There were predictions back 40 years ago. Oh, things like ozone holes, stuff like that. NASA eventually started looking for them, but had some trouble at first. The holes were so f*** large that their computer software was rejecting them as impossible.
I guess that 40 years ago, it would have been within the knowledge and ability of people to predict that cutting down the forests in Africa would cause a drought. Certainly, it's indisputable that humanly-deforested regions have suffered longer, more severe droughts since being deforested than at any time prior.
In recent years, there has been strong evidence that zooplankton levels are inversely proportional to temperature - cooler weather, more plankton; hotter weather, less plankton.
Does this mean that global warming is real? Define real. The globe is warming, that's irrefutable. Is it caused by human activity? Well, define activity - are you including deforestation, pollution, changes in the biological infrastructure of the planet, etc? Or just a select set of these? Also, and this is the billion dollar question, how much does the cause matter? If the planet is warming to the point where the current life is incapable of survival, who gives a damn about the causes? The latency inherent in the system is on the order of decades to centuries - changing the causes today won't be fast enough to stop the planet overheating, even if all causes WERE under human control. Why not take care of the problem right now and address the causes when we've got time?
I do believe humans are the primary cause, because although natural sources are often much greater, they are much more sporadic and much more regional. Humans have generated non-local sustained inputs, and those simply didn't exist before. Nor is the process linear. Not even remotely close. Saying that X is greater than Y by a factor of Z is only useful if you can use Z to make some useful observation. If the system is non-linear with both positive feedback and negative feedback loops that are themselves non-linear, you have what is known as a chaotic system. Chaotic systems have two properties - they are acutely sensitive to initial conditions, so any error in measurement will explode out of all proportion in almost no time at all, and they are non-differentiable, so that you can't accurately solve any given step even if you DID know the initial conditions. This means that you cannot directly equate human activity with natural activity and hope to get useful results. The best you can do is equate mechanisms and distributions to see what MIGHT be comparable.
However, my opinion of human activity is of no consequence. If humans cut out all pollution tomorrow, we would not start to see the benefits until a hundred or so years after global warming reached crisis point. If you want to do something effective, don't target the stuff that is pointless. Fixing human activity is like re-wallpapering a house that's on fire. Some things can be left to later.
A virus bootstrap gets onto the machine by some means, then uses RDMA to transfer the rest of the virus from an infected machine elsewhere. The connection is initiated on the inside and is therefore technically an outbound connection even though the important traffic is going the other way.
A program that is supposed to be running has a feature to connect to some server or other at periodic intervals for updates, genuine-ness certification, etc. Someone poisons a router table or DNS cache and hijacks that traffic in order to feed the program a virus or trojan through its own update mechanism.
You install a program from a source you don't entirely trust. You think it might contain spyware, but your spyware scanner doesn't show anything. To be sure, you want to totally lock it down.
An attack vector is discovered on the firewall such that the attacker can cause the firewall machine to either initiate - or think it initiated - a connection to the attacker.
Are these likely? Well, some have probably already occurred, others might never occur, and the rest might happen to your computer as you are reading this text. The whole point is that you have zero idea what form the next attack will be in because it hasn't happened yet and you have zero idea of what is trustable because the chain of trust is very very long and almost totally outside the scope of us mere mortals to validate. ANY break ANYWHERE in that chain can invalidate the assumptions made on a one-sided firewall. If components outside of the firewall were safe, you wouldn't need the firewall. Since the firewall is of value, you are implying that total trust on outside components is unsafe and by inference that a total trust that the outside will make no attempt to appear to be on the inside is also unsafe.
You label everything as one or more of "sushi", "chowder" or "probably made into soup somewhere". Saves on physiological and genetic analysis, and it's all that Joe Average is likely to care about. (If the average person gave a rat's about conservation or science, we'd be a thousand years more advanced and ten thousand years wiser.)
Besides, in 15 years or less there won't be enough of a food chain in the oceans to sustain most of the organisms that do still exist and without a gene bank capable of storing that kind of volume of information there's no possibility of either having any usable data OR being able to revive the ecology once conditions have returned to saner levels. Collecting photos is all fine and good, but in not that long a time that is ALL we'll have, unless serious efforts are made to either conserve or genetically catalog.
(And, frankly, I can't see the US Government even getting past the planning stages in a mere 15 years - assuming it even got that far. As they're the only group with the clout and the money to build a center capable of analyzing and storing a few hundred million DNA/mtDNA databases in that kind of timeframe, most of the information currently in the oceans is beyond any possibility of recovery.)
Everything in Australia is deadly. The spiders are deadly, the snakes are deadly, the crocodiles are deadly, the plants are deadly, the driving in Sydney is definitely deadly, the TV commercials are lethal... I never did find out what happened to those rabbits that escaped from a research facility on a Government-owned island and made it to shore, back in '95. As I recall, they were being used for some research into some lethal pathogen or other. Since there are Australians still alive, I take it that the crisis was brought under control, but that was cutting it a little fine. I guess we can add the Australian Government to things that are lethal, though.
(John Wyndham's novel still gives me the creeps. That and Day of the Triffids. But as best as I can tell, the survey hasn't turned up any underwater Triffids. Yet.)
The 286 had an instruction to load/save the complete state of the processor from storage in one go. Later processors did not have this capability, which is frankly pathetic. Ideally, you'd have such an instruction on all processors (GPUs, FPUs, CPUs, etc) which would reduce such power-to-known-state operations to a very short time.
The best I've seen, in terms of imaging, is the work on LinuxBIOS. This transfers your initial kernel and initial ramdisk (if any) from Flash, but that's not really improving the method. It only improves what you use the existing method with.
One of the problems mentioned a lot is with initializing hardware. Well, let's be honest about this - you never actually NEED to initialize most of the hardware at boot time anyway. It only needs to be fully operational at first use. Furthermore, most initialization can be done in parallel and can be entirely off-loaded. I see absolutely no reason why you couldn't have a PCI card or an FPGA chip on the motherboard whose sole function was to set up pieces of hardware. Offload the whole damn plug-and-play crap from the kernel and put it in hardware. It's better there, because modifying the state of hardware can disrupt the kernel a whole lot more than it can disrupt a solid-state circuit. Software PnP is dangerous, because the kernel runs on the very hardware that could either be removed or disrupted. (A single power spike could ruin your computer's whole day.)
(In case anyone hasn't noticed, I tend to like hardware solutions. Wafer technology isn't at the point where you could reduce the entire Linux kernel to a series of gates and registers - besides which, my hands aren't steady enough to apply the patches you'd need - but software is only good for things that are too complex or too dynamic for hardware. A full hardware detection sequence, initialization and registration system is still well within the bounds of what hardware can do vastly better than software. Stuff that is so utterly non-standard that it needs its own unique init sequence would be a problem, but since I'd like to see standards violators deported to the far side of the moon, there are certain advantages in having a system that boots rationally-designed hardware the best.)
The "traditional" network stack shuffles data from the application from memory into the CPU then into kernel memory for running through the protocol stack then into the CPU then into kernel memory for running into the actual NIC driver then into the CPU then onto the bus. This assumes that the data doesn't fall out of cache during any of the processing of the packet.
A better design would be to have networked data in well-defined regions which the card can DMA directly out of and into. The PCI bus can handle a 4K transfer as an atomic operation on a single channel, so a 4 channel PCI card can simultaneously send and receive streams of jumbo packets without requiring any CPU intervention. The driver would merely need to be passed two lists of physical page pointers - one for inputs, one for outputs - for each of the open connections and to pass back signals from the board that an entire packet and/or message had been uploaded for a given connection.
(Jumbo packets can go up to 8K, and you can do 8K over two PCI channels, so that's a good unit to be working with.)
The next improvement that can be made to such a system is to improve the buffering. The network will be slower than the computer, almost always, so being able to queue up multiple packets for sending is a Good Thing. Filtering out packets that have not been sent but are no longer worth sending can be done entirely in parallel and does not require anything extra at the end of the pipeline. Not all packets are of equal value, though. You want to deliver the information in the order that will give the best possible benefit - which may not be the order in which the program generates the traffic. Hierarchical Fair Service Curve, Class Based Queueing, and a bunch of other similar techniques, have been developed to fix exactly that sort of problem.
If the ISPs would stop being so bloody stupid, you could also enable protocols such as your basic multicasting (for your UDP stuff) and Scalable Reliable Multicast (for the stuff that needs to reliably get through). That hacks, slashes, butchers and roasts (with just a hint of parsley) problems associated with sending identical state information to multiple end-points.
Bear in mind that Myrinet, Dolphinics, and a bunch of other vendors, use essentially the above mechanisms already and are achieving latencies in the region of 2.5 - 3 microseconds. I say essentially, because I'm not convinced they've optimized quite to the degree I'm suggesting - reliable, scalable multicast RDMA isn't something you'll see a lot of even at a supercomputer fair. True, you're not getting that kind of latency over the Internet whatever you do, but if you can achieve a hard real-time guarantee of 3 microsecond delivery in a LAN party, you WILL notice a difference. At the very least, in the door price, which will now be expressed in exponential notation to fit on the door.
The good news is that since the irrational have created most of the world's problems, and are a primary component of the planetary overpopulation, the problem should eventually correct itself. It generally has, in the past. Mind you, it usually took with it most of civilization, but Slashdot keeps archives so we can recover fairly quickly.
Yeah, that would definitely work. In fact, because there's no distributed locking mechanism (that I know of) in the brain or any packet labeling, a race condition can occur not just between any two processes but also between any two (or more) neurons and between any two (or more) neural pathways. You can also get something similar to a packet collision when backwash from a previous signal corrupts a subsequent one. And as the topology is a mesh, you can get cyclic and self-referential inputs, which means that a race condition need not occur on a specific iteration, only on some iteration.
Blake's 7 got that one right, in that you almost never got the "World War I Dogfight" space combat. Mind you, how much of that was due to skill and how much was due to a budget you could barely feed a canary with is up for debate. The same goes for the few flickers of space combat in The Tomorrow People and most of what you get in Doctor Who in that regard. Most of the other British telefantasy series I can think of that involved space (such as the Quatermas series) avoided the issue entirely.
...based on a remarkably stupid theory. I'm going to type this out slowly, so that the cognitive scientists out there can follow what I'm saying. Deja Vu may occur in any number of possible ways, but the human brain needs to be able to recognize ANY stimulus extremely quickly, for survival reasons. Vision, if anything, should be the least of the senses that gets such verification, because if you're looking at a threat directly, you probably don't need to remember that it was a threat the last time. It should be pretty obvious. Sound, smell, taste, touch -- these contain far less information to start with, so increasing the odds of a false positive, but need to be checked far more thoroughly because potential hazards can be much less obvious.
A false positive is bad, especially if there are far too many, but a false negative can be lethal. This would be more true, say, 100,000 years ago than today, and that's when most of these mechanisms became as finely tuned as they are. Back in the days when hominids were trudging through deadly terrain, you had to remember places and situations that were Bad News with enough time to get clear. In those days, there was a shortage of humvees, so having time to get clear meant having extremely early warning. From that, Deja Vu is a very obvious, direct consequence. In fact, no matter how good humans may have been at avoiding such situations, Deja Vu would always be selected for far more often than against.
(The above can be translated by crypto geeks as follows: The brain has a really crappy but very very fast hashing algorithm used to label sensory data. It's so fast that being crappy doesn't hurt survival chances, but it's crappy enough that we are seeing a very large number of hashing collisions.)
Now, here is where it gets fun. The senses are all cross-linked and cross-referenced in the brain. When the barriers in the brain don't work as expected, we get synaesthesia. Now, it is not at all obvious where the comparison is made, or how the barriers work. For this reason, it is entirely possible to imagine a situation where data from sense A is compared with a prior input from sense B. All it would take is for the barrier to fail to work correctly for recalled data, even if it worked just fine otherwise. This is not "classic" Deja Vu, because the brain is not incorrectly matching an experience with a prior experience of the same sense - it is incorrectly matching totally different types of data. Is this possible? Depends. Any connection that is bi-directional in the brain by nature can fail to mask or block data in either direction, so I can see absolutely no reason why - given synaesthetes are proof that the failure can occur one way - it cannot fail on recall.
(There are soooo many brain disorders associated with inexplicable associations, spooky feelings and false associations that you could fund half the field of neurology for the next fifty years just looking at sensory mismatches and nothing else. Given that, I'd call it almost a flat-out certainty that some of these experiences are cross-sensory errors that involve some of the same matching failures as Deja Vu.)
...but I believe the GPP's intent was to say that lawsuits for otherwise-resolvable, trivial and/or blatantly frivolous matters have produced an elements of skepticism (both in the US and internationally) as to the credibility of the US legal system, fairly or unfairly. This is not to say the matters should not be resolved according to the rule of law, or by any other applicable code that may apply. Rather, it is to say that there appears to be insufficient deterrant for using the courts as a dumping-ground for trivially-resolvable disputes and/or a source of income for professional liars.
eg: The RIAA have tried to sue people who have never owned a computer or an Internet account for file swapping. Firstly, this would appear to be trivially-resolvable by any competent arbitrator, it most definitely does not need to be demanding vast amounts of time from an already-overloaded court system. Secondly, it is the understanding of us non-lawyers that the worst the RIAA can get for wasting the time and money of the legal system is a rap on the knuckles for a frivolous lawsuit - the defendant is most unlikely to be reimbursed for time and costs involved - which directly implies that it is cheaper to sue first and ask questions later.
Because the rewards are perceived to be high (whether they are in practice or not) and the risks are perceived to be low (ditto), the courts appear to have become the first resort, not the last resort. No matter how unjustified such a perception may be in reality, it is nonetheless the perception that has arisen and that is seriously damaging to the credibility of the system as a whole.
Personally, I would like to see the courts have greater power to call bull - whether by the plaintiff or the defendant - and greater flexibility in the handling of what can only be called abuse of court. That should include the ability to impose fines or jail time on plaintiffs (or defendants) even outside of the frivolous lawsuit mechanism or the final verdict. There may also be problems with the public defender system, as they have developed rather a bad reputation over the years. If the courts need to supervise such people, then they should be given the power to do so.
Does this impinge on a person's right to a trial? No. I'm not saying anything about denying a person a right to a trial, but rather that such a right does NOT imply a right to a trial first, OR a right to use the mere act of having a trial as a means of inflicting punishment on a person if that person is innocent, and certainly does not imply a right to use the courts for entertainment or get-rich-quick purposes.
(That second one is tough. Time is money. Even if all other expenses are either taken care of or reimbursed afterwards, if a person is in court and is not on the court's payroll, then they are not at work. For low-income individuals or individuals who don't have much of a buffer for whatever reason, this can make it impossible for that person to argue their case meaningfully or - in some cases - at all. I don't know how you can easily close that loophole, but this is essentially a denial-of-service attack, and the courts should never tolerate being used as a weapon. They are there to judge on matters of law, they are not there as a cheap alternative to hiring a hitman.)
No, the matches were all positive, but they were all in Reed's matchbox. The dangerous thing with unspecifics is that it's not always obvious as to what the unspecifics are unspecific about.
The mathematicians invented nullity (see later slashdot article on dividing by zero), so the physicists now have to one-up them on nothingness. I'm expecting any day now to hear NASA announce the discovery of planets that don't actually exist.
That presumably depends on the nature of the link. If something is forcibly linked by the compiler such that symbols from the object/library being linked to are embedded in the code, then you've got contamination. The same would be true if you include header files, and absolutely definitely unquestionably true for any form of static link.
In the case of readline, that would be an upstream link and I could see potential licensing issues there, as you are essentially including GPL code in a non-GPL object. That would definitely be on the Forbidden List. Downstream is slightly different - you can run GCC under non-GPL'ed OS', even though there must be links GCC must use that are not GPL'ed. (It is possible, I suppose, that Cygwin re-implements the BIOS, has its own screen manager so that X will work, etc, but me thinks not.:)
In this case, a better example might be a use of dlopen(). If person A wrote some code that installed a file of a specific name and called specific functions within that file, with ALL of that interface under the GPL, then if some such file happened to not be GPL, I don't see that you would be retrospectively violating the license. The program has not been changed - on disk or in memory. Everything is exactly as it was, with the sole difference that the pointers now point to something, where that something is wholly external and wholly black-box.
(If you were to ask me if I like closed source - whether as a module or in any other form - I'd say no. Corporations HAVE to compile to the lowest common denominator, which means I can always optimize better than them. Corporations CANNOT include capabilities as fast as the total IT market is capable of creating them, which means that I am better equipt to ensure I have the feature set I need. Corporations also have to make assumptions that may - or may not - apply either the typical user or the stereotypically-dumb user, so I am in an infinitely superior position to have code that functions for me, operates the way I think, follows my mental picture of the system in question. Closed-source, by its very nature, has to be a compromise hack. It can't be anything else. Open source often is a compromise hack, but that is entirely by choice, as stupid as I think such a choice is.)
It depends. Dynamic links are generally not covered by the GPL, as they are not "really" modifications of the code. They are merely calls to an external object, and totally external calls are not usually considered within the remit of the GPL. (By totally external, I mean that no sane person could say that the object was in any way embedded - even at runtime - into the original code.) If you add such a link to an external object that handles Open XML, then the object handling Open XML need not be GPLed, whereas the code that is identifiably OpenOffice would remain GPL.
This would be modular if (and only if) you could remove said link from the code and have it still work. I think the word WinDriver is appropriate here. Microsoft has, in the past, found ways to shift functionality around to break things when not doing things their way, even though "technically" they are not doing so. The hardware in a WinPrinter or WinModem doesn't change when you move it to Linux, it still functions entirely within spec, it's not its fault that Linux lacks the necessary extra code.
Alternatively, Microsoft could overload one of the Open Office functions in a way that makes Open Office run better (or appear to) with the module than without. Or they could make it flakier to use Open Document. There's a million ways they could coerce users into using their module. And, as with the browser wars, all they need is to make themselves appear needed.
Now, will this happen? I'm not sure. Novell seem suspicious of Microsoft, but the test of a trap is not whether you are suspicious of it, but whether you are caught. (Kerr Avon, "Bounty", Blake's 7) It also seems odd that - at a time the community is suspicious of the whole relationship - Novell would be doing this. It seems unhelpful for customer relationships (or anything else) to add fuel to the fire, no matter how innocent the whole thing is. There have simply been too many cases of innocent victims (users and businesses) in the past for people to simply relax. One should not be too relaxed around a vampire, even if they claim to have become vegetarian. (Vegetarian vampire ducks excluded.)
Is this a fork? I don't think it matters what it is - if it's safe, then it's helpful. If it's unsafe, it'll be lethal. The name on the bottle really doesn't count for much.
If the stuff's important, then they can always use a secure VPN tunnel. OpenCA will let you roll your own certs, so those cost nothing, and telling IPSec or SSH to validate on a cert is a piece of cake. It's all transparent to the user, the user doesn't need to learn anything fancy, the automated scripts would not even need touching, but you've now got a level of security that is at least beyond the average 5 year old.
Weather stuff and other public information need to be protected only insofar as they should be read-only. You do NOT want a spotty 16 yr old converting their porn collection into the correct format for the weather database. Well, this IS Slashdot, so maybe I can't be quite so sure about that... Anyways, set up correctly, the system would use mandatory access controls (SELinux would work fine, though they probably have a spare copy of Trusted Irix somewhere) or a read-only network file system for the data with the actual writing done on a secure machine.
This stuff really isn't rocket science......of course, that might be why rocket scientists don't seem very good at it...
...there are no counter-claims being stifled, but I have heard that there have been some unfortunate accidents involving extreme ultraviolet light exposure whilst being nailed (in an ecologically sound fashion) to a rapidly melting iceberg.
Seriously, the press has been full of "counter-claims" from all kinds of nuts, and the only known examples of censorship have all been from the anti-Global-Warming lobby in attempts to stifle NASA scientists, et al. (On that basis alone, one might argue that if the anti lobby was being stifled, it's merely the bad karma they themselves polluted the environment with. If you don't want others to do unto you, it helps to not try to stiff them first.)
The closest you can get to valid metadata is to start by mapping the topology of the information. In the trivial sense, you simply look at all pages that cite or are cited by the page you want to rank, and look at what they discuss. A more complex system does this recursively, with a steadily-falling contribution the greater the link distance becomes. (This would also limit Google Bombing, as you aren't simply working off a single direction over a single step.) You can enhance this further by grouping pages by perhaps shortest distance or median distance to give you one dimension, and then by primary subject matter to give you a second dimension. You then have one keyword per subtopic per planar ring, whos weight is the fraction of that ring that subtopic exists in multiplied by the fraction of the ring that intersects the original page divided by the diameter of the ring.
Oh. You want your search results the same year. Forget it, then. Language analysis on modern computers is way too primitive to give you useful, bomb-proof results over indeterminantly-overlapping web pages where those pages may overlap by different amounts depending on the exact relationship being examined at the time, especially if multi-generational relationships are considered.
You are simply not going to get useful results this side of 2100, unless you are either willing to put up with artificial stupidity (the polar opposite of artificial intelligence) or really bad latency. Artificial stupidity is, sadly, how most search engines operate - usually on the very persuasive grounds that users don't expect anything useful and so don't mind trawling through a billion pages of adverts and fake links in the hope of picking up a nugget or two. Hell, with the click-through model, it might even be more profitable to produce endless junk and a bunch of adverts than to produce something of quality. I could easily see that. Users faced with what they want would have no incentive to give up and look at the banners instead.
Slashdot Mirror
I guess that 40 years ago, it would have been within the knowledge and ability of people to predict that cutting down the forests in Africa would cause a drought. Certainly, it's indisputable that humanly-deforested regions have suffered longer, more severe droughts since being deforested than at any time prior.
In recent years, there has been strong evidence that zooplankton levels are inversely proportional to temperature - cooler weather, more plankton; hotter weather, less plankton.
Does this mean that global warming is real? Define real. The globe is warming, that's irrefutable. Is it caused by human activity? Well, define activity - are you including deforestation, pollution, changes in the biological infrastructure of the planet, etc? Or just a select set of these? Also, and this is the billion dollar question, how much does the cause matter? If the planet is warming to the point where the current life is incapable of survival, who gives a damn about the causes? The latency inherent in the system is on the order of decades to centuries - changing the causes today won't be fast enough to stop the planet overheating, even if all causes WERE under human control. Why not take care of the problem right now and address the causes when we've got time?
I do believe humans are the primary cause, because although natural sources are often much greater, they are much more sporadic and much more regional. Humans have generated non-local sustained inputs, and those simply didn't exist before. Nor is the process linear. Not even remotely close. Saying that X is greater than Y by a factor of Z is only useful if you can use Z to make some useful observation. If the system is non-linear with both positive feedback and negative feedback loops that are themselves non-linear, you have what is known as a chaotic system. Chaotic systems have two properties - they are acutely sensitive to initial conditions, so any error in measurement will explode out of all proportion in almost no time at all, and they are non-differentiable, so that you can't accurately solve any given step even if you DID know the initial conditions. This means that you cannot directly equate human activity with natural activity and hope to get useful results. The best you can do is equate mechanisms and distributions to see what MIGHT be comparable.
However, my opinion of human activity is of no consequence. If humans cut out all pollution tomorrow, we would not start to see the benefits until a hundred or so years after global warming reached crisis point. If you want to do something effective, don't target the stuff that is pointless. Fixing human activity is like re-wallpapering a house that's on fire. Some things can be left to later.
Those sheep are baaaaad news. They could lamb-chop you in half before you could say "mint sauce".
Are these likely? Well, some have probably already occurred, others might never occur, and the rest might happen to your computer as you are reading this text. The whole point is that you have zero idea what form the next attack will be in because it hasn't happened yet and you have zero idea of what is trustable because the chain of trust is very very long and almost totally outside the scope of us mere mortals to validate. ANY break ANYWHERE in that chain can invalidate the assumptions made on a one-sided firewall. If components outside of the firewall were safe, you wouldn't need the firewall. Since the firewall is of value, you are implying that total trust on outside components is unsafe and by inference that a total trust that the outside will make no attempt to appear to be on the inside is also unsafe.
Shouldn't that be: TAE TLA?
Besides, in 15 years or less there won't be enough of a food chain in the oceans to sustain most of the organisms that do still exist and without a gene bank capable of storing that kind of volume of information there's no possibility of either having any usable data OR being able to revive the ecology once conditions have returned to saner levels. Collecting photos is all fine and good, but in not that long a time that is ALL we'll have, unless serious efforts are made to either conserve or genetically catalog.
(And, frankly, I can't see the US Government even getting past the planning stages in a mere 15 years - assuming it even got that far. As they're the only group with the clout and the money to build a center capable of analyzing and storing a few hundred million DNA/mtDNA databases in that kind of timeframe, most of the information currently in the oceans is beyond any possibility of recovery.)
Everything in Australia is deadly. The spiders are deadly, the snakes are deadly, the crocodiles are deadly, the plants are deadly, the driving in Sydney is definitely deadly, the TV commercials are lethal... I never did find out what happened to those rabbits that escaped from a research facility on a Government-owned island and made it to shore, back in '95. As I recall, they were being used for some research into some lethal pathogen or other. Since there are Australians still alive, I take it that the crisis was brought under control, but that was cutting it a little fine. I guess we can add the Australian Government to things that are lethal, though.
(John Wyndham's novel still gives me the creeps. That and Day of the Triffids. But as best as I can tell, the survey hasn't turned up any underwater Triffids. Yet.)
No, they dream in salt water. Now, the salt water might dream in octal, but that's another matter.
Yeah, but the t-shirts'll crinkle due to continental drift!
The best I've seen, in terms of imaging, is the work on LinuxBIOS. This transfers your initial kernel and initial ramdisk (if any) from Flash, but that's not really improving the method. It only improves what you use the existing method with.
One of the problems mentioned a lot is with initializing hardware. Well, let's be honest about this - you never actually NEED to initialize most of the hardware at boot time anyway. It only needs to be fully operational at first use. Furthermore, most initialization can be done in parallel and can be entirely off-loaded. I see absolutely no reason why you couldn't have a PCI card or an FPGA chip on the motherboard whose sole function was to set up pieces of hardware. Offload the whole damn plug-and-play crap from the kernel and put it in hardware. It's better there, because modifying the state of hardware can disrupt the kernel a whole lot more than it can disrupt a solid-state circuit. Software PnP is dangerous, because the kernel runs on the very hardware that could either be removed or disrupted. (A single power spike could ruin your computer's whole day.)
(In case anyone hasn't noticed, I tend to like hardware solutions. Wafer technology isn't at the point where you could reduce the entire Linux kernel to a series of gates and registers - besides which, my hands aren't steady enough to apply the patches you'd need - but software is only good for things that are too complex or too dynamic for hardware. A full hardware detection sequence, initialization and registration system is still well within the bounds of what hardware can do vastly better than software. Stuff that is so utterly non-standard that it needs its own unique init sequence would be a problem, but since I'd like to see standards violators deported to the far side of the moon, there are certain advantages in having a system that boots rationally-designed hardware the best.)
A better design would be to have networked data in well-defined regions which the card can DMA directly out of and into. The PCI bus can handle a 4K transfer as an atomic operation on a single channel, so a 4 channel PCI card can simultaneously send and receive streams of jumbo packets without requiring any CPU intervention. The driver would merely need to be passed two lists of physical page pointers - one for inputs, one for outputs - for each of the open connections and to pass back signals from the board that an entire packet and/or message had been uploaded for a given connection.
(Jumbo packets can go up to 8K, and you can do 8K over two PCI channels, so that's a good unit to be working with.)
The next improvement that can be made to such a system is to improve the buffering. The network will be slower than the computer, almost always, so being able to queue up multiple packets for sending is a Good Thing. Filtering out packets that have not been sent but are no longer worth sending can be done entirely in parallel and does not require anything extra at the end of the pipeline. Not all packets are of equal value, though. You want to deliver the information in the order that will give the best possible benefit - which may not be the order in which the program generates the traffic. Hierarchical Fair Service Curve, Class Based Queueing, and a bunch of other similar techniques, have been developed to fix exactly that sort of problem.
If the ISPs would stop being so bloody stupid, you could also enable protocols such as your basic multicasting (for your UDP stuff) and Scalable Reliable Multicast (for the stuff that needs to reliably get through). That hacks, slashes, butchers and roasts (with just a hint of parsley) problems associated with sending identical state information to multiple end-points.
Bear in mind that Myrinet, Dolphinics, and a bunch of other vendors, use essentially the above mechanisms already and are achieving latencies in the region of 2.5 - 3 microseconds. I say essentially, because I'm not convinced they've optimized quite to the degree I'm suggesting - reliable, scalable multicast RDMA isn't something you'll see a lot of even at a supercomputer fair. True, you're not getting that kind of latency over the Internet whatever you do, but if you can achieve a hard real-time guarantee of 3 microsecond delivery in a LAN party, you WILL notice a difference. At the very least, in the door price, which will now be expressed in exponential notation to fit on the door.
The good news is that since the irrational have created most of the world's problems, and are a primary component of the planetary overpopulation, the problem should eventually correct itself. It generally has, in the past. Mind you, it usually took with it most of civilization, but Slashdot keeps archives so we can recover fairly quickly.
Oversight has several possible definitions, one of which the current administration is extremely good at.
Yeah, that would definitely work. In fact, because there's no distributed locking mechanism (that I know of) in the brain or any packet labeling, a race condition can occur not just between any two processes but also between any two (or more) neurons and between any two (or more) neural pathways. You can also get something similar to a packet collision when backwash from a previous signal corrupts a subsequent one. And as the topology is a mesh, you can get cyclic and self-referential inputs, which means that a race condition need not occur on a specific iteration, only on some iteration.
Blake's 7 got that one right, in that you almost never got the "World War I Dogfight" space combat. Mind you, how much of that was due to skill and how much was due to a budget you could barely feed a canary with is up for debate. The same goes for the few flickers of space combat in The Tomorrow People and most of what you get in Doctor Who in that regard. Most of the other British telefantasy series I can think of that involved space (such as the Quatermas series) avoided the issue entirely.
A false positive is bad, especially if there are far too many, but a false negative can be lethal. This would be more true, say, 100,000 years ago than today, and that's when most of these mechanisms became as finely tuned as they are. Back in the days when hominids were trudging through deadly terrain, you had to remember places and situations that were Bad News with enough time to get clear. In those days, there was a shortage of humvees, so having time to get clear meant having extremely early warning. From that, Deja Vu is a very obvious, direct consequence. In fact, no matter how good humans may have been at avoiding such situations, Deja Vu would always be selected for far more often than against.
(The above can be translated by crypto geeks as follows: The brain has a really crappy but very very fast hashing algorithm used to label sensory data. It's so fast that being crappy doesn't hurt survival chances, but it's crappy enough that we are seeing a very large number of hashing collisions.)
Now, here is where it gets fun. The senses are all cross-linked and cross-referenced in the brain. When the barriers in the brain don't work as expected, we get synaesthesia. Now, it is not at all obvious where the comparison is made, or how the barriers work. For this reason, it is entirely possible to imagine a situation where data from sense A is compared with a prior input from sense B. All it would take is for the barrier to fail to work correctly for recalled data, even if it worked just fine otherwise. This is not "classic" Deja Vu, because the brain is not incorrectly matching an experience with a prior experience of the same sense - it is incorrectly matching totally different types of data. Is this possible? Depends. Any connection that is bi-directional in the brain by nature can fail to mask or block data in either direction, so I can see absolutely no reason why - given synaesthetes are proof that the failure can occur one way - it cannot fail on recall.
(There are soooo many brain disorders associated with inexplicable associations, spooky feelings and false associations that you could fund half the field of neurology for the next fifty years just looking at sensory mismatches and nothing else. Given that, I'd call it almost a flat-out certainty that some of these experiences are cross-sensory errors that involve some of the same matching failures as Deja Vu.)
eg: The RIAA have tried to sue people who have never owned a computer or an Internet account for file swapping. Firstly, this would appear to be trivially-resolvable by any competent arbitrator, it most definitely does not need to be demanding vast amounts of time from an already-overloaded court system. Secondly, it is the understanding of us non-lawyers that the worst the RIAA can get for wasting the time and money of the legal system is a rap on the knuckles for a frivolous lawsuit - the defendant is most unlikely to be reimbursed for time and costs involved - which directly implies that it is cheaper to sue first and ask questions later.
Because the rewards are perceived to be high (whether they are in practice or not) and the risks are perceived to be low (ditto), the courts appear to have become the first resort, not the last resort. No matter how unjustified such a perception may be in reality, it is nonetheless the perception that has arisen and that is seriously damaging to the credibility of the system as a whole.
Personally, I would like to see the courts have greater power to call bull - whether by the plaintiff or the defendant - and greater flexibility in the handling of what can only be called abuse of court. That should include the ability to impose fines or jail time on plaintiffs (or defendants) even outside of the frivolous lawsuit mechanism or the final verdict. There may also be problems with the public defender system, as they have developed rather a bad reputation over the years. If the courts need to supervise such people, then they should be given the power to do so.
Does this impinge on a person's right to a trial? No. I'm not saying anything about denying a person a right to a trial, but rather that such a right does NOT imply a right to a trial first, OR a right to use the mere act of having a trial as a means of inflicting punishment on a person if that person is innocent, and certainly does not imply a right to use the courts for entertainment or get-rich-quick purposes.
(That second one is tough. Time is money. Even if all other expenses are either taken care of or reimbursed afterwards, if a person is in court and is not on the court's payroll, then they are not at work. For low-income individuals or individuals who don't have much of a buffer for whatever reason, this can make it impossible for that person to argue their case meaningfully or - in some cases - at all. I don't know how you can easily close that loophole, but this is essentially a denial-of-service attack, and the courts should never tolerate being used as a weapon. They are there to judge on matters of law, they are not there as a cheap alternative to hiring a hitman.)
No, the matches were all positive, but they were all in Reed's matchbox. The dangerous thing with unspecifics is that it's not always obvious as to what the unspecifics are unspecific about.
The mathematicians invented nullity (see later slashdot article on dividing by zero), so the physicists now have to one-up them on nothingness. I'm expecting any day now to hear NASA announce the discovery of planets that don't actually exist.
In the case of readline, that would be an upstream link and I could see potential licensing issues there, as you are essentially including GPL code in a non-GPL object. That would definitely be on the Forbidden List. Downstream is slightly different - you can run GCC under non-GPL'ed OS', even though there must be links GCC must use that are not GPL'ed. (It is possible, I suppose, that Cygwin re-implements the BIOS, has its own screen manager so that X will work, etc, but me thinks not.
In this case, a better example might be a use of dlopen(). If person A wrote some code that installed a file of a specific name and called specific functions within that file, with ALL of that interface under the GPL, then if some such file happened to not be GPL, I don't see that you would be retrospectively violating the license. The program has not been changed - on disk or in memory. Everything is exactly as it was, with the sole difference that the pointers now point to something, where that something is wholly external and wholly black-box.
(If you were to ask me if I like closed source - whether as a module or in any other form - I'd say no. Corporations HAVE to compile to the lowest common denominator, which means I can always optimize better than them. Corporations CANNOT include capabilities as fast as the total IT market is capable of creating them, which means that I am better equipt to ensure I have the feature set I need. Corporations also have to make assumptions that may - or may not - apply either the typical user or the stereotypically-dumb user, so I am in an infinitely superior position to have code that functions for me, operates the way I think, follows my mental picture of the system in question. Closed-source, by its very nature, has to be a compromise hack. It can't be anything else. Open source often is a compromise hack, but that is entirely by choice, as stupid as I think such a choice is.)
This would be modular if (and only if) you could remove said link from the code and have it still work. I think the word WinDriver is appropriate here. Microsoft has, in the past, found ways to shift functionality around to break things when not doing things their way, even though "technically" they are not doing so. The hardware in a WinPrinter or WinModem doesn't change when you move it to Linux, it still functions entirely within spec, it's not its fault that Linux lacks the necessary extra code.
Alternatively, Microsoft could overload one of the Open Office functions in a way that makes Open Office run better (or appear to) with the module than without. Or they could make it flakier to use Open Document. There's a million ways they could coerce users into using their module. And, as with the browser wars, all they need is to make themselves appear needed.
Now, will this happen? I'm not sure. Novell seem suspicious of Microsoft, but the test of a trap is not whether you are suspicious of it, but whether you are caught. (Kerr Avon, "Bounty", Blake's 7) It also seems odd that - at a time the community is suspicious of the whole relationship - Novell would be doing this. It seems unhelpful for customer relationships (or anything else) to add fuel to the fire, no matter how innocent the whole thing is. There have simply been too many cases of innocent victims (users and businesses) in the past for people to simply relax. One should not be too relaxed around a vampire, even if they claim to have become vegetarian. (Vegetarian vampire ducks excluded.)
Is this a fork? I don't think it matters what it is - if it's safe, then it's helpful. If it's unsafe, it'll be lethal. The name on the bottle really doesn't count for much.
Weather stuff and other public information need to be protected only insofar as they should be read-only. You do NOT want a spotty 16 yr old converting their porn collection into the correct format for the weather database. Well, this IS Slashdot, so maybe I can't be quite so sure about that... Anyways, set up correctly, the system would use mandatory access controls (SELinux would work fine, though they probably have a spare copy of Trusted Irix somewhere) or a read-only network file system for the data with the actual writing done on a secure machine.
This stuff really isn't rocket science...
Seriously, the press has been full of "counter-claims" from all kinds of nuts, and the only known examples of censorship have all been from the anti-Global-Warming lobby in attempts to stifle NASA scientists, et al. (On that basis alone, one might argue that if the anti lobby was being stifled, it's merely the bad karma they themselves polluted the environment with. If you don't want others to do unto you, it helps to not try to stiff them first.)
Oh. You want your search results the same year. Forget it, then. Language analysis on modern computers is way too primitive to give you useful, bomb-proof results over indeterminantly-overlapping web pages where those pages may overlap by different amounts depending on the exact relationship being examined at the time, especially if multi-generational relationships are considered.
You are simply not going to get useful results this side of 2100, unless you are either willing to put up with artificial stupidity (the polar opposite of artificial intelligence) or really bad latency. Artificial stupidity is, sadly, how most search engines operate - usually on the very persuasive grounds that users don't expect anything useful and so don't mind trawling through a billion pages of adverts and fake links in the hope of picking up a nugget or two. Hell, with the click-through model, it might even be more profitable to produce endless junk and a bunch of adverts than to produce something of quality. I could easily see that. Users faced with what they want would have no incentive to give up and look at the banners instead.
I think you've already said that, though.