I certainly think binary sandboxing would be a good idea, though implementing it has a lot of various tricky issues I'd imagine. It wouldn't solve all problems though; for example, the recent OpenSSH root exploit would've been prevented if it had been written in Cyclone, but would not have been prevented by binary sandboxing, since OpenSSH has to run as root (or some other priviliged user) to be useful.
"Selling commercial software" isn't a field of endeavor; it's one method of undertaking an endeavor in any field. The GPL is certainly not designed to protect all methods of endeavors equally. In fact, I would not use it if it were; if I'm going to release my code to you, I'm going to attach conditions to it. With normal commercial software, I get money in return. If I'm going to give you my work for free, I'm going to want something else in return; in the case of the GPL, it's any modifications you make to it. If you want to do whatever you want with no strings attached, write your own damn code.
Really important stuff, like say SSHd, should be written in something safe. Just compiling in bounds-checking in an ad hoc manner is both slower and less safe than writing it safely to begin with.
Though as the other poster mentioned, if people just abandoned C in the first place, we'd solve a lot of the problems. Cyclone is nice in that it's a way for people who still want C's low-level control to abandon C's security holes without using a high-level language like SML.
It seems like Cyclone is designed explicitly for this -- somewhere where safety guarantees are worth some slight (but not major) performance penalties. It's a low-level language designed to be very compatible with C, but adds a bunch of safety features to the language (with a mind towards optimization; for example, you can declare a pointer "never-NULL" to avoid run-time NULL-pointer checking). And it gets rid of pretty much all buffer-overflow or pointer-dereferencing style errors, rather than just catching some of them as these ad hoc approaches do.
There is no field of endeavor or group of people prohibited by the GPL. Any group of people can make derivatives based on GPL'd software provided they accept the license agreements, and anyone in any field of endeavor may choose to do so. The only thing it discriminates against is people who choose not to accept the license agreement, which is a necessary requirement for all licenses (for example, the BSD license discriminates against people who wish to strip copyright information from the files).
Free software or documentation will probably be created to replace them.
Given the generally horrible state of Free Software documentation, I find this unlikely. This lack of free documentation was, in fact, the very problem the GNU FDL was intended to solve.
The point is that almost nobody writes documentation for free. GNU is, in fact, the only major provider of Free Software that consistently provides excellent documentation along with their software. The idea of invariant sections is that you can write documentation, and then provide something about who you are. For example, if you're a company, and you write a book documenting some piece of software, you could release it as a traditional copyrighted book. The purpose of the GNU FDL is to allow you to release it for Free, with an invariant section saying "I wrote this, and its writing was funded by So-and-So Publishers; if you'd like to support our work, buy that version."
If this is disallowed, it will not encourage "more-Free" documentation. It will instead encourage people to take their work and publish it as traditional copyrighted documentation.
There's a lot of gee-whiz techno-nonsense reminiscent of early 1980s cyberpunk. Just because you have a robot, a "hacker," and a scantily-clad schoolgirl in your show doesn't mean you can skip the plot.
If you read the papers at that site, the empirical conclusion is that the gain of caching large files is outweighed by the cache overhead. Always caching small files has two advantages: 1) you can now access any file (on disk or in memory) with a simple one-level pointer indirection, since you always know small files are in memory and large files are on disk, and have something equivalent to the standard inode table holding both; and 2) you now know every file on disk is "big" (for some value of big) and can optimize accordingly, in the process greatly simplifying your filesystem code (currently filled with lots of hacks for optimizing small-file accesses); an obvious win here is using large cluster sizes (say, 512K instead of 1K).
So what we have here is Debian claiming that the FSF is not radical enough in its promotion of Free Software? That seems a bit ridiculous; even the FSF realizes that allowing the author of documentation the right to include some invariant section (such as an attribution) is necessary to get people to write free documentation at all. I never thought I'd see an argument where RMS would be on the side opposing the zealots.
Perhaps I'm doing something wrong, but the AA in the xft build of Mozilla makes fonts unreadable on black backgrounds. It seems it works by basically blurring the fonts and letting the black bleed into the white a bit, which works beautifully when it's black text on a white background (the black of the text bleeds a little into the white of the background, smoothing out pixellation), but is completely unreadable when it's white text on a black background (the black of the background bleeds into the white that was supposed to be the text). It's not just Mozilla either; the gtk2 builds of gaim have the same problem.
1. There's free access available in many many areas at public libraries and schools, and much of it is not taken advantage of.
2. The same people who "can't afford" a computer can somehow afford all sorts of other equally or more expensive things. Cable TV ($40/mo * 12 mo = $480/yr) is one that comes to mind. Expensive car modifications (stereo systems, etc.) are another.
Most people have free access to the internet through their local public library. The exceptions are some people in very rural areas, but they make up quite a small percentage of the total US population (certainly far less than the 40% non-internet users quoted here). It seems a lot of people simply aren't interested in using the internet, despite free access.
And even if it weren't for that, count the number of people who "can't afford" a computer, but someone "can afford" to pay over $400/yr for cable television.
This is a very insular USENet thing adopted by some segments of Free Software culture, and not at all in keeping with past or present common usage in the computer field or wider culture. As noted in one of the other replies, the usage of "cracker" to describe people who break into computers was coined ca. 1985; the usage of "hacker" to describe these same people dates back to the late 1970s, and was already in very common usage by the early 1980s. For the vast majority of the history of computers, this (someone who breaks into computers) has been the primary meaning of the term "hacker."
I'm not sure hackers are the biggest problem here
on
Should You Hire a Hacker?
·
· Score: 2, Insightful
You ever listened to any gangsta rap or seen the movie Catch Me If You Can? Both probably have a much bigger influence on the general public.
Disaster relief. It's been proven before that the internet can be resilient to disasters such as earthquake. Useful maybe?
Well, considering that power is often the first thing that goes out in any disaster (be in earthquake, hurricane, etc.), that might limit the usefulness.
Given that the most popular Windows audio player -- Winamp -- comes with Ogg support built into the default install, how does this differentiate Linux?
If it was, say, a 1.5 GHz overclocked to 2.0 GHz, and sold as "running at 2.0 GHz," then this was a true statement. As for whether it was misleading, it depends on the chip and the overclocking. It's quite conceivable that some chips could be overclocked by 10-20% and still be more reliable than other chips clocked at the manufacturer's spec'd speed. Should it then be illegal to sell overclocked versions of these chips, even though they're still more reliable than other chips legally on the market?
I seem to recall reading an article a while back that Epitaph is actually an RIAA member. I'm not sure why they're not on the list on riaa.org. FWIW, both Epitaph and Fat Wreck Chords are on the RIAA membership list at boycott-riaa.com. Anybody have more details? Maybe they're owned by a parent company on the riaa.org list?
Searle's main thrust seems to be that the Chinese Room is an example of simple lookup table responses, which he claims is not understanding. If there was a full Turing machine running the responses (rather than a simple lookup table), it becomes much less clear that he can say there's no understanding going on.
I believe if one accepts Chomsky's theories of language, that it is actually impossible to encode all possible language behavior as an input-output system. I'm not an expert on the subject so I can't say so for sure though. The thrust, though, would be that language has to be encoded in a manner which is more complex than a stimulus-response system because no stimulus-response system can fully implement language; thus, Searle's example of a Chinese room built on a stimulus-response system would be impossible to construct, making his thought experiment incoherent.
Slashdot Mirror
I certainly think binary sandboxing would be a good idea, though implementing it has a lot of various tricky issues I'd imagine. It wouldn't solve all problems though; for example, the recent OpenSSH root exploit would've been prevented if it had been written in Cyclone, but would not have been prevented by binary sandboxing, since OpenSSH has to run as root (or some other priviliged user) to be useful.
"Selling commercial software" isn't a field of endeavor; it's one method of undertaking an endeavor in any field. The GPL is certainly not designed to protect all methods of endeavors equally. In fact, I would not use it if it were; if I'm going to release my code to you, I'm going to attach conditions to it. With normal commercial software, I get money in return. If I'm going to give you my work for free, I'm going to want something else in return; in the case of the GPL, it's any modifications you make to it. If you want to do whatever you want with no strings attached, write your own damn code.
Really important stuff, like say SSHd, should be written in something safe. Just compiling in bounds-checking in an ad hoc manner is both slower and less safe than writing it safely to begin with.
Though as the other poster mentioned, if people just abandoned C in the first place, we'd solve a lot of the problems. Cyclone is nice in that it's a way for people who still want C's low-level control to abandon C's security holes without using a high-level language like SML.
It seems like Cyclone is designed explicitly for this -- somewhere where safety guarantees are worth some slight (but not major) performance penalties. It's a low-level language designed to be very compatible with C, but adds a bunch of safety features to the language (with a mind towards optimization; for example, you can declare a pointer "never-NULL" to avoid run-time NULL-pointer checking). And it gets rid of pretty much all buffer-overflow or pointer-dereferencing style errors, rather than just catching some of them as these ad hoc approaches do.
There is no field of endeavor or group of people prohibited by the GPL. Any group of people can make derivatives based on GPL'd software provided they accept the license agreements, and anyone in any field of endeavor may choose to do so. The only thing it discriminates against is people who choose not to accept the license agreement, which is a necessary requirement for all licenses (for example, the BSD license discriminates against people who wish to strip copyright information from the files).
Free software or documentation will probably be created to replace them.
Given the generally horrible state of Free Software documentation, I find this unlikely. This lack of free documentation was, in fact, the very problem the GNU FDL was intended to solve.
The point is that almost nobody writes documentation for free. GNU is, in fact, the only major provider of Free Software that consistently provides excellent documentation along with their software. The idea of invariant sections is that you can write documentation, and then provide something about who you are. For example, if you're a company, and you write a book documenting some piece of software, you could release it as a traditional copyrighted book. The purpose of the GNU FDL is to allow you to release it for Free, with an invariant section saying "I wrote this, and its writing was funded by So-and-So Publishers; if you'd like to support our work, buy that version."
If this is disallowed, it will not encourage "more-Free" documentation. It will instead encourage people to take their work and publish it as traditional copyrighted documentation.
There's a lot of gee-whiz techno-nonsense reminiscent of early 1980s cyberpunk. Just because you have a robot, a "hacker," and a scantily-clad schoolgirl in your show doesn't mean you can skip the plot.
I'm going to have trouble fitting them all in, what with the Must-See Star Trek Fan Film Festival this week too.
If you read the papers at that site, the empirical conclusion is that the gain of caching large files is outweighed by the cache overhead. Always caching small files has two advantages: 1) you can now access any file (on disk or in memory) with a simple one-level pointer indirection, since you always know small files are in memory and large files are on disk, and have something equivalent to the standard inode table holding both; and 2) you now know every file on disk is "big" (for some value of big) and can optimize accordingly, in the process greatly simplifying your filesystem code (currently filled with lots of hacks for optimizing small-file accesses); an obvious win here is using large cluster sizes (say, 512K instead of 1K).
So what we have here is Debian claiming that the FSF is not radical enough in its promotion of Free Software? That seems a bit ridiculous; even the FSF realizes that allowing the author of documentation the right to include some invariant section (such as an attribution) is necessary to get people to write free documentation at all. I never thought I'd see an argument where RMS would be on the side opposing the zealots.
Perhaps I'm doing something wrong, but the AA in the xft build of Mozilla makes fonts unreadable on black backgrounds. It seems it works by basically blurring the fonts and letting the black bleed into the white a bit, which works beautifully when it's black text on a white background (the black of the text bleeds a little into the white of the background, smoothing out pixellation), but is completely unreadable when it's white text on a black background (the black of the background bleeds into the white that was supposed to be the text). It's not just Mozilla either; the gtk2 builds of gaim have the same problem.
1. There's free access available in many many areas at public libraries and schools, and much of it is not taken advantage of.
2. The same people who "can't afford" a computer can somehow afford all sorts of other equally or more expensive things. Cable TV ($40/mo * 12 mo = $480/yr) is one that comes to mind. Expensive car modifications (stereo systems, etc.) are another.
Most people have free access to the internet through their local public library. The exceptions are some people in very rural areas, but they make up quite a small percentage of the total US population (certainly far less than the 40% non-internet users quoted here). It seems a lot of people simply aren't interested in using the internet, despite free access.
And even if it weren't for that, count the number of people who "can't afford" a computer, but someone "can afford" to pay over $400/yr for cable television.
This is a very insular USENet thing adopted by some segments of Free Software culture, and not at all in keeping with past or present common usage in the computer field or wider culture. As noted in one of the other replies, the usage of "cracker" to describe people who break into computers was coined ca. 1985; the usage of "hacker" to describe these same people dates back to the late 1970s, and was already in very common usage by the early 1980s. For the vast majority of the history of computers, this (someone who breaks into computers) has been the primary meaning of the term "hacker."
You ever listened to any gangsta rap or seen the movie Catch Me If You Can? Both probably have a much bigger influence on the general public.
Disaster relief. It's been proven before that the internet can be resilient to disasters such as earthquake. Useful maybe?
Well, considering that power is often the first thing that goes out in any disaster (be in earthquake, hurricane, etc.), that might limit the usefulness.
Given that the most popular Windows audio player -- Winamp -- comes with Ogg support built into the default install, how does this differentiate Linux?
[This text here because slashdot is broken]
If it was, say, a 1.5 GHz overclocked to 2.0 GHz, and sold as "running at 2.0 GHz," then this was a true statement. As for whether it was misleading, it depends on the chip and the overclocking. It's quite conceivable that some chips could be overclocked by 10-20% and still be more reliable than other chips clocked at the manufacturer's spec'd speed. Should it then be illegal to sell overclocked versions of these chips, even though they're still more reliable than other chips legally on the market?
People's case mods don't make a fucking racket when they drive down the street.
I seem to recall reading an article a while back that Epitaph is actually an RIAA member. I'm not sure why they're not on the list on riaa.org. FWIW, both Epitaph and Fat Wreck Chords are on the RIAA membership list at boycott-riaa.com. Anybody have more details? Maybe they're owned by a parent company on the riaa.org list?
A very large number of 'indie' labels are RIAA members. Check the list, your favorite indie label is probably on it.
Searle's main thrust seems to be that the Chinese Room is an example of simple lookup table responses, which he claims is not understanding. If there was a full Turing machine running the responses (rather than a simple lookup table), it becomes much less clear that he can say there's no understanding going on.
I believe if one accepts Chomsky's theories of language, that it is actually impossible to encode all possible language behavior as an input-output system. I'm not an expert on the subject so I can't say so for sure though. The thrust, though, would be that language has to be encoded in a manner which is more complex than a stimulus-response system because no stimulus-response system can fully implement language; thus, Searle's example of a Chinese room built on a stimulus-response system would be impossible to construct, making his thought experiment incoherent.