do the kernel, the glibc people are a different bunch, the X consortium [x.org], the ISC [isc.org], Apache Foundation [apache.org], plus all those assorted little libraries, you know the type, it's a kinda neat library, but you've only found 1 app that needs it ... Everyone does their own thing and contributes it to the slushpot, but nobody controls the pot.

So, where the BSD team is some 10-20 people who can all get in a room and hash out details and come out with a coherent ports system, or a standard place to put software (apache goes in /var/www? Wtf patrick?), the Linux world is far too big to do that. Hell, we can't even document stuf coherently -- everything has its own man page, readme, manual, plus linux documentation project [linuxdoc.org]. Compare to FreeBSD's Handbook [freebsd.org].

This is a weakness in the Linux system of cooperation. It's also a strength. Just as no one can take control of the whole thing and fix it, also nobody can break the whole thing. Even if Linux and Cox between them decided to sabotage Linux, they couldn't, whereas one guy with cvs commit privileges on cvsup.freebsd.org could give himself a root shell on every BSD box on the planet. (Okay I exaggerate -- he'd get caught, probably, but that's only because most of the people working on BSD are good guys.)

Re:They're not cooperating...

[Carlos Laviola]

Their motivation is quite clear to me: have another choice of a kernel that is already robust and better than the Linux kernel in certain areas, and have the nice, organized and easily upgradable Debian on top of it. Come on, building *everything* from source is so much more trouble than apt-get dist-upgrade. Think XFree86, Mozilla, etc.

I don't like it

[znu]

The thing I've always really liked about the BSDs is that they're complete and separate systems that include everything from the kernel to the userland tools, all integrated by one team. Compare with the Linux world, where you have a bunch of different distros that many people pretend are all the same OS (in spite of the fact that file systems are arranged differently, boot sequences are different, configuration is different, package management is different, userland tools are often different, etc.) because they happen to use the same kernel. The BSD way has always seemed a lot cleaner to me. The idea of seeing a myriad of distros based on the BSD kernels really isn't one that I like. I believe it's a step in exactly the wrong direction. Open source Unix needs more standardization, not more fragmentation.

how different (from standard netbsd) is it?

[foonf]

Did they just port apt and dpkg, and put up some Debian-packaged NetBSD binaries? Or have they moved to a Sys V init system, ported the Debian administration and configuration tools, and all the other stuff that makes debian distinctive? They explicitly say that NetBSD doesn't support runlevels, and looking at the package list, it doesn't look like much of the debian tools have made it yet.

If its just a different package system, its pointless. Less work, and more immediately useful results, would be modifying apt to work with the current binary package system (which actually does support dependencies, etc.), and the large number of binaries in this format already available.

If not, its a more questionable proposition. Arguably, its not really BSD anymore...it runs NetBSD binaries and uses that kernel, but the userland is basically Debian, ie, just like any Linux distribution. And most people who want that should just assume use Debian with the Linux kernel, which is a far more mature combination. Yes, for VAXen, toasters, slide rules and other more arcane platforms this won't exactly work, but Debian-NetBSD doesn't seem to have package for these platforms anyway.

Transition

[tiny69]

Not too long ago, someone made the comment on slashdot about the general progression of Linux users. As a users becomes more experienced with Linux, they tend to shift from:

Mandrake/RedHat -> Debian/Slackware -> *BSD

It seems that Debian is going to make that last transition a little easier.

ports

[lightfoot jim]

What exactly is the ports system?

More like, what are rpm users missing out on? With rpm -i package.rpm the user may or may not be able to install the intended software. There could be real dependency problems, as in kde2 needs qt2. There could also be bogus dependency problems since you may have compiled qt2 from source but rpm wouldn't know about it.

Enter FreeBSD and ports. A typical FreeBSD install creates a directory called /usr/ports which is a whole tree of makefiles. So to install something, you just cd /usr/ports/category/WhateverYouWantToInstall/ && make && make install. All dependencies are taken care of automagically. The makefiles in these directories are smart enough to download whatever you need and then compile the source on your machine. So installing a new package doesn't take several hours of trolling newsgroups and searching for rpms.

But you don't have to take my word for it. Check this [freebsd.org] out.

My experience is limited to Mandrake, Slackware, FreeBSD and OpenBSD. They each have their strengths and weaknesses, but when I need to get sh*t done, BSD, espescially FreeBSD is my first choice just because the ports tree contains nearly any software I'd want to run, eliminating the bottleneck that software installation sometimes turns into and letting me get to the task at hand.

As an aside, it seems like everything that Mandrake tries to be to "joe sixpack" who is just getting into trying linux on the desktop, BSD is to the sysadmin or programmer who needs to get a *nix platform up and running for a certain task. Compiling a custom kernel, installing software, modifying the init process, etc are at least as easy for the sysadmin on BSD as adjusting the screen fonts and changing the wallpaper are for a newbie in Mandrake.

FreeBSD ports and Sorcerer

[Shiny Metal S.]

So it's basically the same thing which Sorcerer [slashdot.org] wants to achieve, right? Than maybe Sorcerer [wox.org] should just use FreeBSD [freebsd.org] ports instead of reinventing the wheel?

Re:I would prefer the other way around

[Anonymous Coward]

Does linux have support for usb mice yet? The last time I tried installing it, it didn't.

Re:ports

[scrytch]

> How is this any more convenient than apt-get which is also available for RPM based systems?

Because ports does not require packages. Example, if I want to install a KDE app, it does not require that I have the qt package installed. It searches for libqt.so. If it's a gtk package, it runs gtk-config, if it's apache, it runs apxs, and so forth. Now debian's a little better than RPM, which ends up making you force-install just about everything (defeating dependency checking) because you installed something from source, but you still have to intervene when a dependent package isn't present. Ports assumes you know what you're doing, and if the lib is there, it's there, it doesn't need a package manifest to tell you. It does check for the package first, and ports does build a package, so you get a package-based system that degrades gracefully when you don't religiously use the package system.

THAT is why I use ports. Because no sysadmin I know of takes the builds out of the box, they keep their source trees around to tweak and recompile as needed. Oh, and ports lets me do that doo, I just "make get" the port, cd work/packagename, and there's the source tree as if I'd untarred it myself. I can configure && make install it from there, or cd ../..; make install from there and it builds as a package.

Re:I would prefer the other way around

[kan]

IMHO, because if you need high end system in the first place, you will be really better served by buying system from Sun, IBM or HP. Their offerings of enterprise-class systems are much more mature and stable than any of the i386-base alternatives on the market today.

SMPng project is not only about improving SMP performance of FreeBSD. It is also about making kernel fully preemtable, which has its advantages for real-time tasks, responsiveness in desktop and multgimedia environments. Besides, with an upcoming PPC and UltraSPARC ports having better SMP support kinda starts making more sense :)

Re:Noooooooo! Stay away!

[faedle]

I'm sorry if keeping it stable is such a bad thing for you. Fortunately for people who's jobs depend on reliability and stablilty of the systems they maintain, there are people like the Debian developers who think that keeping it stable is a Good Thing.

How quickly we forget that stability is how many of us have sold Linux and *BSD to our employers, and that sudden rash changes and stability problems can leave those who require stability out in a lurch.

I see Debian's methodical approach to "current" as a good thing. Debian is about choice: if you want to be on the bleeding edge, please run unstable, we need the beta testers. If that's not your game, here's the stable tree. If you're really anal (or demented), here's the last version with security patches.

Sounds like a win for everybody. NetBSD gains more users (those who happen to get off on apt-get), Debian gets to work on porting to other kernels (a major cornerstone of the Debian project's defined goals), and everybody gets more choice.

Stay away from system administrators who think that the "latest" is always the "greatest."

Re:Noooooooo! Stay away!

[The Ape With No Name]

My experience with STABLE and Debian GNU/Linux is that it is an outdated tree. FreeBSD pkgs, again in my experience, are right where they need to be, not outdated versions of common apps.