Slashdot Mirror
Jan Schaumann Talks About NetBSD on the Desktop
An anonymous reader writes "Continuing his series of interviews, Emmanuel Dreyfus asks NetBSD's Jan Schaumann about his experience with NetBSD on the desktop. From the article: 'Jan Schaumann has been an important contributor to the NetBSD project for several years. He spent a lot of time working on the NetBSD package system, known as pkgsrc, and he currently uses NetBSD as his desktop system. We will try to learn from his experience during this interview.'"
I have been saying this for a long time. Basically to sum it up. Linux (and netbsd) ARE ready for the desktop. Because the end user wouldnt be installing linux, just like they dont install/upgrade windows. Someone else does that, the administrator, or the kid down the street. The administrative details can still lack, but that is immaterial since the person doing the work is already knowledgable (in theory) As long as their is an easy to use GUI available that makes it easy to get to their mail, the web, and possibly type something up, that satisfies most people's requirements.
The phrase "more better" is acceptable English. suck it grammar Nazis
I used NetBSD as my desktop for over two years, and didn't have any usability issues. Thunderbird for e-mail, Firefox for browsing, OpenOffice for the occasional resume tweak. Plus all the "standard" FOSS stuff: Gimp, Apache, Tomcat, Ethereal, gAIM, etc. VLC for the (very) occasional MP3/DVD playback.
Granted, I'm more of a pure software developer (I don't game, and I don't use my machine for "media" too much), but I can't recall a time when I got "stuck" because I didn't have some piece of software available. I believe both KDE and GNOME are available (I used AfterStep), so there shouldn't be too much confusion switching a Windows user over.
Just junk food for thought...
yes because you only ever install just the linux kernel. jackass.
The phrase "more better" is acceptable English. suck it grammar Nazis
clearly parent is a troll, but a useful point nonetheless. too many distros, too many bsd*s, too many people too willing to run off in their own direction. as a concept, open source has delivered a wide range of altenative to commercial unix, a few theoretical alternatives to Windows for primarily computer hobbyists and business, but has failed to deliver the mind storm that will render the model viable going forward. i think fragmentation is to blame, and sadly the financial crisis affecting openbsd, waning interest in netbsd and a proliferation of truly uninspiringly different linux distros and desktops is a sympton and the type of headline we're going to have to get used to until my demands are met. no more passing off an afternoon's entertainment as a supportable software project.
The only problem is that my desktop is a VAX 4000-90
But yes, compared to VMS on the same hardware, NetBSD is WONDERFUL!
I've been hearing myself say that for years, and it's been true. But we're increasingly concentrating more on our data and, the people with whom we exchange it, than on the applications we use to do that. The maturity of "personal computing" has evolved a short list of apps which resemble each other regardless of developer, with similar UIs. But a diversity of architectures, from phones to notebooks to desktops to big iron, often several of which participate in any one transaction across the network. Yet the app paradigm inherently creates boundaries across which people must communicate, which often doesn't work and is always complex - even when "integrated". While cross-platform Web apps and inclusion of millions of "unsophisticated" users create a demand for things to "just work", without requiring "computer" skills in addition to those required by the actual task at hand.
In short, "personal computing" is getting to be like driving: most people can use most cars more or less the same, with different performance and convenience, on standardized roads, to get where we're going - mostly to get to other people. Applications are like cars, desktops are like dashboards, OS'es are like transmissions, networks are like fuel types, and our data is like the open road. MIME and desktop integrations are making that data the center of user activity. So the question is decreasingly whether "the" app you need is available under an OS on given HW. Rather, whether an app more or less automatically is available to work with your data, on whatever OS/HW is available and connected to the Internet. Since most of that data is for working with other people, convergence of voice and other data will make a lot of idiosyncratic SW, and unique skills using it, go the way of the Model T.
--
make install -not war
If you feel the need to learn more than you wanted to know about unix and building everything from source, then go for it. Install netbsd and fiddle with the (vastly improved) rc.conf stuff to get things to startup and configure you cards. cvs update the latest security fixes to build you are running and remake,reinstall the kernel/os. Install your packages from source (the first time it takes 20 hours to install KDE is fun!). Or use the prebuilt packages for the stable release. Most of the packages in pkgsrc seem to compile and work ok. You'll have to add you apps to menus by hand. If you want your browser to have java plugins, flash, acrobat and such, you'll need to install the suse compat layer and binary packages.
Many many software packages will autodetect what is installed and enable extra functionality. Sometimes pkgsrc packages will have notes about things you might want to install, but that are not dependencies. So you might end up rebuilding kde or pieces of it if you don't have other optional things installed first, or if the binary packages weren't built with those options.
And you can set up a master image that you carefully check dependencies on and make sure it all works, then rsync the OS changes out.
If all that sounds like the way you want to spend your time then go for it. If you can get a job doing all that busy work and getting paid a good salary, even better.
Otherwise a modern binary linux distro (debian,ubuntu,fedora,centos) will likely eliminate much of the tedium and give you just a more user friendly environment with much less overhead. Tools like kickstart,fai and cfengine allow you easily build and keep many systems up to date with little manual intervention (cfengine also works with *BSD).
just my 2 cents. I manage linux, os x and *BSD boxes.
I've used NetBSD on my old server. It was great once I got my head around pkgsrc and its quirks. I now use pkgsrc on my new server with Slackware Linux, because I heard Linux was better on SMP hardware than BSD, although NB3.0 might have chnaged that.
I'd use NetBSD on my desktop if the NVidia X server would run on it. Does anyone play Unreal Tournament on BSD?
Notice that Solaris moved away from M:N and to 1:1. It is WAY simpler, and thus easier to do correctly and with good performance. NetBSD and FreeBSD are the only ones still trying to do M:N threading, because its too complicated, and offers only theoretical benefits that haven't actually been realized in practice.