They haven't "done nothing" - they specified an Evil Bit that attackers are required (by spec) to set. If attackers are failing to set the Evil Bit, it isn't the IETF's fault. That's an implementation issue.
I've a better idea. Let's swap one of the glaciers for LA, and then open the casino on top of the Martian ice. It'll be closer, there's less risk of being considered edible delicacies and it'll solve most of the environmental problems in one go.
Well, if they're going to do DNA analysis, there's various indexes of famous dead people's DNA that you can compare yourself to. On the historic side, it adds credibility to contemporary accounts. (You thought they were more honest than today's newspapers?) On the tourism side, local authorities always love being able to nail signs up saying "so-and-so is buried here" (helps them to increase taxes) and local thieves can sell Copernicus-themed junk to gullible visitors. On the science side, it is probably worthless.
And seriously harden up the electronics. If the Pioneer and Voyager probes can do 30+ years, a modern probe can. Given the fuel efficiency of the ion drive, a probe could also carry enough fuel to perform a great many missions. It may not be able to do everything on the list, but a decent design should be able to tick off a fair few at less cost than one probe for each one.
Some grid software is heavy - Globus could do with being ripped into sensible-sized modules - but the notion of a "Grid" is very simple. You have software where components don't care where other components are located. The Service Location Protocol, Zeroconf, RPC w/ Mobile IP - these are all ways of providing lightweight grids. MOSIX over a wide-area network would be a Grid of sorts. None of these are designed by committee. Now, some Grid systems are. The aforementioned Globus being an example. As committee-designed software systems go, it's actually not that bad. It's overweight and inefficient, but there are far far worse examples, and at least it's maintained and some very nice add-ons (like SASL security) exist.
Yes, that's the stuff, but - yuk - it looks very incomplete. It's unclear as to what's there and what isn't, VHDL isn't an easy format to work in, and it'd take a while for even those who know what they're doing to get that into any reasonable shape. I don't even come close to that level of expertise in signals theory.
I've bookmarked it and will start hacking on the code this weekend.
True, but wouldn't that be just as true for any wireless mesh that tries to handle reliable connections over paths that can change pseudo-randomly on-the-fly? DTN is presumably no more expensive on the compute cycles or memory than any of the 200+ methods on Citeseer for doing the same job specifically for terrestrial wireless, but is presumably much more versatile as it was designed to handle far worse situations.
Talking of noisy links, are there any open source implementations of Turbo Codes? (This is an error-correction scheme that is supposed to be designed specifically for errors that are not scattered randomly but concentrated into chunks. Reed-Solomon is better for scattered bit errors, provided the density is low enough. I can find open source Reed-Solomon implementations, but I haven't seen any for Turbo Codes.)
It would be nice if there were packet drivers for the protocol for Linux, NetBSD and NS-3 (the current network simulator). This would allow people to get a good feel for the behaviour of the new protocol, which may have uses beyond deep space. (It is possible to imagine real-world networks on Planet Earth where the characteristics of the network compare with inter-satellite communications.) Besides, would you rather Europe's answer to GPS use Windows?
It is headed by the only Linux nerd who could afford to chase a rating of 6 or above. (7 is the highest the EAL will go.) Another thing to consider is that EAL ratings are only valid for a combination of OS and hardware. So, running Windows on any box (even if functionally identical) to the configuration tested on makes the tests invalid. The true is arguably the same for Linux, except that you can download LTP and gain some measure of assurance (even if not blessed on that platform) that you've not broken any of the security.
The highest old-style NSA rating (A1) is superior to the current EAL6+, and general-purpose OS' did achieve it. Genesis was one (and, no, not the one with the Phil Collins plugin module). EAL6+ looks to me to be about the same as the Orange Book B3 classification, which Trusted Irix achieved. Linux, if LTP was extended enough, could be provisionally ratified up to this level. If it ever was, then I could see vendors like IBM (who got Red Hat certified up to EAL4+) or private millionaires either individually or (more likely) jointly funding the certification.
Of course, EAL-style security isn't everything you need. Security labels on packets would be good - isn't there some work on this already? Support for hardware MAC (mandatory access controls) for memory would be good, as that protects not only against memory access violations in software, but also against such violations with RDMA. (Of course, if the hardware isn't present, you don't get that security, but likewise if the OS support isn't present, you don't get the security.) Better support for hardware encryption - especially within OpenSSL and IPSec - would improve matters too. Coverity is a decent-enough static checker, but their much-vaunted cooperation with Open Source doesn't seem to be producing much in the way of results - I can't remember the last time anyone covered on Slashdot or LWN any work by them. Are the major Linux vendors considering alternatives like Klockwork or any of the theorum provers listed just the other day?
Linux is already very good, but it hasn't received the severe auditing of OpenBSD (although, arguably, Linux does better when it comes to bugs that aren't security holes and also does better on the feature set and hardware supported). Perhaps a round or eleventy of severe auditing would be good for it. There again, perhaps there are other means of being close enough to that level of effectiveness without cutting back on the flexibility and without demanding unreasonable resources.
First, as others have noted, there is a massive level of sheer scientific curiosity. Prior to this, we didn't know of any planet other than Earth that ever had liquid water on it. We had no idea if such planets were rare or common, or even how to identify them if the water wasn't extremely visible and obvious. This allows us to know so much more about planets and their evolution in early solar systems than we ever knew before.
Then, there is another side. Water, particularly if it is mildly acidic, leaves open the possibility of cave systems. Cave systems make manned exploration a more realistic possibility, as you're better shielded from cosmic radiation, much better shielded from dust devils, and have a (comparatively) easy environment to seal and pressurize.
Finally, the combination of a lower gravity and a lower air pressure (whilst a significant atmosphere lasted) may make for crystals that are very different from those that form naturally on Earth. They should be slightly higher purity, for a start. This would not pay for exploration of Mars, or even significantly offset the costs, but it might well intrigue enough of the uber-rich (who tend to like unique trinkets) to either coerce Governments to fund exploration or provide some of the money themselves, purely for the bragging rights of having superior-grade, all-natural, extraterrestrial gemstones.
HOL is said to be usable, but doesn't have the same opportunities for name abuse, though 2001 fans could come up with a whole bunch of new ideas. Isabelle, however, would provide such opportunities, if you were to combine it with the package that started this thread. ACL2 is the least abusable name of all, so is quite useless.
You are correct, but nobody is going to buy a distro for an Ultrasparc IV (unless there's a LOT of support behind it) and there's a limit to how many who download from their site who will then buy their hardware. Sun has done some amazing things for Open Source (such as opening up the source to their microprocessors, although I still think placing the code on Open Cores would be smart, and they're not exactly forthcoming as to how the community has responded), but ultimately their core survival depends on shifting Big Ticket items. You'd have to sell a LOT of Solaris or Linux CDs to compare with the net profit obtained by shipping even a single T2-based server blade.
This does not mean backing away from Open Source. Far from it. Software is expensive to maintain, generates relatively little revenue and Sun's acceptance of Open Source clearly demonstrates they can no longer compete in that market. What Sun really needs to do is switch their software efforts from original development to QA, porting and optimization, and to gear up where they can weather the storm. IBM has done remarkably well by leveraging their skills in processors to produce low-end chips like the PPC, PPC64 and Cell. A server processor might not be viable for a mobile phone, but I'm willing to bet they could produce a version that was easier to write for than a Cell and better at the bus-intensive operations needed by MMORGs and other modern games than the ix86 architecture, which means they should be able to produce something that could compete in the flourishing console markets (which is what IBM is trying to do with the Cell). Since servers are bus-intensive and it's more important to shove data at the GPUs (which can do all the serious maths these days better than CPUs), starting with a CPU that is designed for massive data processing than one designed for massive number crunching would seem logical enough.
I don't seriously believe Sun can survive if they stay so deeply enmeshed in software, as it is only in hardware that they are distinct and only in hardware that they really have any advantage.
I don't know about anyone else's definition, but I would consider distros marking themselves as Phase 2 Certified as one way you could define "pushing IPv6". Another might be to enable IPv6 by default in the kernel (since we're talking about IPv6 capability in the article, rather than usage), or to use IPv6 by default on all connections where it is supported at the kernel and application level, whether or not it is genuinely supported end-to-end. Far as I can see, very few distros are certified for IPv6 (I can't honestly remember seeing the logo anywhere), those that provide it don't take advantage of it, and those that do don't make it easy to take advantage of it (Ubuntu provides some IPv6 tunneling software, but nothing that works with any of the IPv6 gateways I'm using, and network administration using IPv6 under Ubuntu is a pain -- and it's by far the best distro I've used in this regard).
If I could have gotten away with calling 98% of my exams "junk", not understood the remaining 2% of the questions, and passed with grades so good that people would give me the kind of cash genetics labs are getting, I'd have retired at age 30 to my own island.
...I'm not (yet) convinced of the value of the gene-mapping you can currently buy. $1000+ and you get back a description that is essentially meaningless because they don't really understand how the genes work yet. You get tested for a handful of conditions which have genetic links, but not all. (Genetic studies have shown there to be 7 forms of ME, according to the specific genetic cause, but very few labs will test for any of them yet.) Without knowing more about how genes work, it is impossible to know if what these studies reveal is even an accurate reflection of the genetics behind such conditions.
Alongside that is an argument in the reverse direction. If genes are not necessarily contiguous and/or can have ill-defined boundaries and/or can have components off the main DNA itself, then there is a definite possibility that there may be additional regions that could be useful for deep ancestry and genealogical DNA testing. This could help enormously as current research is pushing the limits of what is knowable using the regions and markers that are currently available. Entire haplogroup trees have been redefined because new information has revealed flaws in the previous models. More data, preferably more data that changes slowly, could be useful in getting these models right rather than continuously patching them.
I've used Ada a fair bit, but the biggest joke in that particular line when I used the language tended to be that none of the compilers complied with all aspects of the specification because it was just too complex. Ada is powerful and I do like many aspects of Ada (including the fact that many common programming errors are impossible in the language) but it is... clunky in some respects and the runtime components tend to be heavy. Modula-2 I don't like at all - it has a tighter structure than Pascal but I've not seen any way in which it does so that is actually useful.
When it comes to programming languages, I am not satisfied with any of the languages currently out there and feel they tend to either be over-engineered or over-reliant on evolution to fix a lack of engineering. There are some that seem to have a better balance, but they tend to be obscure and therefore insufficiently exercised to be sure that this appearance reflects reality. If they were better, it would seem they'd be used more.
I dunno - books describing how to analyze problems, produce flow charts, abstract data types and abstract state machines, common patterns, useful algorithms, jackson structured programming, and the like, could be applied to many situations and give a good grounding that could be used to then take a computer course or any other course that is based on the organization of processes. They are also techniques that require not a whole lot more than a pencil and paper to practice.
The same would be true of books covering cellular automata, truth tables, boolean algebra or even petri nets. Now, the middle two might be more interesting to those who later want to go into EE rather than CS, but they're still worth knowing.
Books describing when and how to use algorithms, herustics and fuzzy logic would be fairly basic to read, require nothing to really practice, and would again be good background material for many courses.
What would be useless would be things that are implementation-specific or which require extensive hands-on experience. So, books on programming in C++ or developing device drivers for Linux 2.6 would probably not be worth a whole lot.
Slashdot Mirror
They haven't "done nothing" - they specified an Evil Bit that attackers are required (by spec) to set. If attackers are failing to set the Evil Bit, it isn't the IETF's fault. That's an implementation issue.
I've a better idea. Let's swap one of the glaciers for LA, and then open the casino on top of the Martian ice. It'll be closer, there's less risk of being considered edible delicacies and it'll solve most of the environmental problems in one go.
Well, if they're going to do DNA analysis, there's various indexes of famous dead people's DNA that you can compare yourself to. On the historic side, it adds credibility to contemporary accounts. (You thought they were more honest than today's newspapers?) On the tourism side, local authorities always love being able to nail signs up saying "so-and-so is buried here" (helps them to increase taxes) and local thieves can sell Copernicus-themed junk to gullible visitors. On the science side, it is probably worthless.
And seriously harden up the electronics. If the Pioneer and Voyager probes can do 30+ years, a modern probe can. Given the fuel efficiency of the ion drive, a probe could also carry enough fuel to perform a great many missions. It may not be able to do everything on the list, but a decent design should be able to tick off a fair few at less cost than one probe for each one.
Some grid software is heavy - Globus could do with being ripped into sensible-sized modules - but the notion of a "Grid" is very simple. You have software where components don't care where other components are located. The Service Location Protocol, Zeroconf, RPC w/ Mobile IP - these are all ways of providing lightweight grids. MOSIX over a wide-area network would be a Grid of sorts. None of these are designed by committee. Now, some Grid systems are. The aforementioned Globus being an example. As committee-designed software systems go, it's actually not that bad. It's overweight and inefficient, but there are far far worse examples, and at least it's maintained and some very nice add-ons (like SASL security) exist.
But... but... what if they're pink?
Yes, that's the stuff, but - yuk - it looks very incomplete. It's unclear as to what's there and what isn't, VHDL isn't an easy format to work in, and it'd take a while for even those who know what they're doing to get that into any reasonable shape. I don't even come close to that level of expertise in signals theory.
I've bookmarked it and will start hacking on the code this weekend.
True, but wouldn't that be just as true for any wireless mesh that tries to handle reliable connections over paths that can change pseudo-randomly on-the-fly? DTN is presumably no more expensive on the compute cycles or memory than any of the 200+ methods on Citeseer for doing the same job specifically for terrestrial wireless, but is presumably much more versatile as it was designed to handle far worse situations.
Talking of noisy links, are there any open source implementations of Turbo Codes? (This is an error-correction scheme that is supposed to be designed specifically for errors that are not scattered randomly but concentrated into chunks. Reed-Solomon is better for scattered bit errors, provided the density is low enough. I can find open source Reed-Solomon implementations, but I haven't seen any for Turbo Codes.)
It would be nice if there were packet drivers for the protocol for Linux, NetBSD and NS-3 (the current network simulator). This would allow people to get a good feel for the behaviour of the new protocol, which may have uses beyond deep space. (It is possible to imagine real-world networks on Planet Earth where the characteristics of the network compare with inter-satellite communications.) Besides, would you rather Europe's answer to GPS use Windows?
Elebenty of tehm! Running Beowulfs!
Hey, that's damn secure! If the corefiles are corrupt or missing, unauthorized users can't figure out how to crash the software.
It is headed by the only Linux nerd who could afford to chase a rating of 6 or above. (7 is the highest the EAL will go.) Another thing to consider is that EAL ratings are only valid for a combination of OS and hardware. So, running Windows on any box (even if functionally identical) to the configuration tested on makes the tests invalid. The true is arguably the same for Linux, except that you can download LTP and gain some measure of assurance (even if not blessed on that platform) that you've not broken any of the security.
The highest old-style NSA rating (A1) is superior to the current EAL6+, and general-purpose OS' did achieve it. Genesis was one (and, no, not the one with the Phil Collins plugin module). EAL6+ looks to me to be about the same as the Orange Book B3 classification, which Trusted Irix achieved. Linux, if LTP was extended enough, could be provisionally ratified up to this level. If it ever was, then I could see vendors like IBM (who got Red Hat certified up to EAL4+) or private millionaires either individually or (more likely) jointly funding the certification.
Of course, EAL-style security isn't everything you need. Security labels on packets would be good - isn't there some work on this already? Support for hardware MAC (mandatory access controls) for memory would be good, as that protects not only against memory access violations in software, but also against such violations with RDMA. (Of course, if the hardware isn't present, you don't get that security, but likewise if the OS support isn't present, you don't get the security.) Better support for hardware encryption - especially within OpenSSL and IPSec - would improve matters too. Coverity is a decent-enough static checker, but their much-vaunted cooperation with Open Source doesn't seem to be producing much in the way of results - I can't remember the last time anyone covered on Slashdot or LWN any work by them. Are the major Linux vendors considering alternatives like Klockwork or any of the theorum provers listed just the other day?
Linux is already very good, but it hasn't received the severe auditing of OpenBSD (although, arguably, Linux does better when it comes to bugs that aren't security holes and also does better on the feature set and hardware supported). Perhaps a round or eleventy of severe auditing would be good for it. There again, perhaps there are other means of being close enough to that level of effectiveness without cutting back on the flexibility and without demanding unreasonable resources.
First, as others have noted, there is a massive level of sheer scientific curiosity. Prior to this, we didn't know of any planet other than Earth that ever had liquid water on it. We had no idea if such planets were rare or common, or even how to identify them if the water wasn't extremely visible and obvious. This allows us to know so much more about planets and their evolution in early solar systems than we ever knew before.
Then, there is another side. Water, particularly if it is mildly acidic, leaves open the possibility of cave systems. Cave systems make manned exploration a more realistic possibility, as you're better shielded from cosmic radiation, much better shielded from dust devils, and have a (comparatively) easy environment to seal and pressurize.
Finally, the combination of a lower gravity and a lower air pressure (whilst a significant atmosphere lasted) may make for crystals that are very different from those that form naturally on Earth. They should be slightly higher purity, for a start. This would not pay for exploration of Mars, or even significantly offset the costs, but it might well intrigue enough of the uber-rich (who tend to like unique trinkets) to either coerce Governments to fund exploration or provide some of the money themselves, purely for the bragging rights of having superior-grade, all-natural, extraterrestrial gemstones.
HOL is said to be usable, but doesn't have the same opportunities for name abuse, though 2001 fans could come up with a whole bunch of new ideas. Isabelle, however, would provide such opportunities, if you were to combine it with the package that started this thread. ACL2 is the least abusable name of all, so is quite useless.
You are correct, but nobody is going to buy a distro for an Ultrasparc IV (unless there's a LOT of support behind it) and there's a limit to how many who download from their site who will then buy their hardware. Sun has done some amazing things for Open Source (such as opening up the source to their microprocessors, although I still think placing the code on Open Cores would be smart, and they're not exactly forthcoming as to how the community has responded), but ultimately their core survival depends on shifting Big Ticket items. You'd have to sell a LOT of Solaris or Linux CDs to compare with the net profit obtained by shipping even a single T2-based server blade.
This does not mean backing away from Open Source. Far from it. Software is expensive to maintain, generates relatively little revenue and Sun's acceptance of Open Source clearly demonstrates they can no longer compete in that market. What Sun really needs to do is switch their software efforts from original development to QA, porting and optimization, and to gear up where they can weather the storm. IBM has done remarkably well by leveraging their skills in processors to produce low-end chips like the PPC, PPC64 and Cell. A server processor might not be viable for a mobile phone, but I'm willing to bet they could produce a version that was easier to write for than a Cell and better at the bus-intensive operations needed by MMORGs and other modern games than the ix86 architecture, which means they should be able to produce something that could compete in the flourishing console markets (which is what IBM is trying to do with the Cell). Since servers are bus-intensive and it's more important to shove data at the GPUs (which can do all the serious maths these days better than CPUs), starting with a CPU that is designed for massive data processing than one designed for massive number crunching would seem logical enough.
I don't seriously believe Sun can survive if they stay so deeply enmeshed in software, as it is only in hardware that they are distinct and only in hardware that they really have any advantage.
I don't know about anyone else's definition, but I would consider distros marking themselves as Phase 2 Certified as one way you could define "pushing IPv6". Another might be to enable IPv6 by default in the kernel (since we're talking about IPv6 capability in the article, rather than usage), or to use IPv6 by default on all connections where it is supported at the kernel and application level, whether or not it is genuinely supported end-to-end. Far as I can see, very few distros are certified for IPv6 (I can't honestly remember seeing the logo anywhere), those that provide it don't take advantage of it, and those that do don't make it easy to take advantage of it (Ubuntu provides some IPv6 tunneling software, but nothing that works with any of the IPv6 gateways I'm using, and network administration using IPv6 under Ubuntu is a pain -- and it's by far the best distro I've used in this regard).
Why is a duck?
If I could have gotten away with calling 98% of my exams "junk", not understood the remaining 2% of the questions, and passed with grades so good that people would give me the kind of cash genetics labs are getting, I'd have retired at age 30 to my own island.
How many genes must a geneticist wear out before we call it a gene?
...I'm not (yet) convinced of the value of the gene-mapping you can currently buy. $1000+ and you get back a description that is essentially meaningless because they don't really understand how the genes work yet. You get tested for a handful of conditions which have genetic links, but not all. (Genetic studies have shown there to be 7 forms of ME, according to the specific genetic cause, but very few labs will test for any of them yet.) Without knowing more about how genes work, it is impossible to know if what these studies reveal is even an accurate reflection of the genetics behind such conditions.
Alongside that is an argument in the reverse direction. If genes are not necessarily contiguous and/or can have ill-defined boundaries and/or can have components off the main DNA itself, then there is a definite possibility that there may be additional regions that could be useful for deep ancestry and genealogical DNA testing. This could help enormously as current research is pushing the limits of what is knowable using the regions and markers that are currently available. Entire haplogroup trees have been redefined because new information has revealed flaws in the previous models. More data, preferably more data that changes slowly, could be useful in getting these models right rather than continuously patching them.
Yes, but Deep Thought says Earth will take some time to figure it out.
I've used Ada a fair bit, but the biggest joke in that particular line when I used the language tended to be that none of the compilers complied with all aspects of the specification because it was just too complex. Ada is powerful and I do like many aspects of Ada (including the fact that many common programming errors are impossible in the language) but it is... clunky in some respects and the runtime components tend to be heavy. Modula-2 I don't like at all - it has a tighter structure than Pascal but I've not seen any way in which it does so that is actually useful.
When it comes to programming languages, I am not satisfied with any of the languages currently out there and feel they tend to either be over-engineered or over-reliant on evolution to fix a lack of engineering. There are some that seem to have a better balance, but they tend to be obscure and therefore insufficiently exercised to be sure that this appearance reflects reality. If they were better, it would seem they'd be used more.
I dunno - books describing how to analyze problems, produce flow charts, abstract data types and abstract state machines, common patterns, useful algorithms, jackson structured programming, and the like, could be applied to many situations and give a good grounding that could be used to then take a computer course or any other course that is based on the organization of processes. They are also techniques that require not a whole lot more than a pencil and paper to practice.
The same would be true of books covering cellular automata, truth tables, boolean algebra or even petri nets. Now, the middle two might be more interesting to those who later want to go into EE rather than CS, but they're still worth knowing.
Books describing when and how to use algorithms, herustics and fuzzy logic would be fairly basic to read, require nothing to really practice, and would again be good background material for many courses.
What would be useless would be things that are implementation-specific or which require extensive hands-on experience. So, books on programming in C++ or developing device drivers for Linux 2.6 would probably not be worth a whole lot.
What about something on the complete ADA specification?
Drinking too much of the tequilla beforehand might impact the observations.