What you want is a matrix. Not the matrix, just a matrix. Assuming a standard registry (similar to IANA) that documents what standard number a given feature should be assigned and what standard number a given manufacturer's reference implementation should be assigned (the formal specification always having a value of 0):
One axis would denote each large scale feature, another would denote each sub-feature, another would denote the major versions of the API and yet another would denote all variants of the API. The node at the intersection of these would define the start and end minor and trivial version numbers supported.
This would be an extremely sparse array, so you'd probably want to use a slightly different structure internally, but this would provide a completely agnostic way of accessing a list of all supported versions of all supported APIs.
An alternative would be to have each and every function self-announcing, such that any external program can query what the exposed API is and be given back a complete list of calls (with parameters), return types and some vendor-neutral representation of the capabilities and limitations of the call. The problem with this approach is that it's a lot more complex to maintain and a lot slower to utilize. Having a static object which can be examined that provides all that information would have enormous advantages. Further, said static object could be on some server tree, not too far removed from how DNS works. That way, the program itself doesn't need to maintain all that information.
It's the Eleventh Commandment, which never got written down again after the original tablets were smashed. "Thou Shalt Not Covet Thy Airline's Ill-Gotten Gains".
One of the reasons my posts tend to be long (and slow to type) is that I try to keep the communication as clear as I can, within the limits of my ability. (And my former English teachers would tell you exactly how limited they thought that ability to be.) A major reason I stick with Slashdot is that even if I am kept around as the modern version of the court jester, I am tolerated and can even persuade myself that people appreciate my views here. That simply is not, and never has been, the case anywhere else. Sad? Yes. Pathetic? Definitely. Doable? Well the good stories make it tolerable and the occasional truly insightful discussion gives me a much-needed boost of feeling like I'm not some alien abandoned on a planet filled with idiots.
Because there are at least five highly intelligent people here, and the several hundred who have listed me as friend must have some willingness to listen to unorthodox opinions that almost certainly differ from their own.
Now, admittedly, the readership of Slashdot is closer to 100,000, making the percentage of potentially civilized people rather small in comparison.
However, given that you could only half-fill a remote Alpine village with all the truly civilized, courteous yet self-respecting, intelligent people on the planet, in absolute terms roughly 10% of them must be regulars on Slashdot.
It's actually a plurality of different mini-nations with a fairly diverse range of languages and genetic groups, from what I understand. The idea that Japan (or indeed China) is a single anything is as absurd as claiming the United States is a single entity. It's a convenient way to present the collective to the outside world, and politicians sometimes get this misguided notion that they should impose this fiction on the population, but other than that it has no basis in reality.
To be honest, I don't know the exact reason Forth is good for hardware. I would imagine that the nature of a stack is part of it - you can go from the current state to any prior state by reversing the operations performed, guaranteeing that the system is always in a valid state. That leads on to the next benefit - you can do state machines in Forth with relative ease. C is not very good for that kind of work precisely because it's very low-level. When you want to do logic that involves applying one of a fixed number of transforms to transition a system from one state to another state, where the list of transforms is state-dependent, you want very high-level logic.
Obviously, such transformations can be concatenated. A-to-B-to-C is the same as A-to-B then B-to-C. You don't want to permanently store the intermediate result, as you will never use it again. So being able to pipe the output from one operation into another directly simplifies things a lot.
Lastly, for things like hardware initialization or controlling a robot's stepper motor, you don't have extensive logic and you usually need high reliability over and above high performance. C gives you speed, but if your average self-steering robot is crawling along at walking pace or slower, you're ending up with a few billion wait states and nothing to show for it. On the other hand, it is extremely easy to make errors in C that can be catastrophic. It's why the C standards body is drawing up documents on how to do C right, but it is also why you almost never see C used in a mission-critical system. It's almost always a language that sacrifices performance for reliability. It is also why, as BIOS developers moved from Forth to C, reliability has dropped.
The military have never released sample compilers for their highly reliable languages and other languages intended for such purpose have traditionally been horribly expensive. As a result, they are under-evolved, under-utilized and under threat. Nobody wants to bother with them. The delusion that you can use 60s mainframe logic ("all the coders belong to billionaire businesses and those businesses shall kneel before us") in 2010 is beyond belief. Well, almost beyond belief. Did you know that a commercial PL/1 compiler for Linux costs $15,000 per user?
Anyways, it is this sheer delinquency and money-grubbing that has traditionally pushed hardware engineers in the direction of what is cheap, usable and practical. This has meant using Forth for the control systems and Fortran for the number crunching.
There is apparently some evidence of a difference in brain structure that shows up on MRI and where there is a difference in brain response showing up on fMRI. One of the problems with this subject is that it would be almost impossible to come up with a test that is both convincing and legal. The other problem is that although high-resolution MRI scanners exist (9.4 T, as opposed to the usual medical 2.5 T), they are unimaginably rare - only three such high-res scanners exist - and they exceed FDA limits on magnetic radiation exposure. There are about 25-30 other ways to scan brain activity, and it is arguable that to produce definitive results you'd need to carry out a meaningful number of tests using each method.
Even then, it is unclear whether you'd be able to absolutely tell the difference between a conscious choice, a subconscious choice or an imposition due to a neurological or biochemical quirk.
But let us say that there were examples of all three, or even examples in which all three variables were present to some degree. What do you plan to do about it? America has adopted the philosophy of standardizing the punishment according to written law. It has no capacity to handle multi-variable situations and very little capacity for handling anything other than a pure binary guilty/not guilty. (In cases where the statute has an automatic fixed penalty - and there are some - then it is pure binary.)
I believe the question of "why" is perhaps the most important - and least-asked - question the legal system could concern itself with. I do not advocate treating people with (or without) leniency according to cause, but I believe the magnitude and not the form of punishment should be what is fixed. I also do not think a fixed form is necessarily useful - stage it so that there is a clear distinction between punishment and therapy/rehabilitation. In cases where therapy/rehabilitation is impossible at the present time, treat as best you can. Further, in no case should punishment be permitted to worsen the mental state of the person, as that risks aggravating whatever caused them to offend in the first place - at best doing nothing of any good or value, and at worst punishing society for the criminal's acts.
There will always be criminals and there will always be people who merely enjoy crimes for no reason other than that they do. But it comes down to divide and conquer. If we can treat that component which is treatable (even in those who otherwise willingly broke the law), you automatically reduce the problem in future. If we can identify that component which is untreatable using existing methods, we know more about where we need to improve. If we can identify unambiguously those who lack the capacity to choose between "right" and "wrong" (whether or not they can distinguish between them), then we can more reliably segregate such cases and not rely on people paid to hold specific opinions. (What possible value is a test for distinguishing between right and wrong? It makes those who intuitively respond correctly but cannot say why Criminally Insane, and makes those who know the difference but cannot act on that knowledge Evil Vermin. What kind of logic is that?)
Forth is extremely good for hardware control. It seems to have lost the edge it used to have, but once upon a time Forth was dominant in radio astronomy, self-contained robots, firmware, scientific lab equipment, etc.
This is the age of multi-core! No longer are you limited to single-thread inductions. With the appropriate locks and barrier operations, 16-way parallelism should be possible.
Ian Bell (Elite), David Braben (Elite, Zarch, Virus, Virus 2000), Jeremy Smith (Thrust, Exile), Peter Irvin (Starship Command, Exile), Nigel Alderton (Chuckie Egg), Geoff Crammond (Revs, Formula One Grand Prix), Tim and Chris Stamper (Knight Lore), Nick Pelling (Firetrack, Frak!, Wing Commander, Mortal Kombat 2, Duke Nukem), Peter Molyneux (Populous). And that's within a few minutes of looking. Of these, Ian Bell and David Braben deserve MBEs for their service to gaming and a lifetime of free psychiatric care to help with their feud).
Actually, the best CEOs are psychopaths and sociopaths. Asperger's makes for awful CEOs. And it is arguments like this that make me feel like psychology should be kicked to the door until it uses causal diagnoses rather than symptoms.
The British police have a unit specifically set up to handle such material. Officers are monitored by psychologists and are required to leave that unit when either the psychologist determines they've hit their limits or a certain maximum exposure has been reached.
I'm not thrilled with the approach, but it seems more rational and better thought-out than to throw college kids in front of the screens until their brains explode.
A benchmark that flooded the cache for the first core would need to reload the cache for the second, third, and so on. The total improvement, owing to the non-shared L1 cache, would likely have been bettered had the real-estate used in the extra cores been used in extra L1 cache space instead.
DNSSEC doesn't need to be complex at all. Basically, any secure communications system of this kind must have a mechanism for authenticating whatever provides the service, authenticating (if/as necessary) the recipients of that service, both encrypting and tamper-proofing both the request and the result, and (where necessary) limiting queries to those authorized for that recipient.
There are plenty of authentication mechanisms out there (Kerberos, SASL, SSL, TLS, S/Key) - some for the server, some for the client, some for both. None of them are rocket-science to set up. Thus, it can be definitively said that any secure channel requiring a more complex setup than the worst of the standard mechanisms out there is far more complex than necessary.
Encryption is definitely a non-issue. For the initial asymmetric key exchange, use one of the standard algorithms. If you insist on avoiding writing code, make a call to IPSEC's IKEv2 facility. Ok, what then? This is a rapid-fire system so there is probably no need to go to the lengths of producing a symmetric key then exchanging that using the asymmetric one. Just use the appropriate asymmetric encryption algorithm to encrypt the request and then to encrypt the response. To provide tamper-proofing, since asymmetric encryption algorithms are generally block encryption, use OMAC for the encryption mode. No configuration is required for any of this. Where DNS-to-DNS tunnels need to be maintained, packets should be encrypted using a regular block cypher with the keys exchanged via the asymmetric mechanism. To provide the switch between the two modes (sporadic or tunnel), you need exactly one flag. Not hard. And the security implications of an incorrect setting are nil. It'll pessimize performance a little but that's it.
This leaves restricting responses. If the original request includes mandatory access control information then that requestor gets access to all public records plus all records in all roles the requestor's access control information authorizes. If there is no MAC information, the requestor only has access to public records. The only servers this would have any meaning for would be those which sit on the boundary of two networks where the servers need to know about both networks AND allow limited visibility from one into the other. In short, practically none. Almost nobody for whom it matters sets up DNS that way. Besides which, hiding the name doesn't hide the server. Hiding the name only stops you from trivially discovering a few things about the topology on the other side of the wall. If you want the information hidden that bad, put it on a different server and not one in plain sight. However, on the off-chance such a facility is necessary, let's consider what it would take. Each non-public record set and each non-public record would need to specify which combinations of permissions would be necessary to access that information. Each requestor would need to be registered on that server with what permissions they had. The server would check those permissions against the record set and then against the record being accessed. 998 times out of the entire 1000 cases worldwide this could possibly ever happen in, all authorized users can access all records. In which case, if the user is registered, they have access. So your setup becomes a username/password table. That's it. The system one should do fine. In the 999th case, a simple bitmask should suffice. For the 1000th case, you would want to have the thread that handled the request chown() itself to the requestor's account, then use the built-in access control system for the OS and/or the database to handle record set and record access controls.
Even in the very very worst, unimaginably horrific scenario imaginable, where you have to have something with each and every one of these features to the limit, configuration of a bi-directionally authenticated connection between two servers should involve a couple of certificates, a couple of digital signatures, the setting of a single bit field on each
Arthur C Clarke's writings on solar sails usually (not always, but usually) didn't involve photonic pressure but rather involved the charged particles of the solar wind. Much more energy is involved and you can operate further from the sun. (Almost no photonic energy by the time you reach Jupiter - the sun is barely distinguishable from any other star at that distance - but the solar wind remains significant until you reach the heliopause.)
No, no, the aliens were just in for their routine medical and the X-Ray scanner went wonky. Nothing to get alarmed about. They were well within the permitted dosage limits, according to medical experts on Omnichron 425.
is the product or service that I desire worth more than that which I would have to give up to obtain it
That is not a useful - or even usable - method of understanding choice because it gives a false choice equal weight to any true choice. It also assumes that A and B are never synonymous. If I give you an apple and you give me back an identical apple (or even the same one), then I have not made any choice that was different from never having given the apple at all. In such a case, I have not chosen to "obtain" the apple over and above "keeping" it because the two are functionally identical. They were the same choice.
(Likewise, if I take an electron off an atom and replace it with another electron, the system will be in an identical state to how it started. It isn't merely similar, it is absolutely the same and no test could ever show that anything had ever happened. It would make absolutely no bloody sense to a physicist if you were to tell them that you chose which electrons were present. The statement has no meaning. It may not be "common sense", but virtually anything in life that is "common sense" is probably wrong.)
I would agree with you on that. No point burning money on Orion at this point. Actually, since the Russian Shuttle seems to be a superior design to the American one, it might not be such a bad idea if the US and Russia got together to do an analysis on whether it would be viable to reuse concepts from it or use it as a starting point in a manned space program.
For commercial space flight, I'd suggest an overhaul of regulations and a sensible international body to monitor it. Mis-communication nearly caused Russia to respond to a civilian rocket launch of, IIRC, a weather satellite with a full salvo of ICBMs after mistaking it for an attack. The last thing we need is for (very normal) corporate mishaps to trigger World War 3. Corporations haven't, as a rule, ever been in a position where that could easily be the outcome of office politics and have no mechanisms in place for dealing with such emergencies. As incompetent as governments are in emergencies at anything practical, they are extremely good at defusing tensions when necessary.
However, I don't think it should stop there. Almost no industry in history has ever successfully gone from being essentially a research program to being a major commercial program without going through a phase at some point where garage developers were able to contribute meaningfully. There are some that have (homebrew has done little for medical electronics or medical equipment in general), but they are exceptions and not the rule. The modern car is largely a product of early automobile engineers experimenting and then racing the results. The modern computer owes more to the surge in inventors playing with the microprocessor in the 70s and then trying to clone the IBM PC in the 1980s than it owes to the Honeywells or DECs of the previous era. (Yes, Ferranti, Honeywell, DEC and IBM were commercial, but they were largely selling to governments. What's more, when they did sell to business it was in a very limited role. The market didn't explode until after the garage era.)
But how is that going to happen? Amateur rockets, even the serious ones, are nothing compared to the engines you need to lift anything significant. This isn't just a matter of scale. Well, in a way it is. There are huge problems to overcome as the nozzle increases in size - one reason the Russians always opted for clusters of smaller rockets. It's technically easier, even though there's efficiency penalties to pay for that.
There are, however, problems that really could use a solution where amateur rocket enthusiasts could probably come up with valuable ideas and innovative solutions in a way neither government nor commerce ever could.
1) Rockets have very low efficiency in a thick atmosphere - one reason NASA has examined everything from turbine-assisted ramjets to ski-jumps as a way to eliminate the initial stage as a rocket. NASA simply hasn't been able to solve that problem, but it is plausible to say that this is because they've had their attention on too many targets and haven't had the resources to solve everything. No existing vehicle could use assisted launches, so it's a solution they could not have used any time soon even if they'd found one.
2) For much the same reason, NASA abandoned in around 1999-2000 its new blended-wing space-plane which would have had the performance of Concorde, twice the capacity of the Airbus 400 and the ability to reach true Low Earth Orbit. (It would have made White Knight look like a wet sardine.) Even if not used for civilian flights it would have been at worst an amazing launch vehicle in the same way White Knight is. A little more oomph and you'd have had a horizontal take-off and landing replacement for the Shuttle. (A modernized HOTOL, basically - a project not only abandoned in the Thatcher era but classified as well.) In consequence of all this abandonment and classification, the sum total of available knowledge in the civilian world is not very much. Although I doubt any private group could hope to build a true space plane, it does mean private groups can obtain knowledge and information about how blended-wing bodies and waveriders work in practice that is not readily available and would be considered far too blue-sky for commerce to go near.
Slashdot Mirror
My guess would be all permutations of "trusted", "Dell" and "technician". There are probably a few other suspect word pairings in there as well.
What you want is a matrix. Not the matrix, just a matrix. Assuming a standard registry (similar to IANA) that documents what standard number a given feature should be assigned and what standard number a given manufacturer's reference implementation should be assigned (the formal specification always having a value of 0):
One axis would denote each large scale feature, another would denote each sub-feature, another would denote the major versions of the API and yet another would denote all variants of the API. The node at the intersection of these would define the start and end minor and trivial version numbers supported.
This would be an extremely sparse array, so you'd probably want to use a slightly different structure internally, but this would provide a completely agnostic way of accessing a list of all supported versions of all supported APIs.
An alternative would be to have each and every function self-announcing, such that any external program can query what the exposed API is and be given back a complete list of calls (with parameters), return types and some vendor-neutral representation of the capabilities and limitations of the call. The problem with this approach is that it's a lot more complex to maintain and a lot slower to utilize. Having a static object which can be examined that provides all that information would have enormous advantages. Further, said static object could be on some server tree, not too far removed from how DNS works. That way, the program itself doesn't need to maintain all that information.
It's the Eleventh Commandment, which never got written down again after the original tablets were smashed. "Thou Shalt Not Covet Thy Airline's Ill-Gotten Gains".
What would they need that for? God can see through clothes - and informs me that CowboyNeil is wearing lace.
One of the reasons my posts tend to be long (and slow to type) is that I try to keep the communication as clear as I can, within the limits of my ability. (And my former English teachers would tell you exactly how limited they thought that ability to be.) A major reason I stick with Slashdot is that even if I am kept around as the modern version of the court jester, I am tolerated and can even persuade myself that people appreciate my views here. That simply is not, and never has been, the case anywhere else. Sad? Yes. Pathetic? Definitely. Doable? Well the good stories make it tolerable and the occasional truly insightful discussion gives me a much-needed boost of feeling like I'm not some alien abandoned on a planet filled with idiots.
Because there are at least five highly intelligent people here, and the several hundred who have listed me as friend must have some willingness to listen to unorthodox opinions that almost certainly differ from their own.
Now, admittedly, the readership of Slashdot is closer to 100,000, making the percentage of potentially civilized people rather small in comparison.
However, given that you could only half-fill a remote Alpine village with all the truly civilized, courteous yet self-respecting, intelligent people on the planet, in absolute terms roughly 10% of them must be regulars on Slashdot.
It's actually a plurality of different mini-nations with a fairly diverse range of languages and genetic groups, from what I understand. The idea that Japan (or indeed China) is a single anything is as absurd as claiming the United States is a single entity. It's a convenient way to present the collective to the outside world, and politicians sometimes get this misguided notion that they should impose this fiction on the population, but other than that it has no basis in reality.
To be honest, I don't know the exact reason Forth is good for hardware. I would imagine that the nature of a stack is part of it - you can go from the current state to any prior state by reversing the operations performed, guaranteeing that the system is always in a valid state. That leads on to the next benefit - you can do state machines in Forth with relative ease. C is not very good for that kind of work precisely because it's very low-level. When you want to do logic that involves applying one of a fixed number of transforms to transition a system from one state to another state, where the list of transforms is state-dependent, you want very high-level logic.
Obviously, such transformations can be concatenated. A-to-B-to-C is the same as A-to-B then B-to-C. You don't want to permanently store the intermediate result, as you will never use it again. So being able to pipe the output from one operation into another directly simplifies things a lot.
Lastly, for things like hardware initialization or controlling a robot's stepper motor, you don't have extensive logic and you usually need high reliability over and above high performance. C gives you speed, but if your average self-steering robot is crawling along at walking pace or slower, you're ending up with a few billion wait states and nothing to show for it. On the other hand, it is extremely easy to make errors in C that can be catastrophic. It's why the C standards body is drawing up documents on how to do C right, but it is also why you almost never see C used in a mission-critical system. It's almost always a language that sacrifices performance for reliability. It is also why, as BIOS developers moved from Forth to C, reliability has dropped.
The military have never released sample compilers for their highly reliable languages and other languages intended for such purpose have traditionally been horribly expensive. As a result, they are under-evolved, under-utilized and under threat. Nobody wants to bother with them. The delusion that you can use 60s mainframe logic ("all the coders belong to billionaire businesses and those businesses shall kneel before us") in 2010 is beyond belief. Well, almost beyond belief. Did you know that a commercial PL/1 compiler for Linux costs $15,000 per user?
Anyways, it is this sheer delinquency and money-grubbing that has traditionally pushed hardware engineers in the direction of what is cheap, usable and practical. This has meant using Forth for the control systems and Fortran for the number crunching.
There is apparently some evidence of a difference in brain structure that shows up on MRI and where there is a difference in brain response showing up on fMRI. One of the problems with this subject is that it would be almost impossible to come up with a test that is both convincing and legal. The other problem is that although high-resolution MRI scanners exist (9.4 T, as opposed to the usual medical 2.5 T), they are unimaginably rare - only three such high-res scanners exist - and they exceed FDA limits on magnetic radiation exposure. There are about 25-30 other ways to scan brain activity, and it is arguable that to produce definitive results you'd need to carry out a meaningful number of tests using each method.
Even then, it is unclear whether you'd be able to absolutely tell the difference between a conscious choice, a subconscious choice or an imposition due to a neurological or biochemical quirk.
But let us say that there were examples of all three, or even examples in which all three variables were present to some degree. What do you plan to do about it? America has adopted the philosophy of standardizing the punishment according to written law. It has no capacity to handle multi-variable situations and very little capacity for handling anything other than a pure binary guilty/not guilty. (In cases where the statute has an automatic fixed penalty - and there are some - then it is pure binary.)
I believe the question of "why" is perhaps the most important - and least-asked - question the legal system could concern itself with. I do not advocate treating people with (or without) leniency according to cause, but I believe the magnitude and not the form of punishment should be what is fixed. I also do not think a fixed form is necessarily useful - stage it so that there is a clear distinction between punishment and therapy/rehabilitation. In cases where therapy/rehabilitation is impossible at the present time, treat as best you can. Further, in no case should punishment be permitted to worsen the mental state of the person, as that risks aggravating whatever caused them to offend in the first place - at best doing nothing of any good or value, and at worst punishing society for the criminal's acts.
There will always be criminals and there will always be people who merely enjoy crimes for no reason other than that they do. But it comes down to divide and conquer. If we can treat that component which is treatable (even in those who otherwise willingly broke the law), you automatically reduce the problem in future. If we can identify that component which is untreatable using existing methods, we know more about where we need to improve. If we can identify unambiguously those who lack the capacity to choose between "right" and "wrong" (whether or not they can distinguish between them), then we can more reliably segregate such cases and not rely on people paid to hold specific opinions. (What possible value is a test for distinguishing between right and wrong? It makes those who intuitively respond correctly but cannot say why Criminally Insane, and makes those who know the difference but cannot act on that knowledge Evil Vermin. What kind of logic is that?)
Forth is extremely good for hardware control. It seems to have lost the edge it used to have, but once upon a time Forth was dominant in radio astronomy, self-contained robots, firmware, scientific lab equipment, etc.
What about the Pike scripting language?
This is the age of multi-core! No longer are you limited to single-thread inductions. With the appropriate locks and barrier operations, 16-way parallelism should be possible.
Ian Bell (Elite), David Braben (Elite, Zarch, Virus, Virus 2000), Jeremy Smith (Thrust, Exile), Peter Irvin (Starship Command, Exile), Nigel Alderton (Chuckie Egg), Geoff Crammond (Revs, Formula One Grand Prix), Tim and Chris Stamper (Knight Lore), Nick Pelling (Firetrack, Frak!, Wing Commander, Mortal Kombat 2, Duke Nukem), Peter Molyneux (Populous). And that's within a few minutes of looking. Of these, Ian Bell and David Braben deserve MBEs for their service to gaming and a lifetime of free psychiatric care to help with their feud).
Actually, the best CEOs are psychopaths and sociopaths. Asperger's makes for awful CEOs. And it is arguments like this that make me feel like psychology should be kicked to the door until it uses causal diagnoses rather than symptoms.
The British police have a unit specifically set up to handle such material. Officers are monitored by psychologists and are required to leave that unit when either the psychologist determines they've hit their limits or a certain maximum exposure has been reached.
I'm not thrilled with the approach, but it seems more rational and better thought-out than to throw college kids in front of the screens until their brains explode.
A benchmark that flooded the cache for the first core would need to reload the cache for the second, third, and so on. The total improvement, owing to the non-shared L1 cache, would likely have been bettered had the real-estate used in the extra cores been used in extra L1 cache space instead.
See: Limit - Amdahl
DNSSEC doesn't need to be complex at all. Basically, any secure communications system of this kind must have a mechanism for authenticating whatever provides the service, authenticating (if/as necessary) the recipients of that service, both encrypting and tamper-proofing both the request and the result, and (where necessary) limiting queries to those authorized for that recipient.
There are plenty of authentication mechanisms out there (Kerberos, SASL, SSL, TLS, S/Key) - some for the server, some for the client, some for both. None of them are rocket-science to set up. Thus, it can be definitively said that any secure channel requiring a more complex setup than the worst of the standard mechanisms out there is far more complex than necessary.
Encryption is definitely a non-issue. For the initial asymmetric key exchange, use one of the standard algorithms. If you insist on avoiding writing code, make a call to IPSEC's IKEv2 facility. Ok, what then? This is a rapid-fire system so there is probably no need to go to the lengths of producing a symmetric key then exchanging that using the asymmetric one. Just use the appropriate asymmetric encryption algorithm to encrypt the request and then to encrypt the response. To provide tamper-proofing, since asymmetric encryption algorithms are generally block encryption, use OMAC for the encryption mode. No configuration is required for any of this. Where DNS-to-DNS tunnels need to be maintained, packets should be encrypted using a regular block cypher with the keys exchanged via the asymmetric mechanism. To provide the switch between the two modes (sporadic or tunnel), you need exactly one flag. Not hard. And the security implications of an incorrect setting are nil. It'll pessimize performance a little but that's it.
This leaves restricting responses. If the original request includes mandatory access control information then that requestor gets access to all public records plus all records in all roles the requestor's access control information authorizes. If there is no MAC information, the requestor only has access to public records. The only servers this would have any meaning for would be those which sit on the boundary of two networks where the servers need to know about both networks AND allow limited visibility from one into the other. In short, practically none. Almost nobody for whom it matters sets up DNS that way. Besides which, hiding the name doesn't hide the server. Hiding the name only stops you from trivially discovering a few things about the topology on the other side of the wall. If you want the information hidden that bad, put it on a different server and not one in plain sight. However, on the off-chance such a facility is necessary, let's consider what it would take. Each non-public record set and each non-public record would need to specify which combinations of permissions would be necessary to access that information. Each requestor would need to be registered on that server with what permissions they had. The server would check those permissions against the record set and then against the record being accessed. 998 times out of the entire 1000 cases worldwide this could possibly ever happen in, all authorized users can access all records. In which case, if the user is registered, they have access. So your setup becomes a username/password table. That's it. The system one should do fine. In the 999th case, a simple bitmask should suffice. For the 1000th case, you would want to have the thread that handled the request chown() itself to the requestor's account, then use the built-in access control system for the OS and/or the database to handle record set and record access controls.
Even in the very very worst, unimaginably horrific scenario imaginable, where you have to have something with each and every one of these features to the limit, configuration of a bi-directionally authenticated connection between two servers should involve a couple of certificates, a couple of digital signatures, the setting of a single bit field on each
Arthur C Clarke's writings on solar sails usually (not always, but usually) didn't involve photonic pressure but rather involved the charged particles of the solar wind. Much more energy is involved and you can operate further from the sun. (Almost no photonic energy by the time you reach Jupiter - the sun is barely distinguishable from any other star at that distance - but the solar wind remains significant until you reach the heliopause.)
In Los Angeles it is.
They're wormholing their network traffic.
Hmmmm. Nope, doesn't look like it, but one has either grown an extra arm and head or NASA is replaying HHGTTG again.
No, no, the aliens were just in for their routine medical and the X-Ray scanner went wonky. Nothing to get alarmed about. They were well within the permitted dosage limits, according to medical experts on Omnichron 425.
is the product or service that I desire worth more than that which I would have to give up to obtain it
That is not a useful - or even usable - method of understanding choice because it gives a false choice equal weight to any true choice. It also assumes that A and B are never synonymous. If I give you an apple and you give me back an identical apple (or even the same one), then I have not made any choice that was different from never having given the apple at all. In such a case, I have not chosen to "obtain" the apple over and above "keeping" it because the two are functionally identical. They were the same choice.
(Likewise, if I take an electron off an atom and replace it with another electron, the system will be in an identical state to how it started. It isn't merely similar, it is absolutely the same and no test could ever show that anything had ever happened. It would make absolutely no bloody sense to a physicist if you were to tell them that you chose which electrons were present. The statement has no meaning. It may not be "common sense", but virtually anything in life that is "common sense" is probably wrong.)
I would agree with you on that. No point burning money on Orion at this point. Actually, since the Russian Shuttle seems to be a superior design to the American one, it might not be such a bad idea if the US and Russia got together to do an analysis on whether it would be viable to reuse concepts from it or use it as a starting point in a manned space program.
For commercial space flight, I'd suggest an overhaul of regulations and a sensible international body to monitor it. Mis-communication nearly caused Russia to respond to a civilian rocket launch of, IIRC, a weather satellite with a full salvo of ICBMs after mistaking it for an attack. The last thing we need is for (very normal) corporate mishaps to trigger World War 3. Corporations haven't, as a rule, ever been in a position where that could easily be the outcome of office politics and have no mechanisms in place for dealing with such emergencies. As incompetent as governments are in emergencies at anything practical, they are extremely good at defusing tensions when necessary.
However, I don't think it should stop there. Almost no industry in history has ever successfully gone from being essentially a research program to being a major commercial program without going through a phase at some point where garage developers were able to contribute meaningfully. There are some that have (homebrew has done little for medical electronics or medical equipment in general), but they are exceptions and not the rule. The modern car is largely a product of early automobile engineers experimenting and then racing the results. The modern computer owes more to the surge in inventors playing with the microprocessor in the 70s and then trying to clone the IBM PC in the 1980s than it owes to the Honeywells or DECs of the previous era. (Yes, Ferranti, Honeywell, DEC and IBM were commercial, but they were largely selling to governments. What's more, when they did sell to business it was in a very limited role. The market didn't explode until after the garage era.)
But how is that going to happen? Amateur rockets, even the serious ones, are nothing compared to the engines you need to lift anything significant. This isn't just a matter of scale. Well, in a way it is. There are huge problems to overcome as the nozzle increases in size - one reason the Russians always opted for clusters of smaller rockets. It's technically easier, even though there's efficiency penalties to pay for that.
There are, however, problems that really could use a solution where amateur rocket enthusiasts could probably come up with valuable ideas and innovative solutions in a way neither government nor commerce ever could.
1) Rockets have very low efficiency in a thick atmosphere - one reason NASA has examined everything from turbine-assisted ramjets to ski-jumps as a way to eliminate the initial stage as a rocket. NASA simply hasn't been able to solve that problem, but it is plausible to say that this is because they've had their attention on too many targets and haven't had the resources to solve everything. No existing vehicle could use assisted launches, so it's a solution they could not have used any time soon even if they'd found one.
2) For much the same reason, NASA abandoned in around 1999-2000 its new blended-wing space-plane which would have had the performance of Concorde, twice the capacity of the Airbus 400 and the ability to reach true Low Earth Orbit. (It would have made White Knight look like a wet sardine.) Even if not used for civilian flights it would have been at worst an amazing launch vehicle in the same way White Knight is. A little more oomph and you'd have had a horizontal take-off and landing replacement for the Shuttle. (A modernized HOTOL, basically - a project not only abandoned in the Thatcher era but classified as well.) In consequence of all this abandonment and classification, the sum total of available knowledge in the civilian world is not very much. Although I doubt any private group could hope to build a true space plane, it does mean private groups can obtain knowledge and information about how blended-wing bodies and waveriders work in practice that is not readily available and would be considered far too blue-sky for commerce to go near.