But drawing people being torn apart or burned or tortured (non-sexually) is not obscene? I suppose I must have missed the memo explaining how X+sex is evil and must be banned, when X is tolerated and "free speech" and what not.
Linus rightly pointed out, with a degree of tact that Theo de Raadt would be proud of, that writing meta-data before the actual data is committed to disk is a colossally stupid idea. If the journal doesn't accurately describe the actual data on the drive then what is the point of the journal? In fact, it can be LESS than useless if you implicitly trust the inconsistent journal and have borked data that is never brought to your attention.
A definite advantage of writing metadata to the journal first is that you could theoretically retain metadata for both the old and new versions of a file. If the order is metadata->journal, file data -> disk, journal-> disk, then once the metadata is in the journal all the new writes to the file system will have valid metadata pointers to them. Once the file data is written, the disk is updated with the journal.
Not that ext3 necessarily does things this way; I think it applies the journal before doing anything else, which would remove the old metadata pointing to the old version of the file. The journal, data, metadata order works best for things like log based file systems or any file system that doesn't overwrite existing data.
The federal lawyers arguing the government's case should be strip searched every day before entering the Supreme Court. It's only fair, after all, to require the same from adults who are entering what is clearly a more important and sensitive building. I bet half of them are smoking crack anyway, and we need to be absolutely sure that they don't bring it into the court room.
You are new to the company (apparently), and you have discovered that licensing is out of compliance. Use this as leverage (with the BSA as the threat) to get whatever you want.
It's the American Way (which is being exported to every other country on earth as we speak, so don't worry if you're a foreigner).
I know the IETF guys aren't very big on NAT, but it does have one (albeit collateral) advantage - security
Have you audited the code for your cheap NAT box to make sure it only passes packets into your local network that exactly match established TCP connections? You checked for any possible bug in the code that allows new incoming FTP connections? Do machines behind NAT get fewer drive-by installs or something?
Basically, NAT only protects machines from attacks against specific open ports (as long as the NAT device has no flaws). That's something a software firewall on each machine would handle just as well, if not better.
The real problem now is that they have caused an incalculable amount of damage to the reputation of our financial system as being a safe place to invest money. The government has to bail all these people out to show that they will stand behind all these too big to fail crooks and make good on their lies in order to maintain confidence.
That is because it was not a safe place to invest money. "Here, put your money in this magic black box, and you simply can't lose! Even if the black box eats your money, we have an even bigger black box that will ensure your (and everyone else's) original investment!" That smacks of inattention to the basic laws of physics, not to mention economics. TANSTAAFL.
As much as it pains us all, these banks really are too big to fail.
There's a quadrillion *pretend* dollars in derivatives; that's the entire point. No one owns the money they think they do on paper. It doesn't exist anywhere in any tangible good. It was an IOU written to investors that could never be paid. The economy is actually poorer than most people think. The money you invested is *gone*. It was spent by rich people and people who got overvalued loans on their home and spent the difference, or who sold their shares in stocks before the crash. That's the reality that people need to understand.
The way to fix it, basically, is massive socialism to carry people through the hard times of losing most of their retirement, their houses, and their jobs. We can move back to a more capitalist system in the future if it ever looks like a good idea.
Eric Spiegel tells of one such Josh, who wears T-shirts with offensive slogans, insults female co-workers and, when asked about documentation, smirks, "What documentation?' Sure, he was whipsmart and could churn out code that saved the company millions, but can we please stop enabling these people?"
For a second I thought they were starting to describe their CEO or other high level manager who instead of saving millions of dollars, was raking in millions of dollars in salary.
The bottom line demands that Offensive Josh the brilliant coder be retained, and the senior management trashing the company be fired. Won't somebody *please* think of big business's bottom line?
"What the deuce is it to me?" he interrupted impatiently: "you say that we go round the sun. If we went round the moon it would not make a pennyworth of difference to me or to my work."
If, for dramatic purposes, Sherlock Holmes needed to solve a case by determining the length of shadows (or some crappy clue like that) cast by the moon and sun on a particular day, I'm sure the Copernican theory would be conveniently remembered...
Or in other words, unless you want to meld your brain to Google and Wikipedia, it helps to remember a few facts.
Quick, what's the proper race code to send in PID-10? What about PID-17 (that was a fun one to standardize)? Not to mention the mess with PID-18, PID-2 and PID-3 across disparate systems, and every mind-boggling combination of ways that different systems treat persons, encounters, orders, results, reports, and images.
Basically, the government will have to throw out or severely limit the use of most medical software, and enforce its replacement with something standard if they want to make health information electronically available to any provider. Otherwise the "solution" will be sending PNGs of medical records back and forth.
False. Induction is different from deduction, because it is a meta-theoretical method of proof. Given a system of logic, deductions follow as single statements from other statements, but to prove that an infinite number of statements follow from another statement would require an infinite number of deductions, so induction is introduced to allow proofs of categorical statements. If you reject the theorem of induction within a system of logic, you can prove any specific statement, but you will not be able to prove a general theorem about all statements of a given form. e.g. you may prove that 1+2 = 2+1 and 5+20 = 20+5 in basic arithmetic, but you cannot prove that a+b = b+a without induction.
Except that God, taken to mean an actual supernatural divine being such as the Christian God I believe in, does not require an explanation for its origin. "I AM" is His answer to the question of what he is and where he came from.
That, I think, is where scientists and believers differ. Scientists would like to know why, exactly, this particular Christian (or Islamic, or Jewish, or Hindu, etc.) God exists, and not one or more of any number of other possible gods (Zeus, Baal, Ra). Does causality have meaning for gods? Do gods always know they are gods, or do some think they are merely an evolved being in a random universe? In theory, any god would work for intelligent design, the outcome would just be different based on their personality. What is the ordinal number of God in relation to other sets, for instance the universal set of positive set theory or new foundations? Do things exist because god causes them to, or is existence an independent property of things? Are there other possible realities where things exist that even an omnipotent god in our universe has no control (or possible even knowledge) of?
Once those simple questions were answered, it would make sense to discover what sort of being god is, e.g. does perfect omniscience preclude free will or merely break causality (which, in a sense, sort of breaks free will to an extent)? Does omnipotence extend to self-altering actions, including perhaps the ability to give up omnipotence? Basically, the ultimate question is whether the domain of god's omniscience includes god himself, or if god is entirely separate from his range of possible actions (e.g. is god unchangeable?) and furthermore whether or not god plans out all the possible quantum futures of the universe when using omniscience, or only looks down the lines he knows (chooses) the world will go?
Beyond that, it would be nice to know if mathematics is an inherent property of the universe, or perhaps of god himself, that would turn out the same given the same axioms in some other universe, and what the question answered by 42 is.
I think the abortion issue will evaporate as NICUs can save younger and younger premature infants. At some point, it's going to be medically possible to just yank the intact fetus out and grow it for adoptive parents (or perhaps reimplant in a host womb, whatever) and either the people who hate abortions will be lining up at the adoption clinics to adopt the little saved fetuses, or they're going to decide they don't like paying taxes to support the millions of unwanted children on welfare and that abortion is okay.
I don't think the problem was saying that the Sun revolved around the Earth via an Earth-based reference frame. The problem was saying that everything revolved around the Earth, because then the planets and comets had to follow really funny orbits and the idea of general gravitation is pretty much incompatible with geocentrism. These are also people who mostly believed the Earth had planar topology instead of spherical, mind you. It's much easier to believe in geocentrism when gravity is a uniform vector field pointing down at the turtles or Atlas.
The classic Christian tradition has always valued rationality and does not hold that faith involves the abandonment of reason or the absence of evidence. Indeed, the Christian tradition is so strong on this matter that it is often difficult to understand where Dawkins got these ideas.
This is amusing, because a) classic christianity is the whole souls/demons/angels/trinity/creationism/transubstantiation/divine-revelation/inspired-scripture thing, and b) all the things I listed have scant evidence if you try to look for it. The rationalism that christianity presents is an attempt to find a set of axioms in the bible that aren't too self-contradictory and that agree with the era's social mores, derive theorems from there, and explain the obvious contradictions as mysteries of faith (which can't subsequently be used to prove anything you want, like in standard logic).
Most importantly, Christianity offers no discernible evidence that it is the correct choice versus any other religion. Judaism is older, Islam is newer, but Christians claim that the Jewish religion was essentially ended by Jesus (despite Jewish denial of such a thing) and that Islam is false.
You can see evidence of how silly the whole situation is when you consider that the Christians disbelieve "The God Delusion", the "Koran", and "The Talmud (traditional Jewish interpretation)", the Muslims disbelieve "The God Delusion", the "Bible", and "The Talmud (traditional Jewish interpretation)", and the Jews disbelieve "The God Delusion", the "Bible", and the "Koran." The only thing the major religions have in common is fending off a single book that can simultaneously discredit all of them, while disagreeing about exactly why the book is wrong. "It's wrong because it denies Jesus!" "No, it's wrong because it denies Muhammad!" "No, it's wrong because it denies YHWH, n00bs!"
He's a scientist using science to claim a "delusion" in God. It's reasonable to assume he's using the scientific term. If he's claiming you can't disprove God, then where is the evidence to the contrary he is implying by the very title of his book?
Definition of "God" error, basically. The definition that Dawkins presents evidence against is a God that actively changes things in the world today and directly created the world 7000 years ago via special creation. Dawkins cannot present evidence against a deistic god that wound up the universe and let it go, and he does not attempt to argue against such a god (which is not much of a god, really).
If anything, Dawkins' book can be read as "The (personal, loving, etc.) God Delusion", because he is challenging the concept many people have of a friendly omnipotent guy (or trio of guys) in the sky who loves us but damns some of us to hell after testing everyone with pain and suffering in our earthly life, gave us rational minds that should be able to decide what is actually true and false and what makes sense and what doesn't make sense, yet requires blind faith (yes, a belief that pain and suffering in life can be justified by the afterlife requires, literally, blind faith; faith whose ultimate results cannot be seen during earthly life) in order to obtain infinite bliss.
It is a fallacy because it is inductive logic, which is not always true. ...
Immanuel Kant proved that you cannot prove God exists or does not exist by Science long ago. Anything else is pure logical fallacies like inductive logic, which Dawkins uses as well as circular references and wishful thinking.
Well, go ahead and explain how Kant's proof is still valid today (and will still be valid tomorrow). I bet you'll say something like "well, clearly logic isn't changing" but I dare you to use anything other than induction to prove such a statement. Humans inherently use induction when they assume that the universe, logic, or anything maintains its form over time. Specifically, you believe that because the proof has always been valid in the past (P(i), i<N, for the current time N), and a valid proof now is a valid proof in the near future (P(N) -> P(N+epsilon)), inductively the same proof will always be valid (P(t) for all times t).
Since when is shared memory too slow for applications? The connection doesn't have to be merely a FIFO, it can be a shared mmap between the parent process and the untrusted child process. Any IPC mechanism works, as long as the syscalls that implement it are safe for untrusted code to use.
OpenGL functions, for instance, would require wrappers to move untrusted memory into trusted memory, sanity check it, and then perform the calls to the real OpenGL API. This is not as bad as it sounds; textures, models, shaders, etc. would be loaded once into trusted memory and checked, and from there the untrusted application could still reference them. Most games preload that information anyway, and most of the dynamic things in the game are the lights and the matrices for manipulating the vertex/surface lists.
I think pre-allocated memory would be the most useful. Most untrusted apps need to have some sort of resource limitation on them anyway, so it would make sense to define a hard limit for memory usage and just allocate it at the beginning. Since most of it would be unused, it wouldn't actually get physical pages until it's used (in Linux, at least). Nice-ing the process before executing the untrusted code would also probably be a good idea. The best thing is that capabilities on the system calls would essentially be a per-process bitmap with as many entries as the system call table, and require only a couple extra instructions to check on each system call. I might do that later today, just for fun...
for target practice. At least some states would know the right way to use them. Or maybe the governator could balance the state budget by selling them as scrap, or even better to third world dictatorships where they would find their true calling in "democratic" elections.
What annoys me is that by adding simple capabilities to the real operating system's syscalls, the operating system could do the same job as CaCl without having to compile programs specially. It's simple:
1. Open a FIFO (or shared memory, or other IPC method). 2. Fork. 3. Close all file descriptors except the FIFO. 4. Free up unused memory. 5. Drop all capabilities to system calls except for sys_read, sys_write, and sys_exit. 6. Read the code to execute from the FIFO. 7. Execute the code.
As long as the OS does its job, the new process is executing untrusted code perfectly safely. The only communication channel is through the open FIFO, which can be controlled completely by the parent process This requires the OS to handle things like the F00F bug properly, and doesn't account for timing attacks against cryptographic implementations, but I don't think CaCl can prevent timing attacks either. Most importantly, this is platform/architecture independent. It just requires the ability to permanently disable certain system calls for a process.
Slashdot Mirror
But drawing people being torn apart or burned or tortured (non-sexually) is not obscene? I suppose I must have missed the memo explaining how X+sex is evil and must be banned, when X is tolerated and "free speech" and what not.
Linus rightly pointed out, with a degree of tact that Theo de Raadt would be proud of, that writing meta-data before the actual data is committed to disk is a colossally stupid idea. If the journal doesn't accurately describe the actual data on the drive then what is the point of the journal? In fact, it can be LESS than useless if you implicitly trust the inconsistent journal and have borked data that is never brought to your attention.
A definite advantage of writing metadata to the journal first is that you could theoretically retain metadata for both the old and new versions of a file. If the order is metadata->journal, file data -> disk, journal-> disk, then once the metadata is in the journal all the new writes to the file system will have valid metadata pointers to them. Once the file data is written, the disk is updated with the journal.
Not that ext3 necessarily does things this way; I think it applies the journal before doing anything else, which would remove the old metadata pointing to the old version of the file. The journal, data, metadata order works best for things like log based file systems or any file system that doesn't overwrite existing data.
The federal lawyers arguing the government's case should be strip searched every day before entering the Supreme Court. It's only fair, after all, to require the same from adults who are entering what is clearly a more important and sensitive building. I bet half of them are smoking crack anyway, and we need to be absolutely sure that they don't bring it into the court room.
You are new to the company (apparently), and you have discovered that licensing is out of compliance. Use this as leverage (with the BSA as the threat) to get whatever you want.
It's the American Way (which is being exported to every other country on earth as we speak, so don't worry if you're a foreigner).
I know the IETF guys aren't very big on NAT, but it does have one (albeit collateral) advantage - security
Have you audited the code for your cheap NAT box to make sure it only passes packets into your local network that exactly match established TCP connections? You checked for any possible bug in the code that allows new incoming FTP connections? Do machines behind NAT get fewer drive-by installs or something?
Basically, NAT only protects machines from attacks against specific open ports (as long as the NAT device has no flaws). That's something a software firewall on each machine would handle just as well, if not better.
The real problem now is that they have caused an incalculable amount of damage to the reputation of our financial system as being a safe place to invest money. The government has to bail all these people out to show that they will stand behind all these too big to fail crooks and make good on their lies in order to maintain confidence.
That is because it was not a safe place to invest money. "Here, put your money in this magic black box, and you simply can't lose! Even if the black box eats your money, we have an even bigger black box that will ensure your (and everyone else's) original investment!" That smacks of inattention to the basic laws of physics, not to mention economics. TANSTAAFL.
As much as it pains us all, these banks really are too big to fail.
There's a quadrillion *pretend* dollars in derivatives; that's the entire point. No one owns the money they think they do on paper. It doesn't exist anywhere in any tangible good. It was an IOU written to investors that could never be paid. The economy is actually poorer than most people think. The money you invested is *gone*. It was spent by rich people and people who got overvalued loans on their home and spent the difference, or who sold their shares in stocks before the crash. That's the reality that people need to understand.
The way to fix it, basically, is massive socialism to carry people through the hard times of losing most of their retirement, their houses, and their jobs. We can move back to a more capitalist system in the future if it ever looks like a good idea.
Eric Spiegel tells of one such Josh, who wears T-shirts with offensive slogans, insults female co-workers and, when asked about documentation, smirks, "What documentation?' Sure, he was whipsmart and could churn out code that saved the company millions, but can we please stop enabling these people?"
For a second I thought they were starting to describe their CEO or other high level manager who instead of saving millions of dollars, was raking in millions of dollars in salary.
The bottom line demands that Offensive Josh the brilliant coder be retained, and the senior management trashing the company be fired. Won't somebody *please* think of big business's bottom line?
seccomp is almost exactly what I was thinking of, but written about 4 years ago. All the good ideas are implemented by the time I think of them...
"What the deuce is it to me?" he interrupted impatiently: "you say that we go round the sun. If we went round the moon it would not make a pennyworth of difference to me or to my work."
If, for dramatic purposes, Sherlock Holmes needed to solve a case by determining the length of shadows (or some crappy clue like that) cast by the moon and sun on a particular day, I'm sure the Copernican theory would be conveniently remembered...
Or in other words, unless you want to meld your brain to Google and Wikipedia, it helps to remember a few facts.
AOL caught on eventually. Compuserve never did, that I'm aware of. Free trials aplenty with a valid check digit.
Quick, what's the proper race code to send in PID-10? What about PID-17 (that was a fun one to standardize)? Not to mention the mess with PID-18, PID-2 and PID-3 across disparate systems, and every mind-boggling combination of ways that different systems treat persons, encounters, orders, results, reports, and images.
Basically, the government will have to throw out or severely limit the use of most medical software, and enforce its replacement with something standard if they want to make health information electronically available to any provider. Otherwise the "solution" will be sending PNGs of medical records back and forth.
Unless users are only given a restricted shell, what prevents them from writing applications in shell script and running them?
It's either a kiosk or a fully functional Universal Turing Machine...
False. Induction is different from deduction, because it is a meta-theoretical method of proof. Given a system of logic, deductions follow as single statements from other statements, but to prove that an infinite number of statements follow from another statement would require an infinite number of deductions, so induction is introduced to allow proofs of categorical statements. If you reject the theorem of induction within a system of logic, you can prove any specific statement, but you will not be able to prove a general theorem about all statements of a given form. e.g. you may prove that 1+2 = 2+1 and 5+20 = 20+5 in basic arithmetic, but you cannot prove that a+b = b+a without induction.
Except that God, taken to mean an actual supernatural divine being such as the Christian God I believe in, does not require an explanation for its origin. "I AM" is His answer to the question of what he is and where he came from.
That, I think, is where scientists and believers differ. Scientists would like to know why, exactly, this particular Christian (or Islamic, or Jewish, or Hindu, etc.) God exists, and not one or more of any number of other possible gods (Zeus, Baal, Ra). Does causality have meaning for gods? Do gods always know they are gods, or do some think they are merely an evolved being in a random universe? In theory, any god would work for intelligent design, the outcome would just be different based on their personality. What is the ordinal number of God in relation to other sets, for instance the universal set of positive set theory or new foundations? Do things exist because god causes them to, or is existence an independent property of things? Are there other possible realities where things exist that even an omnipotent god in our universe has no control (or possible even knowledge) of?
Once those simple questions were answered, it would make sense to discover what sort of being god is, e.g. does perfect omniscience preclude free will or merely break causality (which, in a sense, sort of breaks free will to an extent)? Does omnipotence extend to self-altering actions, including perhaps the ability to give up omnipotence? Basically, the ultimate question is whether the domain of god's omniscience includes god himself, or if god is entirely separate from his range of possible actions (e.g. is god unchangeable?) and furthermore whether or not god plans out all the possible quantum futures of the universe when using omniscience, or only looks down the lines he knows (chooses) the world will go?
Beyond that, it would be nice to know if mathematics is an inherent property of the universe, or perhaps of god himself, that would turn out the same given the same axioms in some other universe, and what the question answered by 42 is.
I think the abortion issue will evaporate as NICUs can save younger and younger premature infants. At some point, it's going to be medically possible to just yank the intact fetus out and grow it for adoptive parents (or perhaps reimplant in a host womb, whatever) and either the people who hate abortions will be lining up at the adoption clinics to adopt the little saved fetuses, or they're going to decide they don't like paying taxes to support the millions of unwanted children on welfare and that abortion is okay.
Oh, for mod points...
Is the cannibalism of your opponent an ad-hominem attack?
I don't think the problem was saying that the Sun revolved around the Earth via an Earth-based reference frame. The problem was saying that everything revolved around the Earth, because then the planets and comets had to follow really funny orbits and the idea of general gravitation is pretty much incompatible with geocentrism. These are also people who mostly believed the Earth had planar topology instead of spherical, mind you. It's much easier to believe in geocentrism when gravity is a uniform vector field pointing down at the turtles or Atlas.
The classic Christian tradition has always valued rationality and does not hold that faith involves the abandonment of reason or the absence of evidence. Indeed, the Christian tradition is so strong on this matter that it is often difficult to understand where Dawkins got these ideas.
This is amusing, because a) classic christianity is the whole souls/demons/angels/trinity/creationism/transubstantiation/divine-revelation/inspired-scripture thing, and b) all the things I listed have scant evidence if you try to look for it. The rationalism that christianity presents is an attempt to find a set of axioms in the bible that aren't too self-contradictory and that agree with the era's social mores, derive theorems from there, and explain the obvious contradictions as mysteries of faith (which can't subsequently be used to prove anything you want, like in standard logic).
Most importantly, Christianity offers no discernible evidence that it is the correct choice versus any other religion. Judaism is older, Islam is newer, but Christians claim that the Jewish religion was essentially ended by Jesus (despite Jewish denial of such a thing) and that Islam is false.
You can see evidence of how silly the whole situation is when you consider that the Christians disbelieve "The God Delusion", the "Koran", and "The Talmud (traditional Jewish interpretation)", the Muslims disbelieve "The God Delusion", the "Bible", and "The Talmud (traditional Jewish interpretation)", and the Jews disbelieve "The God Delusion", the "Bible", and the "Koran." The only thing the major religions have in common is fending off a single book that can simultaneously discredit all of them, while disagreeing about exactly why the book is wrong. "It's wrong because it denies Jesus!" "No, it's wrong because it denies Muhammad!" "No, it's wrong because it denies YHWH, n00bs!"
He's a scientist using science to claim a "delusion" in God. It's reasonable to assume he's using the scientific term. If he's claiming you can't disprove God, then where is the evidence to the contrary he is implying by the very title of his book?
Definition of "God" error, basically. The definition that Dawkins presents evidence against is a God that actively changes things in the world today and directly created the world 7000 years ago via special creation. Dawkins cannot present evidence against a deistic god that wound up the universe and let it go, and he does not attempt to argue against such a god (which is not much of a god, really).
If anything, Dawkins' book can be read as "The (personal, loving, etc.) God Delusion", because he is challenging the concept many people have of a friendly omnipotent guy (or trio of guys) in the sky who loves us but damns some of us to hell after testing everyone with pain and suffering in our earthly life, gave us rational minds that should be able to decide what is actually true and false and what makes sense and what doesn't make sense, yet requires blind faith (yes, a belief that pain and suffering in life can be justified by the afterlife requires, literally, blind faith; faith whose ultimate results cannot be seen during earthly life) in order to obtain infinite bliss.
It is a fallacy because it is inductive logic, which is not always true.
...
Immanuel Kant proved that you cannot prove God exists or does not exist by Science long ago. Anything else is pure logical fallacies like inductive logic, which Dawkins uses as well as circular references and wishful thinking.
Well, go ahead and explain how Kant's proof is still valid today (and will still be valid tomorrow). I bet you'll say something like "well, clearly logic isn't changing" but I dare you to use anything other than induction to prove such a statement. Humans inherently use induction when they assume that the universe, logic, or anything maintains its form over time. Specifically, you believe that because the proof has always been valid in the past (P(i), i<N, for the current time N), and a valid proof now is a valid proof in the near future (P(N) -> P(N+epsilon)), inductively the same proof will always be valid (P(t) for all times t).
Since when is shared memory too slow for applications? The connection doesn't have to be merely a FIFO, it can be a shared mmap between the parent process and the untrusted child process. Any IPC mechanism works, as long as the syscalls that implement it are safe for untrusted code to use.
OpenGL functions, for instance, would require wrappers to move untrusted memory into trusted memory, sanity check it, and then perform the calls to the real OpenGL API. This is not as bad as it sounds; textures, models, shaders, etc. would be loaded once into trusted memory and checked, and from there the untrusted application could still reference them. Most games preload that information anyway, and most of the dynamic things in the game are the lights and the matrices for manipulating the vertex/surface lists.
I think pre-allocated memory would be the most useful. Most untrusted apps need to have some sort of resource limitation on them anyway, so it would make sense to define a hard limit for memory usage and just allocate it at the beginning. Since most of it would be unused, it wouldn't actually get physical pages until it's used (in Linux, at least). Nice-ing the process before executing the untrusted code would also probably be a good idea. The best thing is that capabilities on the system calls would essentially be a per-process bitmap with as many entries as the system call table, and require only a couple extra instructions to check on each system call. I might do that later today, just for fun...
for target practice. At least some states would know the right way to use them. Or maybe the governator could balance the state budget by selling them as scrap, or even better to third world dictatorships where they would find their true calling in "democratic" elections.
What annoys me is that by adding simple capabilities to the real operating system's syscalls, the operating system could do the same job as CaCl without having to compile programs specially. It's simple:
1. Open a FIFO (or shared memory, or other IPC method).
2. Fork.
3. Close all file descriptors except the FIFO.
4. Free up unused memory.
5. Drop all capabilities to system calls except for sys_read, sys_write, and sys_exit.
6. Read the code to execute from the FIFO.
7. Execute the code.
As long as the OS does its job, the new process is executing untrusted code perfectly safely. The only communication channel is through the open FIFO, which can be controlled completely by the parent process This requires the OS to handle things like the F00F bug properly, and doesn't account for timing attacks against cryptographic implementations, but I don't think CaCl can prevent timing attacks either. Most importantly, this is platform/architecture independent. It just requires the ability to permanently disable certain system calls for a process.