Slashdot Mirror

← Back to Stories (view on slashdot.org)

NetBSD 2.0RC2 Released

Posted by timothy on from the will-assimilate-you dept.

An anonymous reader writes "NetBSD 2.0 RC2 has been released. Get it using sup or ftp from one of the mirror sites. NetBSD is used to routinely set transmission-speed records, and is widely considered to be the cleanest of the BSDs. NetBSD is widely portable."

56 comments

  1. Cleanest? by chjones · · Score: 3, Interesting

    Just out of curiosity, who considers NetBSD to be the "cleanest" of the BSDs, and why? I'm not trolling, just wondering---I don't think I've ever heard that (specifically). CDJ

    --

    Christian Jones
    Medicine. Mathematics. Mediocrity.

    1. Re:Cleanest? by redhotchil · · Score: 3, Informative

      Its the most simple, and therefore is easiest to port to all sorts of hardware. Its not huge on features, but rather simplicity, portability, stability and flexibility.

    2. Re:Cleanest? by vga_init · · Score: 4, Informative
      Well, I am not an expert on it, but my gut feeling tells me that it's linked with the concept of portability. We know that it's portable because it's been proven so many times, and I think portability implies an above-average deal of cleanliness. In order to port your software, you need a system that is very logically and uniformly structured. So, your design goals are focused on creating an implementation that contains the least amount of messy or inextricable code. Also, the pressure for portability helps promote the deprecation of code that might otherwise cling.

      In this sense we can think of cleanliness and portability as things that imply the other, though it would be quite a different thing to interpret this particular kind of cleanliness as anything but that; it wouldn't be safe to assume on those grounds alone that the code would be faster/more efficient or be more feature rich/powerful, which I'm thinking some people might want to extrapolate. It's not beyond reason, however, and, as the author points out, there are speed records involved, so its possible that these things are related somehow (but not necessarily).

    3. Re:Cleanest? by burns210 · · Score: 1

      Not to troll, as the grandparent did, I have an honest question.

      Why is out-of-the-box, desktop-ness a low priority for netbsd, while it is so stressed for Linux? Seems like a 'clean' system would more easily be achieved with a 'clean' OS...

      I just wish I had an Ubuntu-like netbsd system using those sweet ports system.

    4. Re:Cleanest? by the+real+darkskye · · Score: 2, Interesting

      Aside from FreeBSD's install system, NetBSD is the easiest I've encountered.
      It is much more friendly (well less intimidating) than OpenBSD's although I do have reservations about it, such as it not storeing any options you set at install time, and not booting with RC_CONFIGURED=false by default to alert you to the fact you need to set options.

      --
      Music is everybody's possession.
      It's only publishers who think that people own it.
      Fuck Beta
      ~John Lenno
    5. Re:Cleanest? by idiotnot · · Score: 1

      In a way, NetBSD is closest to being desktop-ready out-of-the-box, seeing as how they include X as a part of the base system.

      That said, I think NetBSD has a few problems on the desktop.

      1. People ignore NetBSD. Probably about 90% of the world is x86, and FreeBSD seems to rule the roost for x86 BSD. Although I really like NetBSD, I often grab for my FBSD mini-iso on x86 before NetBSD. Other BSD users show up in the form of the "OMG Sekure!" crowd, who gravitate to OpenBSD. So, the platforms left are non-x86, and many of those have commercial Unix that's easy to get, if not free to use. (think Solaris on Sparc) When people do decide to try NetBSD, it's because they've got a toaster they want to use for something small.
      2. Desktop nicities. Things like accelerated X and Java take more work under NetBSD.
      3. Notebook support. NetBSD's tends to lag behind.

    6. Re:Cleanest? by Anonymous Coward · · Score: 1, Insightful

      Why is it that NetBSD users always try to criticise OpenBSD users? The reverse doesn't seem to be true...

    7. Re:Cleanest? by Anonymous Coward · · Score: 0

      It's the cleanest because hackers concentrate on writing code instead of bikeshedding and pushing their personal agenda. Oh, and did I say it's free of assholes like Darling Smorgrav?

    8. Re:Cleanest? by Anonymous Coward · · Score: 2, Informative

      What exactly does NetBSD lack that FreeBSD has?

      usb stack: check (FreeBSD borrowed it)
      SMP : check (getting much better fast)
      RAIDframe: check (FreeBSD killed it on 5.x)
      Kernel assisted threading (Scheduler Activations) : check (works better than FreeBSD's KSE btw)
      FFSv2 : check
      Sendmail and postfix in base: check
      veirexec : check (FreeBSD doesn't have it)
      pkgsrsc: check (works on FreeBSD too)
      rcNG : check (FreeBSD borrowed it)
      arrogant developers: nope, FreeBSD does though, phk@, des@ and some more.
      cross compilable kernel, world *and* X : check

      Find me a more feature rich, portable and clean unix-like system, please...

    9. Re:Cleanest? by 0racle · · Score: 2, Insightful

      Different priorities for different projects. You could also say that OpenBSD has a low priority for desktop 'users' (as opposed to developers) with the stress on absolute security being the over-riding factor, everything else gets added in later. NetBSD works toward a clean implementation and portability of the code across many different platforms, everything else comes second. Not to say that NetBSD doesn't care about security, just that its not the number one driving factor, and that still sounds bad. On the other hand, Linux, as part of the fact that the parts that make a complete OS are created by different people for each part, seems to aim at whatever the most vocal of the parts wish it to do, and thats be a Microsoft killer, hence the appearance to attempt to gear it more to a desktop user.

      --
      "I use a Mac because I'm just better than you are."
    10. Re:Cleanest? by Fweeky · · Score: 4, Informative

      "RAIDframe: check (FreeBSD killed it on 5.x)"
      Options for software RAID on FreeBSD 5 right now: vinum (old and crufty these days), gvinum (GEOM version of vinum), gmirror/gstripe (standalone GEOM RAID-1 and 0 modules), ataraid, ccd (old and crummy, seems mainly an experiment in GEOM porting). Was there a compelling reason to keep RAIDframe?

      "Kernel assisted threading (Scheduler Activations) : check (works better than FreeBSD's KSE btw)"
      References? I'd be especially interested in a MySQL super-smack benchmark on a decent SMP system.

      "FFSv2 : check"
      With snapshots? Obtained from: FreeBSD? :)

      "Sendmail and postfix in base: check"
      Great, now I get two mail daemons I don't like instead of just one. At least syspkg makes them easy to avoid :)

      "veirexec : check (FreeBSD doesn't have it)"
      whatexec?

      "arrogant developers: nope"
      Just the users, huh? ;)

    11. Re:Cleanest? by mnmn · · Score: 3, Interesting

      I've always considered OpenBSD to be the cleanest. It has resisted major change but has been code-audited much more than NetBSD.

      I dont know about portability but OpenBSD was forked from NetBSD, and has been changed less since and audited more.

      But if you mean 'original' by clean, the earlier versions of either would be cleaner.

      --
      "Give orange me give eat orange me eat orange give me eat orange give me you." -Nim Chimpsky
    12. Re:Cleanest? by beholder77 · · Score: 4, Informative

      I've used both NetBSD and FreeBSD systems as desktops for over a year, and I can say from the standpoint of initial setup they're pretty equal. But I can say the same thing for most Linux systems as well. You can get X, KDE, GAIM, Moz, X-Chat, etc working pretty easily on both systems after an initial install.

      However, where they differ is the maintainability. FreeBSD has portupgrade, which doesn't seem to exist in the NetBSD ports world. Portupgrade, although not a perfect port management tool, is damn near close to perfect :)

      Having a ports tree is nice, all the BSD's have that, but having a tool that can intelligently upgrade your existing packages without having to remove and recompile 1/2 your installed packages (try a "make update" sometime on NetBSD, you'll see), is a must.

      As well, updating NetBSD from source has always been an exercise in frustration for me. Most of the time, after updating, I can no longer rebuild the userland. It's always some small problem or other, but it's still a far stretch from FreeBSD-STABLE which has broken maybe 3-4 times in the 5 years I've worked with it.

      Stability of desktop apps seems to be a bit of an issue for me as well. Mozilla in particular (Linux emulated, and native) core dumps on me on a weekly basis. This was with the pre-RC 2.0 beta's though, so they might have worked that issue out. I imagine it can't be threading related, because KDE seemed very solid, and never gave me grief.

      After all this, I still trust NetBSD as a firewall which it has performed perfectly for me for years now. It's also the only OS that I find works well on non i386 hardware. Everything in the Linux world I've tried for Sparc (not Sparc64) has been crap compared to NetBSD in terms of stability. As well, 68k support (Older macs, and older HP workstations) on NetBSD is top notch, and very usable. I actually had a Quadra 610 mac (33Mhz, 24 meg ram, 250 meg SCSI-2 hdd) running apache, php, postgresql and bind 8, and it was still surprisingly fast, and even more amazingly fit well within that tiny hard drive.

      NetBSD can give your older oddball hardware new life. The scalability factor that they push is not only impressive upwardly, but downwardly as well. Can you imagine any Linux distro working well on 25Mhz machines with 250 meg drives? Me neither. When they finally come out with a tool for netbsd like portupgrade, and 2.0 gets a minor revision number, I'll be there with my desktop systems.

      --
      Success is as dangerous as failure, hope as hollow as fear.
    13. Re:Cleanest? by LizardKing · · Score: -1, Troll

      Why is it that NetBSD users always try to criticise OpenBSD users?

      Perhaps it's because it could be argued that OpenBSD developers are a little bit too strident in their claims to have the most secure freenix. If Theo hadn't been such an arsehole on the NetBSD mailing lists then he could have pursued a security audit of the NetBSD code without having to fork it. Instead he's given us another BSD that lacks features like clean upgrades between releases, decent Linux emulation or better than average performance. I tried OpenBSD several times during the later 2.x releases, having swallowed the "super secure" hype, and was so underwhelmed I went straight back to NetBSD.

      Free and Net are the freenixes where interesting system level stuff is developed (SA, SMP, FFSv2, rcNG, etc.). Much of this gets ported between Free, Net and Open. Where Open has made the greatest contribution is in the userland with things like OpenSSH.

      The reverse doesn't seem to be true...

      Theo hasn't popped up on the NetBSD mailing lists in a while, but when he does it's to crow about some percieved superiority of Open versus Net. He has mellowed a bit over the years, but he's still quite happy to f*ck off users with acerbic remarks.

    14. Re:Cleanest? by Anonymous Coward · · Score: 0

      Ok, first the mozilla issue, did you bother reading the MESSAGE in the package dir?

      Upgrading your world is a few commands away. Build a new kernel, boot new kernel, cd /usr/src && ./build.sh -j 3 distribution. Wait, shutdown now. cd /usr/src && ./build.sh install=/ && etcupdate && reboot. Done.

      There's no portupgrade because it isn't needed, install pkg_chk from /usr/pkgsrc/pkgtools/pkg_chk/ and you're done. What you need is done via pkg_chk -i and pkg_chk -u.

    15. Re:Cleanest? by Anonymous Coward · · Score: 0

      > FreeBSD has portupgrade, which doesn't seem to exist
      > in the NetBSD ports world.

      Such tool DOES exist on NetBSD.
      Just try "make update" on a pkgsrc directory,
      All packages will be automatically updated.
      I usually use "env UPDATE_TARGET=package make update",
      so all recreated programs are automatically re-packaged.

      > As well, updating NetBSD from source has always been
      > an exercise in frustration for me. Most of the time,
      > after updating, I can no longer rebuild the userland.

      Hmm, what sort of problem does happens at updating?
      I do similar thing, but I don't have much problem so far.

      > Mozilla in particular (Linux emulated, and native) core dumps
      > on me on a weekly basis.

      FWIW, firefox works fine for me.

    16. Re:Cleanest? by Anonymous Coward · · Score: 0

      No. Clean != Original.

      Clean == Clean == Well-designed == Scalable == Modular == Easier to change

    17. Re:Cleanest? by beholder77 · · Score: 1

      There is no message in the native mozilla package dir, and the one in the linux mozilla just tells me to have COMPAT_LINUX in my kernel... duh. Mozilla core dumps on me after a few days of use. It gets slower and slower until it dies, it's not a core dump on execute.

      I understand the upgrade procedure, the problem is the source breaks quite consistently. I would say that 7 times out of 10 updating my source tree will induce some breakage that is only fixed by whacking the entire tree and re-downloading. Also, the procedure you outline is only for 2.0 systems not 1.6.x systems, which I'm still using. I fully understand why the 2.0 tree was breaking in the past, it's the 1.6 tree breakage that causes me the most grief.

      I installed pkg_chk on my sparc, and it seems to be quite useful. I don't have any packages out of date yet so I can't test the update functionality, but it does have the features I'm looking for. Thanks :)

      --
      Success is as dangerous as failure, hope as hollow as fear.
    18. Re:Cleanest? by Anonymous Coward · · Score: 0

      vinum and gvinum are ok if you're familiar with Veritas' VxVM, else they're unnecessarily complex (plexes, subdisks, volumes, etc). RAIDframe is dead simple to configure (much like ccd) and does the job fine. RAIDframe was removed because phk@'s GEOM ego trip entered the game. It also broke vinum. Ask Greg Lehey how happy he was with that and how difficult dealing with Poul-Henning can be (yet he still has his commit bit intact, unlike Dillon). Tell me about GEOM 2 years from now when phk gets bored of it and "can't maintain it unless someone pays me".

      References? I'd be especially interested in a MySQL super-smack benchmark on a decent SMP system.

      Well, for one KSE is not finished on most arches yet. By working better I actually mean it has fewer bugs. Performance doesn't matter much if the implementation is still buggy.

      "FFSv2 : check" With snapshots? Obtained from: FreeBSD? :)

      Yes, isn't cross pollination great? Btw, LFS is still making progress and might be a trustable replacement in a not so far future.

      "veirexec : check (FreeBSD doesn't have it)" whatexec?

      typo, I meant veriexec

      "arrogant developers: nope" Just the users, huh? ;)

      Not taking Theo into account, how many developers have been driven away from NetBSD due to the attitude of other committers? That's what I thought. Now how many people have been driven away from FreeBSD due to the arrogant attitude of people like Poul-Henning and Dag-Erling? Quite a few already.

    19. Re:Cleanest? by mnmn · · Score: 1

      From online Merriam Webster:

      free from dirt or pollution
      free from contamination or disease
      UNADULTERATED, PURE

      In some meanings it means original, in others it means without the unwanted 'pollution' or bugs.

      So a more thoroughly audited code would be cleaner than 'better designed' which isnt too different between netbsd and openbsd in this case. Theres some very buggy software out there that was designed well, but not implemented properly.

      And the most thoroughly washed and cleaned up BSD, is openbsd although really the difference isnt huge, for cleanliness or good design.

      --
      "Give orange me give eat orange me eat orange give me eat orange give me you." -Nim Chimpsky
    20. Re:Cleanest? by Anonymous Coward · · Score: 0

      Upgrades work quite well from release to release (or snapshot to snapshot).

      I've never had an issue with the Linux emulation.

      OpenBSD's also provided the best free firewall software available, which has been ported to FreeBSD and NetBSD. I'm not sure if anyone is considering porting CARP or not though.

      Theo mentions NetBSD occassionally, and it's never a nice comment. :P

    21. Re:Cleanest? by tedu · · Score: 2, Insightful

      "widely" is a codeword for "i can't prove it, but it sounds good." see also: "everybody else thinks so, what's wrong with you?"

    22. Re:Cleanest? by Anonymous Coward · · Score: 0

      There is pkg_chk in NetBSD's pkgsrc. This can check pkgsrc for any updated packages, and rebuild them along with all their dependencies.

      You can also do this one package at a time with pkg_chk if you just want to upgrade a particular piece of software.

      Very nice, and basically the equivalent of FreeBSD's portupgrade.

    23. Re:Cleanest? by Anonymous Coward · · Score: 0

      Can you imagine any Linux distro working well on 25Mhz machines with 250 meg drives? Me neither.

      Actually, I think this is one advantage to NetBSD... I remember when they went to the new rcorder code for startup scripts, some slowness occurred on older hardware like the Vaxen. It prompted them to look at how to speed it up. Now, on a P4/3Ghz, I'm sure the startup cost was negligable (an extra 1/2 second somewhere is hardly noticable), but on slower boxes things become more obvious.

      So, running on older/slower boxes can a lot of times improve the code by forcing people to examine things that would be hardly noticable on new hardware. Now arguably, the rc startup only happens once, so its not as critical.. but what would happen if the code to say, fork a process, became obviously slower. Even on a fast box, where it might be negligable under low load, it could cause issues when load increases (say, apache get nailed and starts forking more processes to handle the traffic). In situations like this, having things like this become noticable sooner, because of older hardware, becomes an advatage to *starting out* with better design, rather than releasing the OS and finding problems with your "release" version later on.

    24. Re:Cleanest? by beholder77 · · Score: 1

      That's a good point. I like to test my PHP and Python scripts out on older, slower machines so I know that I'm not introducing anything horribly inefficent.

      My work life is spent helping people code ASP.NET pages for a municipality, and I have to say that any organization with money seems more inclined to throw bigger hardware at a performance issue rather than spend any amount of time profiling the applications. We have monster boxes running bad web applications that could be served from really tiny boxes if the original author wouldn't have had a P4 3.0HT, 2gig ram desktop to develop on.

      --
      Success is as dangerous as failure, hope as hollow as fear.
  2. uh huh by syrinx · · Score: 4, Funny

    NetBSD is widely portable.

    In other news: the ocean is wet. :P

    --
    Quidquid latine dictum sit, altum sonatur.
    1. Re:uh huh by Anonymous Coward · · Score: -1, Flamebait

      In yet other news, it is less portable than Linux.

    2. Re:uh huh by Anonymous Coward · · Score: 0

      OMG OMG!!!! *masturbates like crazy*

    3. Re:uh huh by Anonymous Coward · · Score: 0

      Ha ha YHBT.

      Why does it upset you that Linux is more portable than NetBSD?

    4. Re:uh huh by Fritz_the_Cat · · Score: 0, Offtopic

      You were trolling? Nnooooo
      Well you got me, I've been trolled.
      Man, you're just tooooooooo good.

    5. Re:uh huh by Anonymous Coward · · Score: 0

      You're pretty messed up.

      Let me give you some advice - if you don't like the fact that Linux is more portable than NetBSD, then ignore it. No need to start masturbating and carrying on like an idiot.

    6. Re:uh huh by Fritz_the_Cat · · Score: 1

      but I'm a Linux fan.
      i just got a bit enthusiastic, that's all.
      you really showed them netbsd users how good linux is.
      What say we have a group mass masturbation session?

    7. Re:uh huh by Anonymous Coward · · Score: 0

      Well no you can fuck off you faggot.

      I'm actually a BSD fan. I happen to be the only one who can admit than Linux is better than a BSD at something. Considering the rest of them have the collective IQ of a dimwitted chimp, that is understandable though.

    8. Re:uh huh by Fritz_the_Cat · · Score: -1, Flamebait

      Naughty, naughty.
      Hasn't Mother told you before that swearing is rude?
      You'll get a spanking when she finds out, you will.

    9. Re:uh huh by AvantLegion · · Score: 3, Funny
      All I know is that NetBSD will even install on my taco.

      I don't mean an N-Gage. I mean the damn taco I just bought from Taco Bell.

      I accidentally dropped a NetBSD CD into the Taco Bell bag, and when I pulled out my taco, there was a bash prompt on the tortilla shell.

  3. DANGER by ArbitraryConstant · · Score: 2, Funny

    DANGER

    Replying to this post may result in flames no matter what your position.

    I know which BSD is the cleanest. I'm just not going to say it here...

    --
    I rarely criticize things I don't care about.
  4. NetBSD portability vs Linux by Anonymous Coward · · Score: 3, Insightful

    it is less portable than Linux.

    It depends of what you are talking. Linux is just the kernel, NetBSD is a compete OS. You have to pick up a Linux distro and compare its portability to NetNSD.

    NetBSD builds for more than 50 architectures from the same source tree, fully supporting cross-building of the entire system (it's as simple as running a shell script).

    It has the same distribution layout for all the supported architectures, and the same installation system for most of them. It has machine independant drivers (write once, run everwhere), including for things such as the system console.

    NetBSD also has a cross-platform package system (in fact it's even cross-OS, as it works also for other systems) that automates rebuilding from source. The vast majority of the packages are therefore available to all NetBSD architectures.

    Now can you name a Linux distribution that can compete?

    1. Re:NetBSD portability vs Linux by Anonymous Coward · · Score: 0

      I'm talking about Linux versus the NetBSD kernel.

      NetBSD builds for more than 50 architectures from the same source tree

      Well, 50 *system architectures*. About 13 or so CPU ISAs/ABIs (versus IIRC about 18 for Linux).

      Now can you name a Linux distribution that can compete?

      Debian has competitive infrastructure. It supports about 11 architectures. Will be as many as NetBSD when PPC64 and x86-64 get on board (shouldn't be too long, as they are supported by lots of other distros).

    2. Re:NetBSD portability vs Linux by Ezdaloth · · Score: 1

      Actually, NetBSD run on 17 CPU architectures, and linux on 16 :P

    3. Re:NetBSD portability vs Linux by Anonymous Coward · · Score: 1, Informative

      Linux supports:
      alpha, arm, arm26, cris, h8300, i386, ia64, m32r, m68k, mips, parisc, ppc, ppc64, s390, sh, sh64, sparc, sparc64, v850, x86-64

      NetBSD supports:
      alpha, arm, arm26, i386, m68k, mips, ns32k, parisc, ppc, sh, sh64, sparc, sparc64, vax, x86-64

      Actually, I count 20 for Linux vs 15 for NetBSD.

    4. Re:NetBSD portability vs Linux by Ezdaloth · · Score: 1

      You're right there, yes. I only counted which were on each team's hardware page. They seem to disagree about what is a CPU-architecture.

  5. Re: I know which BSD is the cleanest... by Anonymous Coward · · Score: 0

    DragonFly!

  6. Give it up already! by Anonymous Coward · · Score: -1, Troll

    Apart from some very minor, rare setups *BSD is WORTHLESS! Linux owns you on every single front these days. Install Slackware and move on with your life. Your dedication to the rotting corpses that are the *BSD variants is noble but futile. Bury the decaying past and join the Slackware revolution.

  7. actually, RC3 is tagged by jschauma · · Score: 3, Informative

    Two days ago, RC3 was tagged, adding an NFS fix (transfers or directory operations hang under special circumstances).

    --

    -- "Tradition is the illusion of permanence."
  8. Unfortunately, this may be their last release by Anonymous Coward · · Score: -1, Flamebait
    It is official; Netcraft now confirms: *BSD is dying

    One more crippling bombshell hit the already beleaguered *BSD community when IDC confirmed that *BSD market share has dropped yet again, now down to less than a fraction of 1 percent of all servers. Coming on the heels of a recent Netcraft survey which plainly states that *BSD has lost more market share, this news serves to reinforce what we've known all along. *BSD is collapsing in complete disarray, as fittingly exemplified by failing dead last in the recent Sys Admin comprehensive networking test.

    You don't need to be a Kreskin to predict *BSD's future. The hand writing is on the wall: *BSD faces a bleak future. In fact there won't be any future at all for *BSD because *BSD is dying. Things are looking very bad for *BSD. As many of us are already aware, *BSD continues to lose market share. Red ink flows like a river of blood.

    FreeBSD is the most endangered of them all, having lost 93% of its core developers. The sudden and unpleasant departures of long time FreeBSD developers Jordan Hubbard and Mike Smith only serve to underscore the point more clearly. There can no longer be any doubt: FreeBSD is dying.

    Let's keep to the facts and look at the numbers.

    OpenBSD leader Theo states that there are 7000 users of OpenBSD. How many users of NetBSD are there? Let's see. The number of OpenBSD versus NetBSD posts on Usenet is roughly in ratio of 5 to 1. Therefore there are about 7000/5 = 1400 NetBSD users. BSD/OS posts on Usenet are about half of the volume of NetBSD posts. Therefore there are about 700 users of BSD/OS. A recent article put FreeBSD at about 80 percent of the *BSD market. Therefore there are (7000+1400+700)*4 = 36400 FreeBSD users. This is consistent with the number of FreeBSD Usenet posts.

    Due to the troubles of Walnut Creek, abysmal sales and so on, FreeBSD went out of business and was taken over by BSDI who sell another troubled OS. Now BSDI is also dead, its corpse turned over to yet another charnel house.

    All major surveys show that *BSD has steadily declined in market share. *BSD is very sick and its long term survival prospects are very dim. If *BSD is to survive at all it will be among OS dilettante dabblers. *BSD continues to decay. Nothing short of a miracle could save it at this point in time. For all practical purposes, *BSD is dead.

    Fact: *BSD is dying

  9. Status of SMP support? by Greg+Koenig · · Score: 2

    What is the condition of the SMP support in the upcoming NetBSD 2.0 release?

    I know that SMP has been present in -current, at least for architectures like Alpha (and others?), for several years. My impression is that 2.0 is intended to be the first release where SMP is a real priority. I have been a longtime FreeBSD user and have watched their progress (and struggles) as the state of the art has moved from SMP in the 4.x kernels into the 5.x kernels where they are trying to squash down the BGL. How does the SMP work in NetBSD compare to this effort? Is it more akin to the FreeBSD 4.x efforts, are they trying to do something more along the lines of the FreeBSD 5.x work, or something else entirely?

    The reason that I ask is that NetBSD has this reputation for not doing something until they figure out the "right" way to do it. So I guess what I'm really asking is what their take is on the "right" way to do SMP. I really want to look at the source code to figure this out, but haven't had the time to invest in this, so I'm wondering if someone else who has might share their impressions.

    1. Re:Status of SMP support? by Anonymous Coward · · Score: 0

      NetBSD's SMP is like the one in FreeBSD 4.x, i.e. a big kernel lock. Fine grained locks are a 3.0 project. IMHO they should look at what Matt Dillon is doing on DragonFlyBSD, sounds like a simpler and more efficient approach to me. The SMP code in FreeBSD 5.3 is very complex, I doubt more than a few hackers actually understand the subsystems.

    2. Re:Status of SMP support? by Greg+Koenig · · Score: 1

      Thanks for the info, AC.

      The complexity of the SMP code in FreeBSD 5.3 was what I was talking about. IMHO, the FreeBSD hackers have been working very hard to move forward with fine-grained locking and my main worry is that they're making such a complicated system that it's going to be hard to continue moving forward with it.

      I am really interested to see what Matt comes up with in DragonFly and have read a lot of the design info that he has posted as well as taken a look at a fair amount of his source code. He's a smart guy, and I think the hackers working on this project stand a good chance of doing something that will get a lot of attention. I'm not sure how directly applicable the technique is to any of the other BSDs without major work, though.

    3. Re:Status of SMP support? by ArbitraryConstant · · Score: 1

      "The complexity of the SMP code in FreeBSD 5.3 was what I was talking about. IMHO, the FreeBSD hackers have been working very hard to move forward with fine-grained locking and my main worry is that they're making such a complicated system that it's going to be hard to continue moving forward with it."

      I don't think the way DragonFly does SMP is best for everything.

      The problem is that the DF kernel has a great deal less freedom to do what it likes with threads. This may ultimately give DF the advantage because it makes so many things so much easier, but for now, for right now, it increases latency (particularly in relation to 5.x). A desktop system that "feels" responsive is easier to achieve with FreeBSD (or Linux).

      In reality you may not notice the difference, but I can tell pretty easily under high load.

      The DF is probably ultimately better for highly loaded servers, and certainly for maintainability. Also, the way the userspace system call layer works takes it to a whole other level.

      --
      I rarely criticize things I don't care about.
    4. Re:Status of SMP support? by idiotnot · · Score: 1

      I had a beta version of 2.0 running on an AlphaServer 4100 (4x400Alpha/2.5GB ram)a couple of months back (before my biz partner traded the furnace for a PowerMac).

      Ran just fine, including full SMP support. Didn't note any real performance difference between FreeBSD 4 and NetBSD.

      But the NetBSD tree is still supported, and FreeBSD has dropped Alpha support. It's probably a good thing, too, seeing as how I could never get 5.1 or 5.2 to install on that machine.

  10. Get With it Already by ToasterTester · · Score: 1, Offtopic

    NetBSD does it again: after the original Internet2 Land Speed Record set by NetBSD in 2004 May 3 was broken, NetBSD shines again: Once more researchers at the Swedish University Network (SUNET) have broken the Internet2 Land Speed Record, using the upcoming version of NetBSD 2.0. The new records are 124.935 Pbmps in a single stream (was 69.073 Pbmps), and 122.367 Pbmps in multiple streams. NetBSD was used once more due to the ``scalability of it's TCP code''. More information about this record including the NetBSD configuration can be found at: http://proj.sunet.se/LSR3-s/ for single stream and http://proj.sunet.se/LSR3-m/ for multiple streams. And the website of the Internet2 Land Speed Record (I2-LSR) competition is located at: http://lsr.internet2.edu/.

  11. deprecated ! RC3 has landed by Anonymous Coward · · Score: 0

    Well, RC-stage was a bit long to come, but wow, things are going pretty fast: RC3 is now tagged !
    source http://netbsd.org/

  12. Re: I know which BSD is the cleanest... by Cheesy+Fool · · Score: 0, Offtopic

    Nope, it's nitwit (Fawlty Tower's reference).

    --

    Hail to the king, baby!
  13. Re: I know which BSD is the cleanest... by Anonymous Coward · · Score: 0

    what is witnit?

  14. Cleanest OS by uid100 · · Score: 2, Interesting

    I believe that the term "cleanest" is in reference to one of NetBSD's principle goals of creating truly portable code. Code which compiles and runs on as many different arch as possible with a minimum of #ifdef and such.

    --
    ...yup...
  15. yay threading by ChristTrekker · · Score: 1

    I don't know much about SMP programming, but that threading has to be really up to par. I've been looking forward to 2.0, even though I have only a uniprocessor box, so that I can recompile Apache to use threads rather than prefork.

    --

    Constitutionally Correct