Don't think I haven't thought of HTML and HTTP that way, and for some purposes they provide some of the same solution, but... no.
Web browsers aren't the GUI equivalent of the UNIX shell. They're one potential GUI equivalent to a terminal program. And they're more like an IBM 3270 terminals than an interactive terminal you can run a shell in: they could potentially lead to something like IBM JCL, something that does part of the job, part of the way, but is haring down a completely wrong direction.
You just don't get it. You can keep bringing up examples like this all day long. They have nothing to do with Linus' own views
You're right, I don't get it. You made a positive claim about Linus' own views, based on the fact that he used the GPL.
I objected to that generalization.
I haven't said that the claim was wrong, because I don't know how strongly he believes in the GNU project. But I do object to the argument that using the GPL carries the kind of implication about ones beliefs that you keep asserting.
Let me repeat that.
I do not think I have made a single statement that says Linus agrees or disagrees with the GNU project, all I've said is that the evidence I have seen wasn't (and still isn't) strong enough to make any such claim. I presented a quote, you presented a quote. I said your quote was subject to interpretation. you said my quote is subject to interpretation.
Well, yeh, both of them are pretty wide open.
But, and this is important, that's a side issue. It's not what I objected to. And I'm not going to try and defend a claim I haven't made and don't believe not matter how much you want me to.
What I am arguing is merely that using the GPL doesn't necessarily imply agreement with the "founding principles of GNU". that is: one can use the GPL on ones software for reasons that have nothing to do with the reasons in the GNU manifesto or other documents.
I'm sorry that you took that badly, and made the broader claim that ones support of free software is somehow tied in to the GPL and the GNU project. I don't understand that point of view, and I don't appreciate the implication that the BSD license is somehow less free and inferior.
I think I've made *that* point good and hard, and if you don't understand it, I'm not going to get anywhere making it again. Either you accept that the GPL is not the only way to keep free software free, or you don't.
I'm quite happy to defend my interpretation of the GNU Manifesto and the GPL, based on my experience in watching the events unfold. I've provided a link to a document at the GNU site that provides a bit of that background.
But I'm not going to get caught up in an argument about what Linus' beliefs may be. I'm not taking up that gauntlet. Sorry.
Perl hardly refutes it, when in the previous question he gave a laundry list of tools and Perl wasn't on it... and Awk was.
I really think he was evading the answer.
The real answer is that you need a framwork that lets you connect the tools together easily before you can use a software tools approach. For the command line era, that framework was the UNIX shell. For the GUI era, there really hasn't been a popular framwork that's also portable. AREXX, Plan 9, Applescript, these seem to be the best frameworks I've seen so far, and they're all isolated to ghettoes... we're still waiting for the GUI equivalent of the UNIX shell.
Re:Plan 9, Unix may not have it, but another OS do
on
Rob Pike Responds
·
· Score: 2, Interesting
Windows is still built around the idea that each machine is self-sufficient. And, well, it has to be. You can build a distributed system on top of that, but if it's going to be widely adopted you need to be able to do it all in one machine: that is, Plan 9's problem is that it required a network. I once asked Dennis Ritchie if there was any real point to running one Plan 9 computer, and his response dissuaded me from trying it.
No, what you need is a standalone system that you can build up into a distributed system. The UNIX kernel isn't the ideal place to start (ideally, something that would let you move components across a machine or network boundary... a message-passing microkernel, perhaps... would be better), but it will do. The distributed authentication in Windows 2000 is now built on the distributed authentication from the Project Athena at MIT... which was developed on UNIX and VMS. Distributed file systems? There's a plethora of them. Distributed applications? The Plan 9 model works well on a UNIX system call style of interface...
We aren't even sure that the laws apply to everything that is observable.
Actually, we have reason to believe that fundamental physical constants may be different elsewhere in the observable universe, if some current theories of cosmogeny are correct. The fact that they don't appear to be significantly different based on what we can observe is one of the tests applied to these theories.
This includes the possibilities that forces (other than strong, weak, electomagnetic, and gravitational) that are too weak to measure here are strong enough to be detectable elsewhere.
What we call the "laws of physics" may only be a local phenomena
A better way of stating that is "the laws of physics include the possibility that our understanding of the laws of physics is incomplete, and there may be differences elsewhere in the universe or at other times (which comes to the same thing)". For example, there may have been a "strong light retarding force" that has now weakened to the point that it's not detectable, and part of the redshift is due to this force.
But mathematically, that force produces a universe that appears to match the one that would result if it didn't exist, but is uniformly expanding. So what's the point?
Any machine that can handle Win2k should run Linux/X11/KDE just great - in your case I'd definately look for possible hardware problems, misconfigurations, maybe compilation problems... look at the distro.
I built the whole system from the source trees, using only the components I needed. I tried debian to see if it was better, no difference.
I suspect that my idea of "just great" and yours are quite different. As soon as I brought up Mozilla along with the desktop, performance plummeted. With Windowmaker there was room for it.
KDE compares favourably here - instead of just changing the button look, for instance, you can also change the layout
But only for a small fraction of the applications I use, same with Gnome.
If you want a uniform GUI environment, you set up all KDE apps, all GNOME apps, all GNUStep apps (this would be my preference, but at the moment there aren't enough people making those
There aren't enough "all anything apps" under X. That is, there's few enough apps available for X as it is, restricting myself to a single toolkit would leave me unimaginally poorer, and not doing so leaves me with having to keep track of the quirks of every application. So using non-toolkit apps isn't just an occasional thing, it'd be a daily occurrance. XV, XMMS, XTerm,... there's no version of any of these in either of the "desktop" environments I've found acceptable.
And, after all, OS X doesn't require that I restrict myself to Aqua apps either. I started out making sure that I had a good X11 environment before I switched, so I could be sure I wasn't restricted in the tools I could use... and yet I find myself rarely firing up X11, and that usually for remote access or because X11 has less overhead on my computer rather than because I'm missing a tool or because the X11 tool is better than the Aqua one.
I suspect I use Classic more often.
The fact is, not only is OS X more consistent, but there's incomparably more applications available for it. So it's rare (and increasingly so) that I have to look to X11 applications to fill in the gaps.
Intrinsic redshifts explain the anomalous association of high-redshift QSOs with galaxies of much lower redshifts.
I think Arp's material is interesting, but I woudl describe them as "the apparent association of high redshift quasars with galaxies of much lower redshifts".
It always amazes me that otherwise sound scientists start talking about 'belief' whenever certain concepts are mentioned
It would amaze me if it were a regular occurrence, certainly, and not a reflection of the arguments otherwise apparently sound people repeatedly bring up. Physicists and engineers get defensive about the word "nuclear" for the same reason... and so "Nuclear Magnetic Resonance Imaging" becomes "Magnetic Resonance Imaging"... but you don't ask what they're hiding or why they're in denial.
Or maybe you do, I don't know. Most people don't.
The real bugger is that when you let intrinsic redshifts out of the bag (ie. that you could be observing them as a phenomenon, not on the basis of a theoretical model
Then nothing changes. Gravitational redshifts and high intrinsic velocities for individual galaxies are expected and have both been observed. If the quasar's redshift is due to it being a massive object, or an object with a high real velocity, or a cauldron of strange physics, its apparent association with a "normal galaxy" that doesn't share these characteristics is no more or less anomolous than before.
there's nothing preventing expansion from being a purely local effect, if you define "local" as 50 billion light-years across
But, Tom, all cosmogeny can tell us about is the history of the universe within our light cone. Beyond that, well, we may be a singularity... a little bubble of emptiness within a universe filled with magic and mystery and different physical laws, a bubble of matter in an aweful void, or anything in between. It doesn't matter. The history of the observable universe is the same, whether it's representative of a larger space, all there is, or an exceptional case created by whatever means for whatever reason in an incomparably different world... a brief eternity in the equivalent of the spark from a cat's back.
What's outside the observable universe is not, so near as we can tell, observable. The parsimonious assumption is that since it isn't observable... that all we can do is speculate about what it might be with no information at all to prove those speculations one way or another... it's not part of any theory that we can devise.
The instruments in the Hubble are likely to be damaged by the brightness of the Earth.
But don't worry, the Keyhole scopes the US intelligence community use are basically Hubbles pointing at the earth, with appropriate instruments. Of course they don't let anyone see the pictures or admit they exist, but that's a minor detail.
So, do you believe in intrinsic redshifts based on gravity or on something that's mathematically identical to an expanding universe? In the former case, why are there no other gravitational effects? In the latter case, William of Occam wants to talk to you.
There's been a few Physorg stories show up on/. recently, and they've been abstracts of stuff that's old news. I wanna moderate the site down (redundant?).
Google for PCIExtreme. It lets you turn it on for PCI or off for AGP.
I can't imagine running KDE on a P2 with 64M. I have a Libretto, 64M (that's the max it supports) and a 266 MHz CPU. I can run Windowmaker and it's OK, but with any of the "desktops" it's swapping its little heart out. It runs Windows 2000 just fine.
Yes, I like the customizability in X window managers, but the problem is that the customizations stop at the window manager level. Every application below that has its own settings... if you have multiple apps using the same toolkit you can often get them all tweaked together, but otherwise you're stuck. On Mac OS X, change a setting and it changes globally. I've switched from the Aqua look and feel to a theme called "Milk", disabled the metal look for all Cocoa apps (Carbon apps like iTunes are still metal, alas), and everything follows it, all the way down to the buttons in Firefox or Safari web pages.
Show me any alternative that will [...] run on so little hardware?
Well that brings us back around to my original point about the "Mac Tax". Glad to see you agree.:)
The Running Man (A terrifc movie for those who haven't seen it)
Someone made a movie about AIX system administration?
Re:But, how do you really feel?
on
Linus Interviewed
·
· Score: 0, Redundant
but to call them greedy and anti-American Dream is taking things way too far
He called them greedy, but that's hardly something you should find offensive: the laws that govern the behaviour of public companies significantly penalise anything but greedy behaviour. That's a fact of life, and companies that don't act that way are never going to become industry leaders. Open Source actually has an advantage there: if Linus doesn't take advantage of some opportunity to make money he's not going to be hauled into court for neglecting his fiduciary duty to his stockholders.
As for the second part, he didn't say Microsoft was "anti-American Dream". He said that it wasn't the "American Dream". I'm not Linus Torvalds... does that mean I'm anti-Linus Torvalds?
Take Econ101.
You probably ought to review that yourself. Maybe you'd understand what he was getting at when he said they "might be better off with some competition". What's more essential to the American economy than competition, after all?
a pre-existing OS that is still not ready for widespread adoption on desktop systems
Funny, I'm running that pre-existing OS on my desktop, and it's got one of the best user interfaces and the most consistent and easiest to learn applications environment in the world.
It's not Linus' particular implementation of that operating system, no. I personally prefer the one that McKusick and co brought into the spotlight: I did some of the work on that so I'm much more familiar with it and I'm glad Apple picked it rather than Linux. But to say that what Linus did wasn't amazing is churlish and childish.
Heck, even when I think Linus is being a butthead about kernel design, and I'm telling him so, he's still one of my heroes.
Well, at least your Windows 2003 server is at LEAST 17% cheaper to exploit.
Re:Fixing fundamental design mistakes?
on
Linus Interviewed
·
· Score: 1
Thus when you are trying to do things like "sort this directory by type of file", it is much more efficient to do it using file extensions than magic numbers
"UNIX applications don't depend as much on the file extension as Windows applications" doesn't imply "You can't use a standard file extension for your files on UNIX".
Yes, fundamental design mistakes.
on
Linus Interviewed
·
· Score: 2, Insightful
Some folks still think that *nix is inherently virus proof because anything a mere user runs couldn't touch the really important stuff in/bin.
No, UNIX is inherently virus-resistent because it was developed in a multiuser environment where you did things like having professors keeping exam results on the same computers that students had accounts on. You had to, it was too expensive to have separate computers for every group that might have a reason to compromise another's security. At Berkeley, it was fairly common for people to set up trojan horses in their home directory to try and trap people who visited their accounts.
The result of this is that the default behaviour of the standard applications (like word processors, browsers, the file managers) when faced with an unknown object isn't to execute it and see what happens. The few exceptions are well publicised and used as object lessons as to why you shouldn't do things that way.
Windows, on the other hand, has this model of "security zones", and once an object (file, document, web page) gets to a place where it's seen as local... in the "trusted zone"... standard Microsoft applications like IE and Outlook happily let it do anything its little heart desires. In response, Microsoft has spent the past seven years (at least) repeatedly redrawing the boundaries of this "trusted zone" in an attempt to keep bad guys from sneaking something into it.
That's one of the fundamental design mistakes that Linus is talking about. There are others, like the lack of a formal system call mechanism, or the fact that implementing strong local security is so inconvenient that it's normal and accepted for most home users to work with the equivalent of local root privilege all the time (and, yes, that really does matter in UNIX... it's some of the other design flaws in Windows that make this one relatively unimportant).
There are pathways for viruses and other malware to attack UNIX systems, but these pathways are due to bugs: fixing a buffer overflow hole in an image viewer won't change the way legitimate applications work or force the user to change the way they use the computer. Redefining the "trusted zone" in Windows does, because too many applications (including standard utilities and aministration tools shipped with the OS) depend on them... so every "fix" has a ripple effect that requires applications to use different APIs or devise workarounds, and often those workarounds end up being useful to malware authors as well.
And on top of that:
Could there be a future Linux kernel that prevents an image library exploit from modifying your.bashrc?
If you want to run your browser in a sandbox that prevents it from modifying your.bashrc, even if it's completely compromised by a buffer overflow, you can do it now. It's been possible since at least 1978 (the date on my copy of the 6th edition manual, but IIRC 6th edition dates back to 1976): if you run the browser in a chrooted environment, then the only way it can modify your.bashrc is for the guy writing the exploit to also come up with a way to break out ofthe chrooted sandbox.
So... given the security available on a 6th edition UNIX system in 1978, there are tricks that can be pulled to do that, but he'd need to be root first, so he'd have to have a third exploit to become root... from a sandbox where there's no root-owned and root-setuid executables to use to boost his privileges. It's possible there's a kernel attack that can be used, but it's much harder to devise a kernel attack on UNIX because verything has to pass through the same system call interface... you don't have separate call gates for each privileged component... so there's one interface the OS designers can depend on the attacker having to go through.
It could be done, but even with the facilities available in UNIX almost 30 years ago it was much much harder than it is now.
Saying Linus is from Sweden just because he happens to speak Swedish is like like saying all english speaking Americans are actually British.
Every English-speaking American I've ever met turned out to be from England, Australia/New Zealand, or South Africa.
Americans can fool you into thinking they're speaking English, but no matter how long you live here and how well you learn their curious language you inevitably end up running into amazing new constructions they drop like weapons of mass distraction into otherwise syntactically correct English speech.
A lot of people from that part of the world seem to be.
I'd love to hear Garrison Keillor interview Linus.
So long as they didn't get into some kind of understated irony competition, of course. I don't think space-time could take it, you'd end up in some kind of conversational singularity.
Slashdot Mirror
Don't think I haven't thought of HTML and HTTP that way, and for some purposes they provide some of the same solution, but... no.
Web browsers aren't the GUI equivalent of the UNIX shell. They're one potential GUI equivalent to a terminal program. And they're more like an IBM 3270 terminals than an interactive terminal you can run a shell in: they could potentially lead to something like IBM JCL, something that does part of the job, part of the way, but is haring down a completely wrong direction.
Gravitational redshifts? You lost me. :(
Clearly.
You just don't get it. You can keep bringing up examples like this all day long. They have nothing to do with Linus' own views
You're right, I don't get it. You made a positive claim about Linus' own views, based on the fact that he used the GPL.
I objected to that generalization.
I haven't said that the claim was wrong, because I don't know how strongly he believes in the GNU project. But I do object to the argument that using the GPL carries the kind of implication about ones beliefs that you keep asserting.
Let me repeat that.
I do not think I have made a single statement that says Linus agrees or disagrees with the GNU project, all I've said is that the evidence I have seen wasn't (and still isn't) strong enough to make any such claim. I presented a quote, you presented a quote. I said your quote was subject to interpretation. you said my quote is subject to interpretation.
Well, yeh, both of them are pretty wide open.
But, and this is important, that's a side issue. It's not what I objected to. And I'm not going to try and defend a claim I haven't made and don't believe not matter how much you want me to.
What I am arguing is merely that using the GPL doesn't necessarily imply agreement with the "founding principles of GNU". that is: one can use the GPL on ones software for reasons that have nothing to do with the reasons in the GNU manifesto or other documents.
I'm sorry that you took that badly, and made the broader claim that ones support of free software is somehow tied in to the GPL and the GNU project. I don't understand that point of view, and I don't appreciate the implication that the BSD license is somehow less free and inferior.
I think I've made *that* point good and hard, and if you don't understand it, I'm not going to get anywhere making it again. Either you accept that the GPL is not the only way to keep free software free, or you don't.
I'm quite happy to defend my interpretation of the GNU Manifesto and the GPL, based on my experience in watching the events unfold. I've provided a link to a document at the GNU site that provides a bit of that background.
But I'm not going to get caught up in an argument about what Linus' beliefs may be. I'm not taking up that gauntlet. Sorry.
Perl hardly refutes it, when in the previous question he gave a laundry list of tools and Perl wasn't on it... and Awk was.
I really think he was evading the answer.
The real answer is that you need a framwork that lets you connect the tools together easily before you can use a software tools approach. For the command line era, that framework was the UNIX shell. For the GUI era, there really hasn't been a popular framwork that's also portable. AREXX, Plan 9, Applescript, these seem to be the best frameworks I've seen so far, and they're all isolated to ghettoes... we're still waiting for the GUI equivalent of the UNIX shell.
Windows is still built around the idea that each machine is self-sufficient. And, well, it has to be. You can build a distributed system on top of that, but if it's going to be widely adopted you need to be able to do it all in one machine: that is, Plan 9's problem is that it required a network. I once asked Dennis Ritchie if there was any real point to running one Plan 9 computer, and his response dissuaded me from trying it.
No, what you need is a standalone system that you can build up into a distributed system. The UNIX kernel isn't the ideal place to start (ideally, something that would let you move components across a machine or network boundary... a message-passing microkernel, perhaps... would be better), but it will do. The distributed authentication in Windows 2000 is now built on the distributed authentication from the Project Athena at MIT... which was developed on UNIX and VMS. Distributed file systems? There's a plethora of them. Distributed applications? The Plan 9 model works well on a UNIX system call style of interface...
Plan9 would be some anachronistic romanticism like Pfizner or Elgar.
Steady on, old man, steady on. You're getting awfully carried away...
---
I like to use lines like that. "There hasn't been any good music since Joplin died... no, I mean Scott Joplin..."
Windows XP is like the Monkees. It's not just commercial pap, it's old commercial pap.
Plan 9? Plan 9 is Jazz.
We aren't even sure that the laws apply to everything that is observable.
Actually, we have reason to believe that fundamental physical constants may be different elsewhere in the observable universe, if some current theories of cosmogeny are correct. The fact that they don't appear to be significantly different based on what we can observe is one of the tests applied to these theories.
This includes the possibilities that forces (other than strong, weak, electomagnetic, and gravitational) that are too weak to measure here are strong enough to be detectable elsewhere.
What we call the "laws of physics" may only be a local phenomena
A better way of stating that is "the laws of physics include the possibility that our understanding of the laws of physics is incomplete, and there may be differences elsewhere in the universe or at other times (which comes to the same thing)". For example, there may have been a "strong light retarding force" that has now weakened to the point that it's not detectable, and part of the redshift is due to this force.
But mathematically, that force produces a universe that appears to match the one that would result if it didn't exist, but is uniformly expanding. So what's the point?
Any machine that can handle Win2k should run Linux/X11/KDE just great - in your case I'd definately look for possible hardware problems, misconfigurations, maybe compilation problems... look at the distro.
... there's no version of any of these in either of the "desktop" environments I've found acceptable.
I built the whole system from the source trees, using only the components I needed. I tried debian to see if it was better, no difference.
I suspect that my idea of "just great" and yours are quite different. As soon as I brought up Mozilla along with the desktop, performance plummeted. With Windowmaker there was room for it.
KDE compares favourably here - instead of just changing the button look, for instance, you can also change the layout
But only for a small fraction of the applications I use, same with Gnome.
If you want a uniform GUI environment, you set up all KDE apps, all GNOME apps, all GNUStep apps (this would be my preference, but at the moment there aren't enough people making those
There aren't enough "all anything apps" under X. That is, there's few enough apps available for X as it is, restricting myself to a single toolkit would leave me unimaginally poorer, and not doing so leaves me with having to keep track of the quirks of every application. So using non-toolkit apps isn't just an occasional thing, it'd be a daily occurrance. XV, XMMS, XTerm,
And, after all, OS X doesn't require that I restrict myself to Aqua apps either. I started out making sure that I had a good X11 environment before I switched, so I could be sure I wasn't restricted in the tools I could use... and yet I find myself rarely firing up X11, and that usually for remote access or because X11 has less overhead on my computer rather than because I'm missing a tool or because the X11 tool is better than the Aqua one.
I suspect I use Classic more often.
The fact is, not only is OS X more consistent, but there's incomparably more applications available for it. So it's rare (and increasingly so) that I have to look to X11 applications to fill in the gaps.
Intrinsic redshifts explain the anomalous association of high-redshift QSOs with galaxies of much lower redshifts.
I think Arp's material is interesting, but I woudl describe them as "the apparent association of high redshift quasars with galaxies of much lower redshifts".
It always amazes me that otherwise sound scientists start talking about 'belief' whenever certain concepts are mentioned
It would amaze me if it were a regular occurrence, certainly, and not a reflection of the arguments otherwise apparently sound people repeatedly bring up. Physicists and engineers get defensive about the word "nuclear" for the same reason... and so "Nuclear Magnetic Resonance Imaging" becomes "Magnetic Resonance Imaging"... but you don't ask what they're hiding or why they're in denial.
Or maybe you do, I don't know. Most people don't.
The real bugger is that when you let intrinsic redshifts out of the bag (ie. that you could be observing them as a phenomenon, not on the basis of a theoretical model
Then nothing changes. Gravitational redshifts and high intrinsic velocities for individual galaxies are expected and have both been observed. If the quasar's redshift is due to it being a massive object, or an object with a high real velocity, or a cauldron of strange physics, its apparent association with a "normal galaxy" that doesn't share these characteristics is no more or less anomolous than before.
there's nothing preventing expansion from being a purely local effect, if you define "local" as 50 billion light-years across
But, Tom, all cosmogeny can tell us about is the history of the universe within our light cone. Beyond that, well, we may be a singularity... a little bubble of emptiness within a universe filled with magic and mystery and different physical laws, a bubble of matter in an aweful void, or anything in between. It doesn't matter. The history of the observable universe is the same, whether it's representative of a larger space, all there is, or an exceptional case created by whatever means for whatever reason in an incomparably different world... a brief eternity in the equivalent of the spark from a cat's back.
What's outside the observable universe is not, so near as we can tell, observable. The parsimonious assumption is that since it isn't observable... that all we can do is speculate about what it might be with no information at all to prove those speculations one way or another... it's not part of any theory that we can devise.
there are a lot of good causes that all these dollars could be going to right here on Earth
Sure. But would they be?
The instruments in the Hubble are likely to be damaged by the brightness of the Earth.
But don't worry, the Keyhole scopes the US intelligence community use are basically Hubbles pointing at the earth, with appropriate instruments. Of course they don't let anyone see the pictures or admit they exist, but that's a minor detail.
So, do you believe in intrinsic redshifts based on gravity or on something that's mathematically identical to an expanding universe? In the former case, why are there no other gravitational effects? In the latter case, William of Occam wants to talk to you.
There's been a few Physorg stories show up on /. recently, and they've been abstracts of stuff that's old news. I wanna moderate the site down (redundant?).
That's a bloody annoying site. Abstractes that provide little more detail than the Slashdot precis, and no links to sources.
Turn off quartz extreme?
:)
Google for PCIExtreme. It lets you turn it on for PCI or off for AGP.
I can't imagine running KDE on a P2 with 64M. I have a Libretto, 64M (that's the max it supports) and a 266 MHz CPU. I can run Windowmaker and it's OK, but with any of the "desktops" it's swapping its little heart out. It runs Windows 2000 just fine.
Yes, I like the customizability in X window managers, but the problem is that the customizations stop at the window manager level. Every application below that has its own settings... if you have multiple apps using the same toolkit you can often get them all tweaked together, but otherwise you're stuck. On Mac OS X, change a setting and it changes globally. I've switched from the Aqua look and feel to a theme called "Milk", disabled the metal look for all Cocoa apps (Carbon apps like iTunes are still metal, alas), and everything follows it, all the way down to the buttons in Firefox or Safari web pages.
Show me any alternative that will [...] run on so little hardware?
Well that brings us back around to my original point about the "Mac Tax". Glad to see you agree.
The Running Man (A terrifc movie for those who haven't seen it)
Someone made a movie about AIX system administration?
but to call them greedy and anti-American Dream is taking things way too far
He called them greedy, but that's hardly something you should find offensive: the laws that govern the behaviour of public companies significantly penalise anything but greedy behaviour. That's a fact of life, and companies that don't act that way are never going to become industry leaders. Open Source actually has an advantage there: if Linus doesn't take advantage of some opportunity to make money he's not going to be hauled into court for neglecting his fiduciary duty to his stockholders.
As for the second part, he didn't say Microsoft was "anti-American Dream". He said that it wasn't the "American Dream". I'm not Linus Torvalds... does that mean I'm anti-Linus Torvalds?
Take Econ101.
You probably ought to review that yourself. Maybe you'd understand what he was getting at when he said they "might be better off with some competition". What's more essential to the American economy than competition, after all?
That is a weakness of the open source approach to building an OS, but ultimately it will get there.
:)
It is there. The best desktop you can buy today is running on top of an open source OS.
It's not an open source desktop, no, but as you say the GUI layer isn't the OS.
a pre-existing OS that is still not ready for widespread adoption on desktop systems
Funny, I'm running that pre-existing OS on my desktop, and it's got one of the best user interfaces and the most consistent and easiest to learn applications environment in the world.
It's not Linus' particular implementation of that operating system, no. I personally prefer the one that McKusick and co brought into the spotlight: I did some of the work on that so I'm much more familiar with it and I'm glad Apple picked it rather than Linux. But to say that what Linus did wasn't amazing is churlish and childish.
Heck, even when I think Linus is being a butthead about kernel design, and I'm telling him so, he's still one of my heroes.
Well, at least your Windows 2003 server is at LEAST 17% cheaper to exploit.
Thus when you are trying to do things like "sort this directory by type of file", it is much more efficient to do it using file extensions than magic numbers
"UNIX applications don't depend as much on the file extension as Windows applications" doesn't imply "You can't use a standard file extension for your files on UNIX".
Some folks still think that *nix is inherently virus proof because anything a mere user runs couldn't touch the really important stuff in /bin.
.bashrc?
.bashrc, even if it's completely compromised by a buffer overflow, you can do it now. It's been possible since at least 1978 (the date on my copy of the 6th edition manual, but IIRC 6th edition dates back to 1976): if you run the browser in a chrooted environment, then the only way it can modify your .bashrc is for the guy writing the exploit to also come up with a way to break out ofthe chrooted sandbox.
No, UNIX is inherently virus-resistent because it was developed in a multiuser environment where you did things like having professors keeping exam results on the same computers that students had accounts on. You had to, it was too expensive to have separate computers for every group that might have a reason to compromise another's security. At Berkeley, it was fairly common for people to set up trojan horses in their home directory to try and trap people who visited their accounts.
The result of this is that the default behaviour of the standard applications (like word processors, browsers, the file managers) when faced with an unknown object isn't to execute it and see what happens. The few exceptions are well publicised and used as object lessons as to why you shouldn't do things that way.
Windows, on the other hand, has this model of "security zones", and once an object (file, document, web page) gets to a place where it's seen as local... in the "trusted zone"... standard Microsoft applications like IE and Outlook happily let it do anything its little heart desires. In response, Microsoft has spent the past seven years (at least) repeatedly redrawing the boundaries of this "trusted zone" in an attempt to keep bad guys from sneaking something into it.
That's one of the fundamental design mistakes that Linus is talking about. There are others, like the lack of a formal system call mechanism, or the fact that implementing strong local security is so inconvenient that it's normal and accepted for most home users to work with the equivalent of local root privilege all the time (and, yes, that really does matter in UNIX... it's some of the other design flaws in Windows that make this one relatively unimportant).
There are pathways for viruses and other malware to attack UNIX systems, but these pathways are due to bugs: fixing a buffer overflow hole in an image viewer won't change the way legitimate applications work or force the user to change the way they use the computer. Redefining the "trusted zone" in Windows does, because too many applications (including standard utilities and aministration tools shipped with the OS) depend on them... so every "fix" has a ripple effect that requires applications to use different APIs or devise workarounds, and often those workarounds end up being useful to malware authors as well.
And on top of that:
Could there be a future Linux kernel that prevents an image library exploit from modifying your
If you want to run your browser in a sandbox that prevents it from modifying your
So... given the security available on a 6th edition UNIX system in 1978, there are tricks that can be pulled to do that, but he'd need to be root first, so he'd have to have a third exploit to become root... from a sandbox where there's no root-owned and root-setuid executables to use to boost his privileges. It's possible there's a kernel attack that can be used, but it's much harder to devise a kernel attack on UNIX because verything has to pass through the same system call interface... you don't have separate call gates for each privileged component... so there's one interface the OS designers can depend on the attacker having to go through.
It could be done, but even with the facilities available in UNIX almost 30 years ago it was much much harder than it is now.
Today,
Saying Linus is from Sweden just because he happens to speak Swedish is like like saying all english speaking Americans are actually British.
Every English-speaking American I've ever met turned out to be from England, Australia/New Zealand, or South Africa.
Americans can fool you into thinking they're speaking English, but no matter how long you live here and how well you learn their curious language you inevitably end up running into amazing new constructions they drop like weapons of mass distraction into otherwise syntactically correct English speech.
Linus would be very at home in Minnesota
A lot of people from that part of the world seem to be.
I'd love to hear Garrison Keillor interview Linus.
So long as they didn't get into some kind of understated irony competition, of course. I don't think space-time could take it, you'd end up in some kind of conversational singularity.