An interesting summary from the 2004 Annual NetBSD Group Meeting was posted to the NetBSD -announce mailing list. The report discusses past achievements as well as future goals.
NetBSD: Report From the 2004 Annual Meeting
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Indeed NetBSD has become much more viable a choice after FreeBSD’s choices during 4.x->5.x transition. NetBSD 2.0 came out very impressive both in terms of performance and code maintainability whereas FreeBSD’s approach is vastly criticized (and Dragonfly fork occurred). See for instance:
http://www.feyrer.de/NetBSD/gmcgarry/
On top of this FreeBSD had problems with KSE+ULE scheduler
yet NetBSD’s scheduler activation performed very solid. I wonder how both of these systems would compare in terms of SMP. I remember NetBSD’s code was mentioned to be mostly still under giant lock.
I only wish that, just like NetBSD team is doing right now, FreeBSD developers somehow reflect on their approach and reiterated their goals especially drawing some lessons from DragonFlyBSD’s successes.
…I will try NetBSD on one of my 16MHz Macs, just for fun
NetBSD is just great. I don’t have an SMP system, but it’s great anyway because everybody is talking about SMP. And some time in the future I’ll have an SMP system,too. And than it will all pay off what others are talking about now…
NetBSD has SMP.
Erg, accidently submitted too early… As I was saying…
NetBSD has SMP. The SMP is simple though, it is the same kind of deal with OpenBSD, a very inefficient way of doing it. If you’re looking for good SMP you’re out of luck.
Give me a way to stay up-to-date without compiling from source.
Please update more frequently than “quarters of year” the pkgsrc.
Example: If Firefox1.1 is released sould haveI to wait for 3 months ?
And I will switch immediately to NetBSD.
I can’t compile from sources, my machine is “slow”.
>I can’t compile from sources, my machine is “slow”.
You have to watch it compile ?
$ make
Now go to bed.
Or join the mailing lists. Get some people you meet there to compile them for you.
Just follow the latest and greatest pkg_src instead of fixed version and you’ll be fine.
Regarding compiling you can use binary packages (pkg_add <url>).
Anyone know how it went with the nvidia drivers in NetBSD? Only “early” hits I get with google are from 2003. Can you use the FreeBSD ones or not? And also I wonder how hard it would be for Nvidia to make drivers for NetBSD aswell since they already do FreeBSD ones and I think there is a special version of them for DragonFLY now aswell.
(also cedega would be great
)
there are plenty of filesystems for linux, and personally i think a good, relyable, journaling, fast filesystem is a no. 1 feature for any server system. the bsd’s ufs2 gets a little bit outdated…
there are plenty of filesystems for linux, and personally i think a good, relyable, journaling, fast filesystem is a no. 1 feature for any server system. the bsd’s ufs2 gets a little bit outdated…
I don’t know why UFS wouldn’t fit. It’s good, it’s reliable, it’s fast. It’s just not journaling. But journaling is a means, not a goal (btw on UFS you have soft updates, which serve a similar goal).
And outdated, I’ll interpret that one as proven, which is another important factor for production use.
I’ve had a lot of bad luck with journaling filesystems on linux. Reiserfs has died at least 5 times under home workstation and small server conditions as has ext3 w/metadata journaling. On the other hand, FreeBSD’s UFS and now UFS2 (with softupdates) have been rock solid under the same conditions since 99 when I started using FreeBSD. I haven’t used NetBSD for any length of time, but since the filesystems share a very, very similiar codebase, I would assume similiar performance.
On a side note, I have never had problems with XFS under various Linux distros, and that seems to have more problems than any of the four major journaling filesystems in linux. I guess my workload is atypical.
In the beginning (around when I starting using them at work full time in the early 90s), there were two camps – linux and bsd. one turned into a industry unto itself, the other splintered into four projects (open/net/free/dragonfly) with dubious futures.
OpenBSD seems to have the best reason to live right now – at least they have a clear viable differentiating point from linux.
NetBSD is great because it works on all the hardware you don’t own nor will you ever own because you don’t buy computers from NASA garage sales.
FreeBSD has turned political and periodically releases terrible products (3.x, early 5.x) even though it claims stability above all else.
DragonFly is what FreeBSD should have been, but really, are you going to run code that someone built as a protest project?
Don’t talk about what you don’t understand.
NetBSD is not just for portability. Their code is incredibly clean and the fact that it has to remain portable insures it will stay that way. This has many many positive side effects.
FreeBSD took on a huge project with 5.x. They have had poor releases…so has everything else. I personally am very mixed in my views on their design for 5.x, but have no doubt they will get their stuff together. FreeBSD 4.x was a great OS, they will get back to that.
“are you going to run code that someone built as a protest project?”
You have _got_ to be kidding. Do you realize how many software projects started as “protest” projects? Please…
>You have to watch it compile ?
>$ make
>Now go to bed.
Ok.
> Regarding compiling you can use binary packages
> (pkg_add <url>).
But suppose I’m going to install OpenOffice:
ftp://ftp.netbsd.org/pub/NetBSD/packages/pkgsrc/misc/openoffice/RE…
ftp://ftp.netbsd.org/pub/NetBSD/packages/pkgsrc/misc/openoffice-li…
I can’t find any binary packages for it.
What am I missing ?
AFAIK, NetBSD doesn’t have native binary packages for OpenOffice but you can use the package source tree (which you will first need to download) to install the suse9 meta-package and the openoffice-linux package. This will give you a binary installation of OpenOffice.
http://www.netbsd.org/Documentation/pkgsrc/
http://www.netbsd.org/guide/en/chap-linux.html
“You have _got_ to be kidding. Do you realize how many software projects started as “protest” projects? Please…”
True that! For instance, wasn’t Linux a “protest” project? Wasn’t Linux created because Linus thought that he can do a better Minix? Didn’t he have those heated discussions with his professor and coded Linux just to “show” him?
I’m sure there were other reasons that he wrote Linux, but I’m sure the above was one of them. At least that’s what I read.
This goes to show, that “protest” projects can be a good thing. You definately have to give credit to those who think they can do it better and actually do, rather than those that whine about it not being better.
You know what I mean?