Slashdot Mirror
Yet Another BSD vs Linux article
Lazaru5 writes "Technology writer Simson L. Garfinkel wrote this article for the Boston Globe Online Business section. " It's something of an incendiary article, but I think it's great to see the amount of press that *BSD has been getting lately.
Seems you don't get out of the *BSD camp enough to get your facts
straight about Linux. Here are some clues...
> While Linux can only run on a few kinds of computers,
> NetBSD can run on more than 22, including Intel-compatible PCs, Amigas, old
> 68000-based Macs, Digital VAXes, and even those sleek but defunct NeXT
> workstations
Sorry to disappoint you, but Linux runs on just about everything NetBSD
does plus Palm Pilots!
> This has made the system popular at places like MIT and NASA,
> both of which have large menageries of computer systems from a multitude of
> vendors.
I don't know about MIT, but Linux has a stronger presence in NASA.
> Earlier this year there were a number of well-publicized security problems
> involving the Linux operating system. During that time my computer was
> frequently attacked. However, since I wasn't running Linux, I wasn't
> vulnerable.
And there were less-publicized ones for *BSD which means many users
probably still don't know. Atleast the Linux developers can admit to
their mistakes. BTW, their TCP/IP stacks are different and most attacks
aren't portable across platforms. You know, Solaris had a hole once,
however, since I wasn't running it, I wasn't vulnerable. That's so weak.
> Linux is the favored operating system for most of the
> attackers on the Internet, which is another reason I don't use it.
Right. And now that you're promoting *BSD, they'll switch. Going to have
to move to something else then, eh?
> Linux developers, overall, are more interested in just putting
> together something that works.
BS.
> Where the BSD projects are largely the work of individuals, businesses
> are now the driving force in the Linux community.
BS.
> This may be one reason proponents of Linux are frequently slow to
> admit the debt they owe to the Computer Science Research Group at Berkeley,
> which created BSD, and to the Free Software Foundation at MIT, the charitable
> organization that raised money for Project GNU
This is complete BS. Everyone knows who the FSF is. Everyone knows Linux
is just the kernel and all the software on top is GNU. In fact, since you
don't seem to know this, people have been fighting to call it GNU/Linux!
> although Linux is a fine operating system, I would encourage businesses
> as well as advanced enthusiasts to take a serious look at the other choices.
How nice of you to say that after you rubs it's face in the dirt.
I didn't see anything incendiary in the article.
Of course, I am not a religious follower of any of the freenix creeds. I run several of them.
And Windows 98 as well. All have their uses.
Almost anything is incendiary on Slashdot these days.
My impression of this article was that the author was just trying to spread the word of BSD 'nixs.
:)
Every OS has its advantages/disavantages. That's why the real computer professional knows when and where to use each OS.
Slashdot has a tendency to Defend Linux no matter what is said.. Even if it is not "slamming" content.
And you have to admit, Linux is the OS of choice for 13 year old script kiddies.
At any rate, let's keep up the good work on the fight to get OpenSource OS's on top!
Signed,
Avid OpenSource User.
What?!? This is sheer idiocy. All modern UNIXes use SysV init, because it's hands-down better. You have the ability to use the backwards, old-fashioned BSD way by lazily sticking stuff into rc.local, rc.sysinit, etc.
How can you seriously argue that the flexibility that runlevels provide is somehow a disadvantage?
Sheesh. It's stuff like this that makes me think less of BSD advocates. Crabby old men, indeed.
--
Interested in XFMail? New XFMail home page
Since people seem to be interested in a comparison of some aspects of Linux & *BSD, I'll post my experiences & knowledge on the subject. I've been using Linux for nearly 4 years (mostly Debian), FreeBSD for about 1 year and NetBSD for about 4 months, and I enjoy all of them. Also, note that I have these OS's on different machines, not all on the same box. I'm using all 3 on production servers.
/usr/pkgsrc and /usr/ports, respectively, as well as precompiled packages. The /usr/ports and /usr/pkgsrc directories contain directories such as /usr/ports/net/ncftp .. To download the source & compile it, you enter that directory & type make. To install, make install. Any packages the tool you're compiling depends on will be downloaded, compiled & installed automatically as part of the process. 1 caveat here, though. At least on FreeBSD, if you use this for things that require X, it will want to download, compile & install X, even if you've got the "base distro" X installed. I use the precompiled versions of this type of stuff instead. The command-line tool 'fetch', which is the backend for file retrieval works nicely through web proxies, also. :)
/etc/passwd in-place, leaving a networked machine's root account open until copied back (which it does automatically, but, make sure to disable incoming connections during this time) Neither of these methods fetch through web proxies, so you'll have to manually download the files in that case.
;)
;)
:)
I'll start off with installation & package selection. I've exclusively done network installations, so that's what I'm comparing. In this area I've found Debian to be the most robust. For instance, installing on a corporate intranet where only a web proxy is available is no problem as Debian's installation and upgrade tools understand web proxies. If you lose a connection in the process of downloading, Debian will resume at the byte it left off. Also, it allows you to select the primary function of the machine so a corresponding set of packages can be installed.
FreeBSD has a user-friendly (full screen curses) install tool, but it doesn't grok web proxies, so it's a bit more difficult to install onto an intranet. Also, the packages selection tool is nicely hierarchically laid out so you can, for instance, select net -> ncftp to get the ncftp package - instead of having 1 huge screen like Debian's dselect does.
NetBSD's install is more spartan, but still gets the job done with a direct internet connection (I haven't tried it through a firewall/proxy) There was a fullscreen curses install tool last time I installed on sparc, though, so it seems to be getting "easier" on this front.
Neither of the BSD's seem to have an installation-type download recovery mechanism as robust as Debian's in case of a lost connection.
Getting packages after install:
NetBSD & FreeBSD have
Debian is about as easy - apt-get install packagename does the trick.
OS Upgrades: Debian is nice here, as it has friendly package scripts that try to (and mostly succeed in) making upgrades smooth & integrating changes well. Upgrades through web proxies work fine. apt-get update ; apt-get dist-upgrade does the trick.
For FreeBSD the most "correct" method seems to be CVS, but I haven't tried it yet. I've used sysinstall, which still could use a bit of improvement. My biggest gripe is that it downloads
Haven't upgraded NetBSD yet.
Regarding the actual systems, some quick comparisons:
Linux & FreeBSD utilize multiple processors in a machine, NetBSD does not.
NetBSD & FreeBSD both currently support NFSv3 in stable releases. Linux is supposed to RSN...
NetBSD uses 32 bit UID's, Linux is supposed to in 2.4 (as of 2.2.x, it's 16 bit), FreeBSD 3.2 is still on 16 bit, but I'm unsure of their future plans.
FreeBSD can do host-based (i.e., pci controller) hardware raid on the Alpha architecture, Linux & NetBSD cannot.
The BSD's support very large files (don't have the #'s), the largest file size on ext2 is 2gb. (Which is being worked on).
On Alpha, Linux supports quite a few vga video cards. NetBSD only supports TGA. (haven't looked at FreeBSD here)
Documentation:
FreeBSD & NetBSD both have their documentation more centralized, but there tends to be less task-oriented stuff than for Linux. NetBSD's task-oriented stuff tends to be a bit sparse. The mailing list archives are searchable, though, and often already have the answer you're after.
Stability: I've achieved > 100 day uptimes achieved with each of them.
The features overlap and tend to leapfrog quite a bit. Proposing one of the systems is "best" reminds me of the perl camel book's definition of binary: "Some people think in binary. You can tell by the questions they ask. 'Should we program everything in perl or C'".
I think that some people in the BSD community need to recognize that Linux is not Red Hat, btw. A good amount of the negative BSD advocacy I've seen has been pointed at Red Hat problems, not Linux problems.
If anything I've said here is wrong, I welcome corrections.
Garfinkel took a few FUDdy cheap shots (Linux as 'a cracker's OS', too commercial, etc.) but the article was pretty good when he stuck to discussing the merits of the BSD's. In a way, he does everyone a favor by pointing out in a mainstream publication that Linux is a kernel, and that the components of a Linux distribution also come from GNU, BSD, et al. To listen to other articles in mundane media, one might think that Linus wrote the entire OS himself.
---------------------
the SlashDot spellchecker:
---------------------
John 3:16 - God's Public License
The Yahoo! founders are reputed to have had a look at Linux around the time of the 1.x.x kernels, but FreeBSD was a far more suitable option at the time.
This far down the line, FreeBSD is pretty much entrenched at Yahoo! - they have people there who can hack the kernel for specialised requirements, and a huge codebase that would be a nightmare to port.
Chris Wareham
Huh? I though hotmail was based on Solaris (or SunOS?) servers. And doesn't Yahoo use Linux as well as *BSD?
:)
He mentioned that some Linux security holes were found and people were hammering his box, but there are plenty of *BSD holes too. Just because Linux exploits didn't work on a *BSD box, doesn't mean that BSD is any more (or less) secure. Most NT/Solaris/Irix/AIX/HP-UX/etc. exploits won't work on *BSD boxes either
Why does the good OS have to be Unix?
Don't get me wrong, I use Linux Mandrake on my home computer, FreeBSD on my firewall and I'm trying to get some Red Hat boxen installed at work.
but. . . (and it's a big butt) I believe the crucial element is the "goodness" of the OS, not the heritage or design philosophy. If an OS is good, use it. If it isn't, don't. The Palm OS is currently my favorite OS. It's not Unix but it does its job admirably. Let's not ignore any OS based on its "Unixness".
Free Mac Mini. Yes, I'm
Good idea. Windows 2000, right? What, you don't want that? That's what happens when you don't have competition. That's why we need Apple, and BeOS, and Sun, and SCO, and Linux and the BSD's.
That's why having Gnome and KDE exist as seperate projects is so important. They stimulate each other into creating a better and better tool. Think of Windows. It has a window manager into the OS. What innovation takes place? Only that which is needed to sucker innocent users into thinking it's better. But that's what we'll get if we don't have two Window managers being developed in parallel. The market *needs* competition, even if it is Open Source.
Or at least some system of making sure that no matter what platform I'm using, I can run ANY software available with no compatibility issues.A noble desire, but again one that isn't desirable. Sure, Java provides binary compatibility, and POSIX.1 provides source compatibility. These are fine for general type applications. But for optimized applications you need to take advantage of the hardware you're running on. Sun hardware, IBM hardware, SGI hardware. Yes, you can run the same apps on them, but you're not getting all the power you can out of them that way.
If I run a server on specific hardware and OS, I want it to be the best possible. If I run my Database on Sun's and Solaris, then I don't care how it runs on Alpha and Linux. However, easily being able to get it to run on different platforms if needed is a plus. For instance, say I'm a consultant and I develop a database for Sun and Solaris. If the next customer comes along and needs it on Alpha and Linux, then I need to be able to be responsive to my customer. Need I mention that if I'd developed my product on Windows, I'd have to completely rewrite it? Not so with the current Server OS's. That's the way I think it should be. Not fragmented, just optimisable.
-Brent--
Garfinkel discussed several issues in his article. On the niggling technical points he's correct, or at least not provably incorrect. Some things are simply a matter of taste.
However, he failed to mention at least two Big Things in his article, even though he was touching on the periphery of one of them.
Garfinkel characterizes the difference between the philosophies of *BSD and Linux as one of "correctness". The snottiness of this distinction aside, he's not quite accurate. The distinction is between old-fashioned release engineering as represented by *BSD, and a strange model of free contribution used by the GNU project and, more recently, Linux. While BSD's release engineering might make some system architects happier, it's not necessarily better, any more than tight government control of the economy is necessarily going to result in a stronger economy. The free-software approach may have flaws, but Garfinkel can't plausibly state which approach is more "correct".
The other, bigger problem he entirely avoids: The general suckiness of Un*x. I firmly believe that Un*x is the best thing going, but IT STILL SUCKS. There is no system-wide abstraction, and a lot less conformance to standards than most people think. NeXT was an interesting attempt to remove some of the suckiness from Un*x, and it may yet prevail under Apple's aegis, but I'll be more impressed when I see it win.
Given that there is no really good OS out there, dwelling on the differences between two fairly similar flavors of Un*x seems stupid and pointless. Let's work instead on making something that doesn't suck so much. That might derive from BSD (as Mac OS X does), or from Linux, or from any other damn thing -- it really doesn't matter that much. Just make sure it doesn't suck!
--
Some keywords for the NSA in the Lord of the Rings universe: One Ring bind find Sauron quest Nazgul freedom
What a pithy summary of the bazaar. It reeks of elitism and arrogance. It was very clever to attack ESR and his religion, did you think of that all by yourself? Thanks, but I prefer foul-mouthed Linux zealots.
The bazaar is important because it's surprising that it works at all. It suggests that people can usefully collaborate in a self organizing fashion, which is something *new* in the software world. Some of us think that this concept has importance in terms of promoting freedom and escaping corporate development nightmares. You don't need to agree, but please, don't come off like some aristocrat too good for the peasants. Your envy is showing.
It's really a shame both OSs have to put up with people who are more interested in being thought elite and correct than doing something to improve their OS. Both OSs deserve far better advocacy.
Why does it seem that every time an article about BSD or Linux or Microsoft or Be or anything shows up, the author takes up a militant stance against who ever he or she thinks is the the competition? It seems that most operating systems have good points to them, and although the need for propoganda is great for any operating system, open sourced or not, it becomes a slugfest, of us against them, and we must prevail.
If you like a system, fine, use it, and tell us all about what it can offer, but don't try to deride others in the process. In the end, it's all about choice and freedom, and that's what we all want, right?
Saw it written and I saw it say, pink moon is on its way. None of you will stand so tall, pink moon is gonna get ye al
There is a fairly recent effort to port Linux to HP's PA/RISC architecture.
The Power (not PowerPC) port is being worked on by Cort Dougan and supported by IBM. (Any URL ?)
The port to Hitachi's SuperH architecture (sh3, sh3e, sh4, as earlier models have no MMU) was done by Yutaka Niibe and is included in the 2.3 development kernel series.
There are rumours about the 64-bit MIPS and PPC ports being worked on.
Well,
I work for an ISP that has presence in 45 states(and a good bit of Europe) and I would say that 75 percent of our business are on *BSD boxes. Free and BSDI. With the leaning towards Free-BSD all around.
Of course we will rent you whatever box including Linux BUT our tech time is spent more on Linux and NT. The BSD boxes chug along quietly.
I run win98, Linux, and BSD at home. Linux has been a primer for Unix and me to enter the BSD world.
I think BSD is a corporate choice. But it is only chosen by corporations that have the in house skills to implement.
The first is the assumption that "the media equals reality". Linux exploits announced in the media and they don't affect NetBSD? Conclusion: NetBSD has no security weaknesses! Hah to you, Penguin! "Linux is the favorite tool of crackers [sic]" I suppose because criminals prefer the automobile we should all stick to mini-vans. Lots of companies announce Linux support and products? Obviously corporations are now driving Linux development! Boy, won't the developers themselves be surprised to hear that.
The second is the assumption that "correctness is superior to functionality". When it's correctness at the expense of functionality (as is arguably the case with certain of the *BSD lines, in certain cases), I take exception to being told "cope".
I do consider this article to be FUD. If you can't argue something's merits without falling back on attacking the perceived competition, you discredit yourself and the subject of your article.
Nothing worth doing is worth doing today.
> This has made the system popular at places like MIT and NASA,
> both of which have large menageries of computer systems from a multitude of
> vendors.
Umm, after spending 5 years at MIT, I can only recall seeing 1 *BSD box, although there certainly might be more....
The campus computers these days are almost entirely Sun's and SGI's (running their respective commercial unices). Over the last five years, the campus computers that have been phased out have been IBM RS6000's (AIX), Sparc Classics (Solaris), DEC 5000's (ultrix), and VAX Station 3100's (ultrix). None of them *BSD.
I worked in both CS and EE labs, the CS lab was a SUN shop, the EE lab used Linux for the main server, and a combo of Solaris/Linux/95 for everything else.
Residential computers are a mix of 95/98/NT/and Linux, with a couple macs thrown in for good measure. Most of the technically orientated people I knew, if they weren't running Linux, had a second box running Linux under there desk. (The place I lived at had ~1.3 computers/person, which was fairly common).
If anyone had a different experience, pipe in.
Stop the holy wars now - you can't argue which philosophy is better than another - only which implimentation is better.
--
"But I'm not rejoicing for the ascendancy of Linux or its penguin mascot."
In other words, if it's not BSD, he doesn't want it to succeed. We don't want users to have a choice. No, we just want them to have BSD.
" If I had to pick out the single difference between the BSD community as a whole and the proponents of Linux, I would say it is something called ''correctness.''"
Ok...if that is true, then that is a very important point. But is it true? How can we know? Not from this article. It doesn't cite any ways in which BSD is more "correct" than Linux. I guess we're just supposed to trust Mr. Garfinkel.
" This may be one reason proponents of Linux are frequently slow to admit the debt they owe to the Computer Science Research Group at Berkeley, which created BSD, and to the Free Software Foundation at MIT"
Oh, now I get it. I know Mr. Garfinkel's true identity. *wink* But why is he calling it Linux rather than GNU/Linux?
"Although Linux is a fine operating system, I would encourage businesses as well as advanced enthusiasts to take a serious look at the other choices."
Now, I can agree with this point. However, that "Linux is a fine operating system" is not the impression that I get from the rest of the article.
This could have been a really good article. It included some real meat, unlike most BSD articles in mainstream mags. However, it is ruined by the author's arrogance. An arrogance which unfortunately seems to be abundant among BSD users.
In some ways, I think the biggest (certainly in the mainstream world's eyes) difference between Linux and BSD are the types of users you encounter. Immature flamers or arrogant snobs. (Betcha can't guess which is which). Most of the people I know who run either OS are really cool people, who I'm glad to know. But those types of people aren't as loud as the stereotypes. Possibly the biggest advantage of Linux over BSD is that it's most stereotypical users (immature, foul-mouthed flamers) don't get published in the mainstream press.
...article. One that the article is complete FUD and the other is that the guy is doing some kind of service to the community.
/is/ tightly controlled by Linus Tovalds and it is a very *pragmatic* design. There are always tradeoffs to be made in the design of any architecure, and Linux is simply one approach. One that, IMHO, works very well.
I fall in the middle... I think an article espousing the virtues of the *BSD systems over Linux is good...there aren't many that I've seen in the mainstream press.
however, the article does contain a LOT of FUD.
The bit about "correctness" really struck me. While Linux is by no means designed to be the ultimately correct perfect academic example of operating system design (for instance, Linux is a monolithic design [the kernel is one big program], vs. the "correct" academic microkernel architecture [small, independent components with tightly-controlled communications linking them]), the design and architecture
The bit about Linux being very much a subject of attack for 'crackers': sure, Linux gets more attacks. But its probably because there are more Linux boxes than *BSD boxes. Windows is attacked far more than Linux (witness recent developments such as Melissa and ExploreZip). Why? Because there are more Windows boxes than Linux boxes. If BSD were more popular than Linux, I'm sure it would be the subject of more attacks.
But some of the things about OpenBSDs strong security vs. FreeBSDs excellent support for threads are very good points. But the guy definitely is spreading FUD: if you can't build up support for your operating system by pointing out its virtues, I guess you're left to attacking your competition. Of course, this tactic is simply tasteless, disgusting and simply downright childish.
My journal has hot
I've never used *BSD extensively, but I respect it (I'm much more accustomed to both Linux and Solaris). And I'm glad to see an article espousing the benefits of *BSD over Linux (gasp! shock! horror!)
This is because the whole goal of Linux is choice, which having a viable *BSD option provides. One more good Unix OS is always a good thing.
So, fellow Linux users/admins, please don't flame the dude. Respect him as a colleague.
"You can never have too many elephants on your team."
I just don't see much behind the advantages the author claims. Still, I'd usually rather have a FreeBSD 2.2.8 box than a RH6.0 . I simple cannot abide SysV /etc/rc.d .
As for `correct` vs `working`, I don't want to get into the old sync debate. It boils down to a choice you make. Linux certainly has more hardware supported, even if `less correctly'.
As for "less commercial", I beg to differ. The BSD licence is wide open for source being taken private. The GPL virus prevents this in Linux.
-- Robert
I have been an avid Linux user since 1990. I think my first ISP used FreeBSD, so I had a shell account on that server. That is the limit of my experience with any sort of BSD. However, as much as I am devoted to Linux, I am still open-minded about BSD. I just really have not had any reason to dual/triple/quad boot my machine.
But I digress. BSD has been around a long time. So of course it is going to be stabler and more robust than Linux in certain applications. I have no problem with that. It is nice that the BSD's are getting more press. More power to us all, I say. But I think the quality of that press is as important as the recognition that BSD (or Linux) gets.
My problem is with articles like these. Aside from a couple token points of recognition to Linux, he lavishes disputable credits to BSD.
For instance, BSD is ported to more architectures. The author sites 22. But I also know for a fact that if you go to the Linux Documentation Project you will see that Linux is ported or being ported to *at least* 22 other architectures, more if you count the various subsets.
His "argument" that BSD developers are more concerned with "correctness" than Linux developers is pure flamebait and unsupported. That is a slap in the face. If this were true, Linus would never have started coding a new kernel nearly a decade ago and it would not have gotten the support it did.
"Businesses are now the driving force in the Linux community." I think this statement is a combination of jealousy over Linux's success and a statement of ignorance of Linux's roots. Ever hear of Debian: one of the stablest distributions with arguably the most advanced package management tools around, owing *nothing* to business? Is he trying to tell me that if IBM and Oracle decided tomorrow to drop their support of Linux and switch to one of the BSD's that the BSD developers would say "no thanks"?
I think the author also tries to capitalize on the whole GNU/Linux debacle by stating that Linux users neglect what Linux owes to BSD and GNU and that Linus "cobbled" the OS together. I think this is a diversionary tactic. What do you think GNU is for? GNU was envisioned as a non-Unix unix. How could it be inappropriate that the Linux kernel work with GNU software? Or BSD software for that matter? I may not *say* GNU/Linux or GNU/Linux-with-BSD-utilities, but I know where Linux came from, and I support GNU and BSD. I don't appreciate the author labeling Linux users/developers as essentially code thieves and ingrateful hacks.
Finally, the author gives no attention to how old Linux is compared to BSD. Sure, he sites a book proclaiming BSD's 20 year vintage. But does he mention that Linux is barely 10 years old? He seems to neglect the astonishing rate at which Linux has developed, with or without the support of business. Let's wait another 10 years and see how mature and stable and secure Linux is then.
If he wants to caution businesses about implementing Linux, that is fine. Any business *should* weight all the alternatives. Linux does have a long way to go, but it has also come a long way in a short time. It can *only* get better.
I think the author himself should have weighed all the evidence before writing such a flagrantly biased piece.
Nothing can possiblai go wrong. Er...possibly go wrong.
Strange, that's the first thing that's ever gone wrong.
Tyler's words coming out of my mouth.
Interestingly, another major security expert, Bruce Schneier, in his Sept. 15 CRYPTOGRAM , praises Linux for its relative security over Solaris. (Schneier declines even to compare Linux to Windows.)
Schneier attributes Linux's enhanced security to its being open source.
So to say that Linux is insecure gives rise to the question, "Insecure as compared to what?".
Of the three BSD's, OpenBSD is the most secure. It is also Canadian, free from US export restrictions.
Since, as Schneier suggests, open source enhances security and since OpenBSD is the most secure, we might conclude that in some broad sense the term "open source" extends to freedom from export controls as well as freedom from various intellectual property restraints.
Not too bad an article. There's nothing wrong with advocacy, or saying you like something. A couple points though:
-- BSD is not a decendant of the GNU Project. It uses some GNU tools such as gcc, but largely it has an independent lineage.
-- Stability and security. This is the same thing Linux partisans talked about for a long time. The reply was always, "but what about applications and support". Now with the commericalization of Linux, it's the Linux camp bragging about apps and support. This is really where Linux has an edge today.
The contrast between BSD and Linux is the contrast between the Cathedral and the Bazaar. Clearly, Linux development is more anarchic, and one might expect that BSD would have some temporary advantages because of that. However, Richard Gabriel wrote an interesting essay, Worse is Better, explaining why C and Unix had overtaken Lisp, etc. The title is somewhat facetious but the observation is a fine one. The key point (missed, I think, by Gabriel), is not that the weaknesses of C vs. Lisp contributed to its success, but that the "get it working, then get it right" nature, and the openness of C and Unix let it evolve and reach perfection faster than striving for absolute perfection right off would have.
Both approaches have a good deal of merit, and one is not more right than the other. However, I would suspect that Linux will advance at a faster rate than BSD. Perhaps not always in useful directions, perhaps not always doing the One Right Thing, but over time, it will get there.
Currently Linux is actively developed for Alpha, ARM, IA64, x86 (IA32), PPC, MIPS, m68k, and sparc(64). There is also a sh3 port, but I'm not sure how active it is (9 architectures).
NetBSD currently runs on Alpha, m68k, ARM, PPC, ix86 (IA32), MIPS, ns32k, sh3, sparc(64), and vax (10 architectures).
Note these are chip architectures, the "kinds of computers" is much larger. Under NetBSD there are explicit ports to different computers running the same chip, for instance a macppc port and a ofppc port even though they both use PowerPC chips. Linux doesn't differentiate ports like this, so it would appear that Linux is ported to far less machines than NetBSD.
99.5% (give or take) of all exploits for the Linux OS are distribution binary exploits, not kernel exploits.
This means that if you had SSH installed on your box and a security announcement regarding SSH on Linux was put out, chances are you would be vulnerable as well. The real difference is that exploit code examples for Linux are far more common than for *BSD.
I would almost say though that a lot of the daemons *BSD uses are typically higher quality than what the Linux world uses, but nothing really stops someone from packaging say an OpenBSD FTP server with a Linux distribution (I believe Debian does now).
I've seen this argument a lot, but I have yet to see a benchmark performed on any modern kernel. The last benchmark I saw as for a 1.2.x kernel which was quite a while ago.
Linus Torvalds is one of the most anal retentive people on this planet (no offense Linus). You see him all the time rejecting patches because of poor architecture design. Of course, he only handles (for the most part) the intel port & generic linux system, but the other subsystem heads are just as bad.
Really, unless you are a kernel developer (ie, you've had patches accepted), you really can't begin to understand the pain and torture that one has to go through to get a patch accepted, especially when they implement new features
Really, when it all comes down to it, how different is Linux from *BSD? I mean, if you took a *BSD system and stuck a Linux kernel instead of a *BSD kernel and changed any type of incompatibilities... would you think it still inferior?
Of course, that's just my opinion, I could be wrong
--
The world is neither black nor white nor good nor evil, only many shades of CowboyNeal.