Slashdot Mirror
NetBSD Q3/Q4 Status Report Published
Anonymous Reader writes "The NetBSD Foundation published its first quarterly status report in 2006, covering the months July though December of 2005. Among many other things, this status report includes the release of both NetBSD 2.1 and NetBSD 3.0, a summary of the NetBSD Project's participation in Google's Summer of Code and the release of two stable pkgsrc branches."
In 10 words? Sorry, no.
FreeBSD and its fork Dragonfly focus mainly on high-performance production servers, a little like Linux.
OpenBSD focuses on the best-ever security on earth. This also includes clean code and good documentation (though even FreeBSD is quite good, IMHO compared to Linux/GNU).
NetBSD focuses on exceptionally clean code/architecture, excellent documentation, but also on interesting features, and of course awesome portability. This includes having very clearly-written, modular drivers, where other systems sometimes only have drivers for, say, a specific device when it's behind a specific bus, but not in general...
I like NetBSD, because it's fast enough IMHO, and very clean and well documented. If you want maximum performance, and maybe more multimedia drivers, get FreeBSD; if you're paranoid, use OpenBSD.
...despite their relatively small developer and user community.
The Q3/4 status report indicates what seems to be the modus operandi for the NetBSD team: consistent incremental gains.
I have an immense amount of respect for the NetBSD project. OpenBSD drives the BSDs (and Linux) to be more secure. FreeBSD traditionally has shown us what a great administrative user experience should be like. NetBSD continues to show us the way with respect to proper system architecture.
For example, NetBSD and FreeBSD (and OpenBSD?) natively support the same wireless hardware that Linux does. The difference? I can configure WEP and/or WPA through the exact same ifconfig that I use to configure a wired ethernet interface. No madwifi drivers. No 'download' wpa_supplicant. No difference that I'm setting up "different" network hardware. It's all just network hardware.
In my opinion Linux's weakest point is its kernel. The userland is great for the most part, but the kernel and the parts of the userland that deal directly with the kernel seem to be its major flaw. To follow the networking example: because the underlying wireless system is so fragmentary in Linux, NetworkManager (a good attempt at a friendly gui network profile configurator) feels like a bit of a bubblegum and bailing wire solution. This isn't NetworkManager's fault, it's Linux's for not providing a consistent system API for wireless. In NetBSD (FreeBSD & OpenBSD?) this isn't the case.
Alas, Linux (and it's collection of cool features like boot splash screens, polished user interfaces and installers, good binary OpenGL video drivers, great hardware detection utilities, commercial support on the server side, native Sun Java support, etc.) enjoys ubiquity while well architected systems like NetBSD languish in relative obscurity.
The cool Linux features often feel hackish (have you ever built an isolinux splash screen? NetBSD has always struck me as a natural choice for building a user-oriented/workstation distribution. Some of the little features are missing in NetBSD, but they could be added easily by a team focused on such a task. If a Mark Shuttleworth style billionaire pulled an Ubuntu with NetBSD, I think the world would generally be a better place.
-Peter
. Penguins Surely Ca