Submitted by Matthias Scheler, on behalf of the NetBSD Release engineering team, announced that the release process for NetBSD 3.0 has officially begun. The first release candidate is available for download
here.
It is expected to have RC2 available on the FTP server in about a week, which hopefully will be the final release candidate. So if all goes well the release of 3.0 is approximately three weeks away.
Always nice to see the NetBSD project moving forward. But couldn’t help to wonder what exactly warrants the major version bump (2->3), and so soon after the 2.1 release? Well, here ya go:
http://www.netbsd.org/Changes/changes-3.0.html
Don’t know why, but this kind of documentation always feels hard-to-find on the NetBSD site ;-(
Doesn’t seem much revolutionary stuff to me, mostly added drivers, support for newer chipsets and (relatively) small subsystem changes. Oh, and Xen 2.0 support. More evolutionary, but maybe it just looks that way…
Enough ofcourse to break things and do this in a development branch, but major version bump? Ah well, what am I bitching about? No personal experience with NetBSD, but still one of the more interesting OS’es out there IMHO. Anyone know some up-to-date liveCD’s ready for download (say, based on 2.1)? And I don’t mean ‘bootable install-disk’, but something to explore a bit, a la Knoppix?
OpenPAM support can be considered the most noticeable change.
Well, the NetBSD folks say that http://www.netbsd.org/gallery/products.html#netbsd-live“>one …
ftp://ftp.netbsd.org/pub/NetBSD/misc/xtraeme/
NetBSD Live CD based on 3.99.7 (-current development branch, but some months old)
It is i nice demonstration of a NetBSD desktop. My only gripe being that you have to do the following to get KDE to start up:
echo “exec /usr/local/bin/startkde” > ~/.xinitrc
For a livecd, that should already have been done for the included user.
One of the nice things about NetBSD (and I’m not sure how much longer it will be the case; see below) is that damned near all supported drivers are built into the kernel, meaning that supported devices work out of the box, with no configuration or messing around. Sound works out of the box, as do USB mice etc., and they don’t require some external (to the kernel like kudzu) hardware detection in order to get things working. Because of their elegant driver framework, including most of the drivers in the GENERIC kernel doesn’t eat up a fantastic amount of memory or disk space.
There has been more and more talk on some lists regarding making the NetBSD kernel more modular; specifically to build more of it as modules like FreeBSD, DragonFly and most Linux distributions do, which I think would be quite a shame because what they have now is relatively small and works beautifully, with no messing around.
Sometimes things that seem like a good idea (modularization) do not in reality, bring progress. But I digress. NetBSD is a fantastic system for people who like Unix.
Edited 2005-11-19 19:55
I tried the live-cd based on 2.0 a while back and it’s worked quite well out of the box (don’t remember having to do anything special to run it). It must be the slowest live-cd I’ve ever seen (by far), but hey, it worked !
to answer the previous posters – netbsd 3 has been in development for a good while now- the releas e of 2.1 and 2.0.3 are differet development strands. people have been using 2.99 (pre 3) for its Xen suppor for a hile now.
hope that guy that did the netBSD-office cd grabs this and rolls out a new “easy” office netBSD offering….
Did that make any sense?
Most significant change for users is that NetBSD now uses PAM for authentication. Purists object to that, I’m sure, but getting older NetBSD machines to work with LDAP authentication wasn’t necessarily the easiest thing in the world.
Otherwise, it’s more of what you’d expect from NetBSD. Stable, clean, etc. The only real “bug” I’ve encountered is that altq does not work with pf. From reading the mailing lists, this has to do with the decision to continue supporting ipfilter, but it’s a bit of an annoyance. There is a patch to fix it, but it then breaks ipfilter completely (which I don’t care about, because I like pf better).
Heh. I’ve read nothing but good things about pf, and all four of the big BSDs have it. Perhaps it’s time for FreeBSD, NetBSD and DragonFly to push it as the default now, instead of IPFW2 for FreeBSD and DragonFly, and IPFilter for NetBSD. It seems that the other BSDs adopted pf cause it’s better, not to give people a choice… but then, I could just be insane…
AFAIK, FreeBSD has dropped support for ipfw, as it doesn’t work with the new SMP/Scheduler stuff very well.
Ipfilter has licensing issues, which is why OpenBSD created pf.
PF is nice in that everything is done in one place (filtering, prioritizing, nat). The only annoyance I’ve had with it is that the ftp-proxy program needed doesn’t work very well with some older windows FTP apps (causing me to have to switch one of the network gateways back to loonix at my former job).
I think the FTP proxy app was just replaced. Try it with the new one.
Actually, FreeBSD have IPFW2 and pf as the main firewalls. IPF has some issues with SMP that is well known. IPF is till supported tho but not recommended for SMP.
IPFW2 works fine with SMP and can (due to lack of support for dummynet by pf) do a few tricks pf can’t do and vice versa
AFAIK, FreeBSD has dropped support for ipfw, as it doesn’t work with the new SMP/Scheduler stuff very well.
You have things completely backwards. It’s IPFilter that doesn’t work well with SMPng on FreeBSD 5+. IPFilter works fine on UP systems, but causes all kinds of issues on SMP systems.
IPFW is the primary firewall in FreeBSD. And PF is supported in FreeBSD 5.3+. Which one you use is up to you.
IPFilter is on its way out.
I’m curious, as I haven’t seen any benchmarks in a loooong while: anyone know of any that compares the BSDs and Linuxes and is recent?
I’d also like to see some benchmarks, especially using a web server and httperf. I did my own tests with that and the difference was dramatic (linux-2.6.13 was like 250% faster than FreeBSD-5.4), but being no expert and owning just one computer those tests might well be completely wrong (I’m not biased towards one or the other, just wanted to find out how they performed as web servers). By the way, I used Lighttpd as a web server, not Apache.