Dec 12, 2009

NetBSD: Bankrupt Software Distribution

Hot on the heels of their NetBSD 5.0.1 release, the NetBSD organization is gearing up for NetBSD 6.0, due in about half a year ("The sixth major release for the six month of 2010!").

To make that happen, NetBSD is asking its industry partners, users, and anyone with spare change to contribute US$60,000. Matt Thomas, of NetBSD's core group, says the money will allow for "network performance improvements and embedded and realtime optimization," meaning NetBSD can finally move onto specialized hardware, an area NetBSD has struggled with in the past.

But is the ultimate goal of $60,000 appropriate for the BSD family's middle child? Is NetBSD a good long-term investment of that kind of money? Is NetBSD even worth the nearly $8,000 raised so far? Why invest in a bankrupt operating system?

To answer any of those questions, we need to identify NetBSD's target audience, though that is about as clear as obfuscated kernel code since NetBSD focuses on tiny or rare hardware. It's still mainly optimized for the VAX and Motorola 68k platforms and has problems handling over 512 MiB of RAM, so NetBSD's not for anyone developing on cutting-edge hardware. They're pretty squarely stuck in 32-bit land, so NetBSD is also not for anyone interested in high-end servers or even low-range workstations.

This all goes back to the fact that the NetBSD team was a splinter of developers from the FreeBSD team who were unhappy with the mainstream track the operating system had taken. They took some code from FreeBSD and vowed to bring it "back to its roots," reintegrated some of the 386BSD project, and continued tinkering in contentment.

So a bunch of tinkerers and code-hoarders on antiquated hardware serving their own interests doesn't resemble anything with real commercial interest. And to further the point, anyone with a real interest in a professional BSD would either choose FreeBSD or Mac OS X. Clearly, then, NetBSD is focused on noncommercial hobby-hacking and research. So does the research ever trickle down—or up, for the matter—into other projects?

According to the last several FreeBSD audits, the answer is an unqualified no. Of the almost 4 GiB of FreeBSD source code, just under 1%, some 40 MiB, of code was in-flow from NetBSD. That means that basically nothing from NetBSD has either been put or taken back by the FreeBSD project. That belies a lack of need, since FreeBSD is more or less like NetBSD on steroids.

This clearly demonstrates that there is a lack of professionalism on the NetBSD team's part: you might think they'd want to share their stuff back, but they simply haven't. In fact, they even deleted sendmail from their source code over a political tiff. Wtf!

To test out what this means in the real world, I installed both NetBSD 5.0.1 and FreeBSD 8.0, the latest respective production releases, on my 2.93 GHz HyperThreaded octocore workstation with 32 GiB RAM.

FreeBSD installed quickly, ran eve faster, and never crashed while supporting all of my hardware. NetBSD hung on install, ran sluggishly, and had no support for any of my cards or peripherals. In fact, it only saw eight of the sixteen logical HyperThreaded cores and 4 GiB RAM. I wanted to throw the machine out the window after an hour of running NetBSD.

Forget games or serving anything too. FreeBSD excelled at serving, since its threading model matches mysql natively. It started both KDE and GNOME in a heartbeat. NetBSD's threading model is old and outdated and complex POS sites barely trickled into my browser. FreeBSD was also able to run some pretty impressive OpenGL games while NetBSD fell down once again and made me shake my head in frustration. NetBSD was a truly wretched experience.

If this wasn't enough, even one of its four original developers says that NetBSD sucks! With such a ringing endorsement, it's really time to question what is going on and why anyone should take NetBSD seriously anymore.

On a positive note, at least NetBSD is not OpenBSD, an even more obscure fork run by a megalomaniacal narcissist named Theo de Raadt who personally harasses and stalks anyone who doesn't agree with him and his coding choices. But when the best point for NetBSD is that it's not something worse, it's hard to say that NetBSD is worth anything at all.

And there are other, better options. Neither Apple nor the FreeBSD Foundation are asking for tens of thousands of dollars in handouts. Download and install FreeBSD for free or shell out a trifling US$30 for Mac OS X—both products just work, unlike the travesty that is NetBSD.

In the end, the only sensible choice is to not donate to the NetBSD foundation. Supporting this moribund operating system is not worth the money—save yourself the trouble and avoid the sinking ship that is NetBSD.

17 comments:

  1. You might check a few of your facts before you post:

    1. FreeBSD was a fork of NetBSD which appeared earlier.

    2. NetBSD has supported 64 bit hardware for longer than Linux or FreeBSD.

    3. NetBSD is not mainly optimised for the VAX or m68k, that's frankly absurd and belies a shocking lack of knowledge.

    4. NetBSD does not have problems with more than 512MB RAM, in fact my NetBSD laptop on which I am typing right now has 2GB.

    5. Sendmail was not deleted as part of a politcal tiff. It was deleted because it suffered from continual security problems over its long life and was deemed to be low quality software that shouldn't be shipped with the OS given that there was already a perfectly reasonable replacement.

    There are more errors in your post, of course.

    You may very well have complaints about NetBSD as I am sure that many people do. But, before you post them, it might be interesting to do a little tiny bit of research and try to keep the complaints within the realm of the true.

    ReplyDelete
  2. If you were correct, I would be really amazed that the high-speed records in long-distance networks are regularly achieved with NetBSD.

    You are, as the previous poster pointed out, mainly incorrect, of course :)


    PS: I don't use NetBSD, but I like to check my facts before going on a rant :)

    ReplyDelete
  3. Horrible article -- I'll skip the lies and just answer the question whether it worths or not.

    With the money that were raised during the previous donation campagne, TNF hired Andrew Doran to work full time on SMP and his efforts resulted in NetBSD's 5.0 great SMP performance.

    So, yes, it made a huge difference.

    ReplyDelete
  4. The article is so wrong, on multiple levels.

    ReplyDelete
  5. So What! Can you build it better yourself ? Or better yet why don't contribute code to the project to improve all the issues you've pointed out.

    ReplyDelete
  6. wow, the troll has an axe to grind.

    might want to have look at http://www.netbsd.org/gallery/products.html

    Plenty of hardware has trouble with various *BSD distros, I had motherboard that wouldn't boot openbsd or freebsd 5.x, but netbsd did the trick. proving exactly nothing same as your little whiny anecdote.

    So you think FreeBSD is cat's meow. Since 6.x days still carries warning that their overcomplicated and unordered smp locking mess can sieze up tighter than lead in a pencil, I've done that on 4-way and 8-way systems. Maybe someday FreeBSD team will unfutz their crapball, or do what NetBSD does and go for correctness first before rushing into high profile but little used scaling

    ReplyDelete
  7. Wow. Just, wow! Your recommendation to buy Mac OS X for the one kind of computer it runs on, only it turns out that a goodly bit of the userland and some kernel interfaces come from...ahem...that project. Gee, some people don't even try, do they? And yes, it's not to say that NetBSD is perfect, but it's my favourite *BSD just *because* they try, and try hard, to make things right. Read the code, judge for yourself; networking, excellent multiplatform support, redeeming POSIX interop/2E support, splendid integration of new processor features (NX, VT). It rocks, even if it won't go on my desktop systems as hardware is, yes, not right at the top. Oh, and my Apple Airport Base Station/Time Capsule is running, uh...work it out for yourself.

    Cheers,
    Sabahattin

    ReplyDelete
  8. This rant is mentally retarded. It claims that OpenBSD is more obscure than NetBSD when it fact FreeBSD and OpenBSD both have more users and are more well known than NetBSD. He attacks Theo personally when in fact objectively many security experts say that OpenBSD is the most secure operating system on the planet and financial giants rely on it to protect their gateways. Also NetBSD is the most flexible and customizable OS I have ever used. That is nothing to sneeze at. I hope Trollaxor gets hit by a train and wins a Darwin award for it.

    ReplyDelete
  9. FreeBSD was not a fork of NetBSD. NetBSD and FreeBSD both grew out of the 386BSD patchkit days. Years later, OpenBSD forked from NetBSD.

    ReplyDelete
  10. @Anonymous 24 Jan 2010 21:08, I find this is a good read for OpenBSD-is-super-secure idiots:
    http://allthatiswrong.wordpress.com/2010/01/20/the-insecurity-of-openbsd/

    As for the author, shame on you. Get your facts straight before you post anything.

    ReplyDelete
  11. No, anonymous, your article criticizing OpenBSD was written by someone ignorant of security, systems administration, and BSD in general and is a "bad read". Besides the core installed system, the packages (that commonly used subset of the ports for which they provide binaries) of OpenBSD are also audited and have fixes applied by the development team.

    The internal ACL-type, attribute-type and SE-Linux type systems touted by that author provide no additional security whatsoever and only serve to fine tune access rights by authenticated users. Proper policy making on such systems proves too complex and security holes are invariably created. They quite often interfere with programs (not to mention are buggy themselves and so open additional security holes) and so are usually disabled in practice anyway. They don't harden against internal exploits nor external attacks. Piling on an access control framework on top of a system not designed with security in mind doesn't make a secure system

    ReplyDelete
  12. You have a huge number of falsehoods here aside from the totally incorrect assertions about the origins of NetBSD, 64 Bit support, and others that have already been pointed out. FreeBSD's /etc/rc system is taken _directly_ from NetBSD as well as many critical drivers. It might not make up a high percentage of the code, but critical and complex code within the kernel is often not large (take the scheduler, for example). The only 1% I can see is the amount of truth or factually correct statements you've made.

    ReplyDelete
  13. According to another post on your blog you just gave them 100,000 USD. yet you imply that the NetBSD project is a joke and a “travesty”.

    I think you and this entire website are a lying waste of electrons.

    ReplyDelete
  14. I think everyone who replied here has successfully been trolled. Thank you for entertaining me.

    ReplyDelete
  15. If you knew how much stuff runs on NetBSD at Nasa and Skunk Works. Your head would implode.

    Almost most ISS experiments are driven from NetBSD.

    ReplyDelete
    Replies
    1. The author does not know anything about BSD destros except the names of the destros.
      Free BSD - Desktop general purpose (competes more with linux than with other BSDs), Good free documents are available for users. No need to buy any books.
      OpenBsd - Security, Secure servers, Several credit card companies run their systems on openBSD.
      netBsd - flexibility, embedded systems, support for most hardware, performance.

      Most of the computing and performance related research is one on netBsd.

      All these *BSDs are very good. Each one is more suited for different needs.

      Delete