Slashdot Mirror
FreeBSD 5.2 Review
JigSaw writes "OSNews published a review of FreeBSD 5.2. They found the OS very solid as a server but pretty lacking as a desktop. The author finds FreeBSD very fast overall, easy to configure and that it feels integrated and mature. On the other hand, it has limited modern hardware support, small annoyances at places and that not many binary packages are available and so compilations from ports may take long time."
They found the OS very solid as a server but pretty lacking as a desktop.
So, are they going to refer people to Windows as something with a "good desktop".
Operating systems, to me, are a lot like buildings. The kernel is the foundations, and everything that sits on top of it are floors in the building. Where would you prefer the weak link to be? Near the bottom of the building, where tapping on a support column on the first floor makes everything come crashing down (read, BSOD)? Or would it be better to have parts of the building that are higher up be weak, in which case part of the building is still left in tact?
I realize that this analogy isn't entirely true, as with the WTC the weight of the top part of the building entirely decimated the bottom part. But supposing that the bottom underlying part of an operating system is bulletproof, all the abstracting layers on top of it that come crashing down won't kill the whole thing.
I see this as BSD. They're making sure their foundations are rock solid before building on top of them. It's good practice. The rest of the infrastructure will come with time.
/^[A-Z0-9._%+-]+@[A-Z0-9.-]+\.[A-Z]{2,4}$/i
We still have a couple print servers around here that are running Pentium Pro's with FreeBSD 3.4 from five years ago. Yeah its probably time we replaced them, but they've been reliable.
I mean for desktop, we use Mac OS X because that what its designed for, at the end of the day its the right tool for the right job.
"The problem with socialism is eventually you run out of other people's money" - Thatcher.
I love the install. I've been using FreeBSD for several years now, and I decided to throw RedHat on a machine to check it out. Suffice it to say, I won't be making that mistake again.
Wellllll, it's like this. Imagine that Linux is a VW Beetle body. It comes off the assembly line in one version only. To acutally use it you have to add the same type of wheels and tires. The engine is basically the same model but there are thousands of hop-up parts and techniques you can use. Then hundreds of different distributors mold different fiberglass shape-changing parts to these things, and give them stunning airbrushed clear-coated paintjobs, and send them out. That's Linux. All the same and all different. Just choose the one you want. If it needs fixing, you'd do best to take it back to the dealer.
FreeBSD is the same, except you pick up a body, a box of fifty million optional parts, a free set of tools, and a manual. There's no dealer, but it's designed for self-maintenance, so one day you realize you don't need a dealer.
Each has its place and its audience. It's all good.
The reviewer does hit a nail on the head, "If you are after an easy-to-use desktop system that doesn't require you to learn anything new, then you better look elsewhere."
This is the arrogance/beauty of FreeBSD, it is designed/engineered/distributed as an O/S to get the job done like no other. The Bauhaus school of software design. It is an SOB to get a new user going on, but once they see the light, good luck prying it from their hands. Good things are rarely easy.
The best thing ever to happen to FreeBSD was Linux, the best thing ever to happen to Linux was FreeBSD. A good, clean, honest competition which leaves both sides stronger.
"To those who are overly cautious, everything is impossible. "
I remember a few years ago, I read that FreeBSD was far superior to the Linux kernel for a heavily-loaded server. Supposedly you could run a server at about 100% CPU load, for days, without any problem if you used FreeBSD, while a Linux kernel would have problems.
Now that Linux 2.6 is released, has Linux caught up with FreeBSD, or is FreeBSD still better?
(And play nice, folks, please. I'm not trying to start a flame war here.)
steveha
lf(1): it's like ls(1) but sorts filenames by extension, tersely
For me, the advantages of compiling everything from scratch justify the investment in compile time. If I can get a piece of software to build on the same machine that I intend to run it on, I've found that I have far fewer problems overall. Most potential problems get caught at compile time. This is what kept me using FreeBSD/OpenBSD for years.
Now that I've been using Gentoo for about a year, I've come to believe that this is 90% of what made the BSDs better than any linux distro I'd used back in 1999. If you want a program you build it... it's built for your architecture, with your optimization settings, exactly the way you want it. If the program you want isn't in ports/portage, you can usually add it yourself by changing a couple lines in an existing port. If the developer updates the source to a program without changing the build process too much, you can just rebuild. No need to hunt down rpms or debs.
Because you're building your own binaries, you're also afforded some small amount of protection from scripted security exploits that target known builds of programs... but that's another subject.
Sadly, I took one glance at the screenshot in the upper-right hand corner, and knew I would be reading a Eugenia article.
/. posters. FreeBSD 5.2 is rock solid, as any Unix, Linux, box would be. Every port in the ports tree has a pkd_add to go along with it. That's 10,000 precompiles binary ready to download. They all install, deinstall, with near zero user interaction.
She continues to base a useable desktop by how many windows she can open at once. Skip the article, and read the reviews from the
In short, use FreeBSD.
I really hate Dan Patrick.
Okay, so my question is, if Linux is such a superior desktop, why doesn't it have a greater marketshare?
This is a faulty argument. Quality does not have a 100% correlate to marketshare, and often marketshare has a very small correlation to quality. A 100% correlation assumes a market with zero friction everywhere. That market does not, and can not, exist. The current OS landscape (partly technical reasons, partly others) is very very far away from a zero friction marketplace.
Disclaimer: I have only read the first page of the article. What he says about kuser is true, it probably should /not/ have been included if it was a known problem. But also bear in mind that 5.2 is Beta quality code, and is a testing release. Not a production release. All the same, I'm here to chip in my two cents.
/do/ have to take disagreement with the 'not ready for desktop' bit. FreeBSD seems to take this attitude that if you take the time to learn something, it will pay off. If you're going to use ports, check out sysutils/portupgrade. Once you learn how to use it, and how to use pkgtools.conf, you will save yourself *hours* of configuration and twiddling time. Just tonight, I spent two hours swapping versions of OpenSSL on a web server (hosting about 200 sites), and Apache was down for all of about 45 seconds. The ports tree is one of the best things about FreeBSD, and I don't see any binary distribution system even coming *close* to competing with it.
;)
First off, stability. It's true, FreeBSD is *rock* solid. I used Solaris and Linux (Debian/Slackware/Redhat) before coming to BSD. And I am astounded daily by the differences between the three. Our residential web server at work has been slashdotted three times in the past year, and the only reason we noticed was because someone saw the posting on Slashdot. And it's not fancy hardware -- P4, 1.6GHz, 512MB RAM. And while being slashdotted, we were still doing work on it, without really noticing the load increase.
But I
(And yes, I have used apt-get extensively. I have not, however, used Gentoo yet, but I have heard that its packaging system does indeed rival the ports tree.)
I use FreeBSD on my desktop (obviously). And to be quite honest, I could easily port my environment over to a Linux desktop, and not really notice a massive difference in functionality. If an app compiles on FreeBSD, chances are, it compiles on Linux (and vice-versa). I just use FBSD because I'm used to it, and because I know it a bit better -- it's more comfortable for me. Purely from a desktop usability viewpoint, I don't think it's much worse or better than any given Linux distribution, so long as I can keep my current setup. But again, it's that ports tree that sets it apart.
I, as well, moved to FreeBSD from Linux. That was shortly before 2.4 came out, and I still don't trust any kernels beyond 2.2, really. But I run 5.2-R and -CURRENT on my two desktops, and 5.2-R on one of our servers. Yes, it has its gotchas right now, but remember: this is considered Beta quality code right now. If you're looking for something slick, together, and perfect, wait for 5.3.
Not that FreeBSD isn't already that.
I wouldn't say that Linux is more stable or anything, but don't pretend that BSD is perfect. Even OpenBSD has its little idiosyncrasies, and it's supposed to have had more code review than any other open source Unix(-like OS).
Windows NT does certainly have plenty of security issues, but in general, it is a fairly reliable operating system. While in some ways it has gone downhill in that regard since NT 3.51 (though it's gotten better since 4.0, FWIW) because of decisions made to improve the speed of the system, NT is built on a solid architecture. The problem with NT is not the design of the foundation so much as the implementation, and the fact that PC hardware is, for the most part, utter crap, and it makes a commitment to supporting (nearly) all of it in a way that enables hardware developers to publish closed drivers and yet still have them work and deliver good performance, and not be tied to an individual version of the kernel. You can service pack NT and end up with upgraded hal and kernel, and still use many of the same drivers across NT 5.0 and 5.1's assorted builds.
Obviously NT is not without flaws, but they are not (in general) as great as you suggest, especially given what they are trying to do with it. And clearly, Windows is improving dramatically. I expect XP sp2 to help a great deal. And, as Microsoft implements more of the non-core functionality of the OS in .NET, it should ease transitions from one version to another, because they will be depending on a fixed API rather than juggling things around so heavily with each new version, mostly because the applications will not be coupled so closely to the OS itself.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"