ERGM and Excalibur are GPS-guided, MRM has a dual mode SAL/IR seeker (SAL is semi-active laser). I've held MRM in my hands and seen it after a flight test (them things is *sturdy*). Wrote some of the test interface software.
Surviving launch shocks of 12-14 kilogees is no joke, and is only now becoming reliable. I sooooooo doubt that there is anything that will survive 45kgee.
Not to say they shouldn't try: definitely go for it. Hell, it would probably easier just to further harden a few of the existing projectiles than develop from scratch a new one (though they fly a lot faster than what the existing things go, so aerodynamic constraints might necessitate a redesign).
What interests me is how well electronics inside a shell will survive that intense a magnetic field. Monstrous Faraday shield around the round? Can you really seal the whole round but yet make it possible to deploy fins or canards for maneuvering?
Even more super dooper: will you have to make the round both super acceleration safe (launch shock) and then also rad-hard (it's not in space long, but neither is EKV and that's rad-hard). Course an EKV firing might be during a, uhm, "excitable" period when space has rather more radiation than normal.
What I especially like is that basically every defense contractor is on the front page of that Powerpoint presentation except the people who've gotten rounds to fly. In fact, ATK was found guilty of fraud while developing their version of MRM. They claimed to have hit a tank, but they ballistically dialed the round in beforehand: no guidance was needed to hit.
Oh God, Rockwell Collins is on that list. Well, if their GPS "expertise" is being utilized, no worries about the round hitting a goddamn thing. Yay, nothing like getting Segmentation Faults from your navigation unit during flight. We really enjoyed that. Nice job you clowns. L3's not much better.
I was at my folk's for Christmas, and somehow we got around to my dad showing me his APL language reference something or other. It, too, has ridiculous numbers of symbols, and requires a specific keyboard configuration to use. Check it out at Wikipedia (keyboard image is about halfway down). It's wacky, I mean I never imagined that floor() and ceil() would ever have their own operators.
Lawyers are evil pieces of shit who should be carved into bleeding chunks with fishing line and monstrous tension. We'll vitrify the chunks, build a probe, and launch the solid result toward the nearest blackhole to remove their contamination from the universe altogether.
The jury pool is comprised of citizens of the United States (largely: one presumes the assorted illegal is sucked into the system). Unless we're willing to have some sort of jury qualification exam, i.e. disenfranchise a chunk of the citizenry, which I think's largely a non-starter and probably should be (that's a hell of a consequence for false negatives). However, I think that there are certain *suits* that it is reasonable to remove people from, and if you could rely on lawyers to not be a toxin in the bloodstream of humanity (oh, darn), then you could hope that this selection would be a decent process.
But it's not. The company I work for is the largest private contractor (non-government) in the south of the *state*, and is 80% engineers. But for all that, almost everyone I know is instantly removed from the jury pool: saying "engineer" is like announcing that you have the Monopoly "Get Out of Jury Duty Free" card. Oh my GOD, they might consider the evidence! They might be capable of understanding complicated matters and not voting based on what they had for lunch!
This happens because, for lawyers, there is no incentive not to game and fuck the system. No, no one expects them to act like humans and simply consider what is proper and correct for justice's sake! You can't have the defense and prosecution teams get together and, like DECENT GODDAMN HUMANS, simply say, "Unfit, fit, unfit, fit" based purely on intellectual, etc. merits. No, it has to be this goddamn caricature of a process where each looks for the person mostly likely to be swayed to their point of view, regardless of the consequencs.
And there is no oversight to say, "Oh, bullshit, you picked 4 illiterate hicks from the pool of 60 people in the profession under discussion?" NOTE: you could probably easily stick the 4 illiterate hicks in various criminal cases where their judgment, if unlikely to be considered "as good", should be "sufficient". But we/never/ do that. Lawyers being disbarred? Don't make me fucking laugh: there's no oversight there. It's lawyers disciplining lawyers. Like Congress with ethics violations. Beautiful goddamn idea. Do we really do that for any other professions: answer, NO. You can say "doctors", but that's obviously stupid -- *how* many malpractice attorneys are there?
This doesn't happen in engineering. Here is Feynman on the Challenger Shuttle debacle:
For a successful technology, reality must take precedence over public relations, for nature cannot be fooled. And it is rarely in the financial (and thus, corrosive) interests of engineers to fuck people. Definitely it happens, but ask if institutions run by engineers are as fucking TOXIC as organizations run by lawyers. Anyone disagree?
And I have worked with engineers I personally loathed and whose work and careers it would absolutely tickle me to destroy, but I don't, because the end result is a bad product, and you cannot HIDE that. Also, I have a soul. Also, there are consequences. Can you say the same about the law? I don't think so.
My job is produce things that work. A lawyer's job should be to produce justice. But we've decided that only lawyers can be in charge of the incentives to behave for lawyers, and that's simply a receipe for disaster for everyone except lawyers.
Work with me, dude: we have these wacky things called batteries. Or supercapacitors. Or super-bitchin' flywheels (my personal favorite). Or heating a fluid in an insulated vessel. Or separating water.
While I agree that the market may not work for everything, I think a lot of that is because of the incentives we have set up in the law really distort things.
Note, however, that corrective forces are at work. There are a bunch of insurance companies that are forcing heavy payments onto companies in at-risk areas. The problem is, the government heavily subsidizes citizens in places like Florida or New Orleans. If they didn't, then we would see (a) fewer [non-wealthy] people living on the beach (b) more pressure to reduce things that cause environmental problems.
The problem currently is one of information: how do we include *all* the costs of burning petroleum into its price, including environmental effects? I think the EU (European Union) has made a brave start with their carbon market (and California, now, too): now we need to figure out at what level to set carbon emissions targets (which is in progress now planning targets for the next few years, I believe). So far that's been gamed, though: the allowed level of emissions was *higher* than the combined emissions of industry for 2005 and 2006 (sayeth the Economist).
So, the market can do it (at least more ably), but only if you are able to fully include the information as to cost. It's still a problem, of course: how do you keep the current generation from deferring the cost to those who have no voice? But there are better ways to do it then heavy-handed regulation by bureaucrats who have never had a real job.
Uhm, hey. You should let the PS2 play it for you. I didn't figure this out myself until after driving a damn Miata (MX-5) around and around. You can do this either at the beginning of the race, or at a pit stop. You want to switch to "B-Spec". This makes you more of a manager, and changes the points you get or whatever, if you care. In this mode, you can accelerate the flow of time up to 3x, and thus finish in only (!) 8 hours. However, time reverts back to 1x afer every pit stop, and I think I usually got about 3-4 laps per set of tires in the Nurburgring, so that's about once every 6-10 minutes (remember, 3x) you have to come back and nudge things. Or just wander away and the PS2 will run things at 1x all the way through.
However, as you know driver intelligence is miserable, so you may have to be careful. Usually I build up a big lead in the first few laps, or choose the most inequitable car matching I can, and only then hand over my machine to the computer retard.
As for security matters: If anyone wanted to create havoc, they'd take one glance at the report and burn down the sites responsible for the largest outages listed. "National infrastructure" is described in painstaking detail. It wouldn't take a criminal mastermind - only a couple of drunk high school kids.
Might you then have to consider (scary thought ahead!) ACTUALLY FUCKING PROTECTING THE NATIONAL INFRASTRUCTURE?!
Oh noes, it costs us money to do that and we would far rather censor the reports than attempt to protect against "a couple of drunk high school kids".
Thank god that people like you, sir, are here to Secure by Obscurity(tm) the pathetic risk reduction and safety habits of our corporate overlords!
That stuff sounds kinda neat, but after reading the SLAM summary paper, it looks pretty similar to the Stanford Checker extensions to gcc, which eventually turned into Coverity. Seems like there are some differences: I don't think that the Checker has to do the rewrite to the Boolean language thing, but the whole idea of rules like "match lock with unlock" seems analagous. Recent Linux kernels have featured some reasonably neat gcc stuff that warns if you don't check return codes and the like.
The Vulcan thing sounds pretty cool, though. Wonder what the overhead is?
Er, how about not making everything such a gigantic, monolithic, unwieldy piece-o'-crap that you have subtle bugs propogate through your system? How about using the magic of 'make' and not rebuilding everything? I know that there's value in a 'make clean ; make', but I'm guessing that if my stuff took hours to build, I'd invest a lot of effort to minimize those needs.
And how 'bout distributed compilation? I mean, I'm basically my own administrator and chief developer, and I setup icecream (http://en.opensuse.org/Icecream) on about 30 processors in only a few minutes. I know there's other stuff with distcc and some caching junk to speed things up even more, but I only have about 14k lines of code in my main project (well, soon to be at least an order of magnitude greater once I begin integration), and it takes only seconds to compile. Surely this is a job for oodles of $400 blades with Athlons and dual Gig-E NICs that you can channel bond for bitchin' throughput.
To me it sounds like the whole build process needs a hell of a rethink by some people who know what they're about. I mean, I think it took an hour to bootstrap gcc 3.4 on a P4 2.5GHz a couple of years ago. With 30 processors and perfect parallelization, I should be able to compile 30x that much source in an hour. I'll just say that processor improvements cancel out less-than-linear scaling and that the same is true today. Just how much bigger is your tool than gcc, and *why*? Seriously. I'm not building Linux from scratch each time I want to upgrade gcc. Libraries and modularization are good.
Okay, here's the only specifics I saw in the article:
This includes not just software for computers and networks but, in some cases, programs for military aircraft, missile guidance, and battlefield management systems.
Okay, I can believe that "battlefield management systems" could have some commercial junk that came from somewhere, but otherwise I find large parts of this less than convincing.
About the missile guidance part I say: bullshit. Hell, for a lot of missiles, particularly older ones, the processor is custom, and so is the operating system: I repeat, no part of the missile was developed by people without citizenship and clearances. A more modern missile might run VxWorks (for example), but so what? Everything under the operating system (the board support package necessary to make VxWorks run on custom hardware), and everything on top of the operating system is largely custom, classified, and written by clearance-bearing people. Certainly you don't just buy Missile Guidance v3.2 (now with support for your uncooled IR focal plane array!) from Habeeb's Software Hut. And missile software has what might be considered a, you know, reasonably thorough testing process, where some care is taken to verify inputs and not just take things on faith.
And how do you subvert these systems? What, you DoS a Tomahawk? Or it accepts your logon attempt and you then your run your buffer overrun 'sploit and become root? There are a rather finite set of inputs to the system and access to them is controlled. I think missiles are pretty safe. Most are too single purpose to care about things that cause more general purpose computers problems.
So, yes, maybe software systems like battlefield control systems or radar control that use commercial databases or the like -- but one does not put these on the 'Net. And I'd hope that the NSA is providing some pretty rigorous guidance like "use NSA-Linux".
Signed on October 17th. Look in Section 212, which has this:
(A) by striking `Director of the Defense Advanced Research Projects Agency' and inserting `Director of Defense Research and Engineering and the service acquisition executive for each military department'; and
Emphasis mine. You can see that now they have to add a dude (assuming that Director of Defense of DARPA is now "Director of Defense Research and Engineering", otherwise it's out of DARPA's hands all together). Maybe it's just a matter of signatures, but I can see how they have been forced to put the award on hold until they can, you know, obey the law as Congress has fiddled with it. And I confess that I haven't looked at the legislation that this section amends, which is:
Subsection (a) of section 2374a of title 10, United States Code
I can't see the referenced webpage, but the only SBIRS (I think there's usually an 'S' on the end to indicate Satellite) I know of is the planned system of missile plume detection satellites. These satellites are looking down for the (apparently) distinctive plumes of a missile launch in order to allow for the launch of an interceptor like EKV (Exo-atmospheric Kill Vehicle), KEI (Kinetic Energy Interceptor), SM3 (the Navy's Standard Missile 3 for use with Aegis), or even potential future programs like NFIRE (Near-field Infrared Experiment).
Last I heard SBIRS was ridiculously over-budget ($10 billion vs originally planned $4 or something), so this is probably the world's most expensive CCD by an order of magnitude or two.
Oh, bullshit. Sure, a lot of standard gear is from U.S. companies, but I doubt that much of it is made here, or inspected with any kind of rigor. Where I work we mostly use HP machines, but I'm putting a bunch of machines from IWill into a closed area with nary a peep. I put serial cards from Axxon into some computers, and though it was annoying to get them, it was mostly due to it being an "International Purchase Order" than because they are Fiendish Canadians.
Yeah, U.S. companies are preferred and are largely required for software, but the inspection process is anything but rigorous -- don't make me laugh. Those clowns at DSS wouldn't know an exploit if it lept up and bit them in the ass. Maybe there are clued folks at the NSA helping in the background (not that NSA seems particularly clued-in to me anymore), but I imagine that Suse Linux, which was verboten, is probably now okay since it's Novell. Think the codebase has changed? Does any of that matter for Linux anyway?
And how does one transmit information since by definition classified machines are not connected to the outside world in any way? Is it N-rays?
It's all ridiculous paranoia by people who all type with two fingers while staring at the keyboard, or who have their email printed by their secretaries and don't understand that there are other means to edit text besides Word (and I don't mean Notepad).
It might get exciting if they fly over/around the White Sands Missile Range (WSMR). They shoot some pretty impressive missiles there, including THAAD, which is exoatmospheric. THAAD has to pull some pretty crazy maneuvers (tight spirals) right after launch to burn off enough fuel to stay within the range. It has a flight termination device (self-destruct), too, I'm sure. But accidents, they do happen.
Also fired there: ESSM, Standard Missile, various other explosive critters. I didn't look to see exactly where the spaceport will be, but WSMR is near Las Cruces: Alamagordo is on the far side.
You don't need the whole missile. You need the guidance electronics and the seeker. I've seen THAAD's HIL.
Yes, you absolutely can project IR imagery. Look for the word "mirror". You can make mirrors that are RADAR transparent. I've seen them.
I don't the details of THAAD, but for the lower stages (since they simulate from launch), you intercept fin commands and use that information within a simulation to calculate updated attitude and position information. This is then used in returning data to the missile (altering IR imagery, etc.).
For the upper stage, they can detect the commands to set off the little "poppers", and they jitter the image with a high-speed mirror. They adjust the position, too.
For my missile, we don't feel a HIL is necessary: we will inject everything, including the video, and fully simulate the inertial measurement unit (IMU). But what if the Army wants to take a missile out of its container and test it? You can't tell them to take the whole damn thing apart, but you can usually get them to take the motor off. Then it goes in the table. It's not strictly necessary, since you can provide modes and still simulate everything, but they have millions of dollars and it gives them warm fuzzies, and they are the customer.
For THAAD, they may have rate sensors on their seeker that need to be fooled by putting them on a table. They may want to combine real and synthesized IMU data (real data for rates, synthesized data for unsimulatable acceleration (gravity, thrust)). Javelin and AIM-9x have to do this for various reasons: I'm not sure of them all. I think Javelin has *no* IMU: because it's lock-on before launch, you can figure out how you're moving based on how the seeker gimbals are pointed. Something like that.
You do HIL tests to make sure things work. For the early models, you want to make damn sure your seeker is good before you shoot off a $1 million missle in a test that costs $30 million (range time is *damn* expensive, and THAAD can only use White Sands cuz they shoot so far). In general, including as much of the flight hardware as you can is good. If you can't mount the CAS on the 3-axis (or whatever: I've seen up to 5-axis (3 for missile, 2 for target (no roll)) table, then you can put it on the floor in a test fixture and apply torques to the fins to simulate pressures and stresses. Depends upon what your requirements are.
In general, though, you use what we call a CIL: Computer-in-the-loop. No seeker: imagery is all injected, no IMU: all simulated, etc. This is purely for flight software and electronics hardware check out. I make these, too.
and I hate them all with a passion. I've been fighting with software installation on the older Origin 2200 (8 400MHz processors, 6GB of RAM). SGI's crap compiler can't bootstrap gcc 4.0.2, their versions of common Unix tools like grep, etc., suck (forcing you to upgrade to the GNU versions, if their stupid compiler can build them), and IRIX has been at release 6.5 since 1998 or something. Sure, they want you to move to their new Linux-based Prism machines, and I've got one of those, too. Yippee, Itanics! What a super swell processor! I have an 8 processor Origin 300 where the total power consumption of all 8 processors is less than the consumption of 1 of the Itanics! See also, the poor code produced by gcc for this processor.
So, anyway. Upgrading SGIs sucks, their hardware is immensely fragile, its very persnickety about its environment (god forbid the temperature in the room not be in the 60s), licensing all their tools is hellish, their debugger is ancient and decrepit, my tech is a retard who tried to cable together the Origin 300 incorrectly and I had to fix it for him, and get this -- 8GB of RAM for an Origin 300 cost $25,000. That's right: $25k. You know what it is: it's PC3200 with some goddamn proprietary bullshit thrown in so you have to order your parts from SGI.
I'm glad you're dying. You've made every misstep possible: lets sell Windows NT machines! You sell Fuels in regular ATX cases with rockin' 800MHz processors that start at something like $10k. Your video offerings, once your strong suit, suck -- all you offer is older ATI cards in crap configurations -- $40k for two cards since I needed a new node (didn't buy it, duh).
The only reason to buy an SGI in the last five years or so is because of the good realtime performance of IRIX: I can sustain 16us interrupt times pretty much forever. But that's it. I'm not paying $130k for another slow-ass computer without even a damn video card for a console. And I don't need to: Ingo Molnar's realtime patches are coming along, and my quad Opteron box wipes the floor with the Origin and cost, oh yeah: $19,992 including shipping, and $7k of that is pimpin' SCSI disks.
Some contracts are "cost plus", where there is a guaranteed level of profit: all extra costs are eaten by the government. This makes okay sense, since usually the government jerks companies around (oh, now we want it in purple and to cure cancer...). This guaranteed fee usually results in profits of less than 10% (the government is looking to bump this somewhat since they can't find takers for some of the work they want done). Take a look at commercial companies and see what kinds of margins they make.
Other contracts are "fixed fee": build it to the defined price or you take it in the keister. Government still jerks you around.
As far as competition goes: it's there for some things, and not for others. But how many people can you expect to make fighters when you only ask for a new one every 10-20 years? Between F-22 and JSF is about 15 years or so. Lockheed Martin and Boeing have had to combine their satellite launch operations because they can't make money off military launches.
And EADS (European Aerospace something xyz) is now competing in the U.S.
Frankly, I'm tired of hearing about this "subsidized this, blah that" from people who know jackshit about how the industry works. The government can sign contracts, then just as quickly obviate them with no repercussions (see the F-22 contract, or DDX (the next gen Navy destroyer), or just about every other damn thing). The government defines who things can be sold to and under what conditions, and requires a fuck ton of paperwork to sell even things like radios (ITAR regulations).
Do your homework before you spout off. There's a difference between AirBus saying "gimme money" and the EU saying "how much?" and Boeing fighting off LockMart, Northrop Grumman, Raytheon, EADS, ATK, etc. for a chance at a contract. I can't speak to the subsidies from various states (Washington, etc.), but I notice Alabama is offering substantial financial incentives to AirBus to built a plant in Mobile, so it looks like that assistance is basically equivalent.
Slashdot Mirror
No: that's what they *want*, not what they have.
I know the guys who build:
ERGM
Excalibur
MRM
ERGM and Excalibur are GPS-guided, MRM has a dual mode SAL/IR seeker (SAL is semi-active laser). I've held MRM in my hands and seen it after a flight test (them things is *sturdy*). Wrote some of the test interface software.
Surviving launch shocks of 12-14 kilogees is no joke, and is only now becoming reliable. I sooooooo doubt that there is anything that will survive 45kgee.
Not to say they shouldn't try: definitely go for it. Hell, it would probably easier just to further harden a few of the existing projectiles than develop from scratch a new one (though they fly a lot faster than what the existing things go, so aerodynamic constraints might necessitate a redesign).
What interests me is how well electronics inside a shell will survive that intense a magnetic field. Monstrous Faraday shield around the round? Can you really seal the whole round but yet make it possible to deploy fins or canards for maneuvering?
Even more super dooper: will you have to make the round both super acceleration safe (launch shock) and then also rad-hard (it's not in space long, but neither is EKV and that's rad-hard). Course an EKV firing might be during a, uhm, "excitable" period when space has rather more radiation than normal.
What I especially like is that basically every defense contractor is on the front page of that Powerpoint presentation except the people who've gotten rounds to fly. In fact, ATK was found guilty of fraud while developing their version of MRM. They claimed to have hit a tank, but they ballistically dialed the round in beforehand: no guidance was needed to hit.
Oh God, Rockwell Collins is on that list. Well, if their GPS "expertise" is being utilized, no worries about the round hitting a goddamn thing. Yay, nothing like getting Segmentation Faults from your navigation unit during flight. We really enjoyed that. Nice job you clowns. L3's not much better.
I was at my folk's for Christmas, and somehow we got around to my dad showing me his APL language reference something or other. It, too, has ridiculous numbers of symbols, and requires a specific keyboard configuration to use. Check it out at Wikipedia (keyboard image is about halfway down). It's wacky, I mean I never imagined that floor() and ceil() would ever have their own operators.
The jury pool is comprised of citizens of the United States (largely: one presumes the assorted illegal is sucked into the system). Unless we're willing to have some sort of jury qualification exam, i.e. disenfranchise a chunk of the citizenry, which I think's largely a non-starter and probably should be (that's a hell of a consequence for false negatives). However, I think that there are certain *suits* that it is reasonable to remove people from, and if you could rely on lawyers to not be a toxin in the bloodstream of humanity (oh, darn), then you could hope that this selection would be a decent process.
But it's not. The company I work for is the largest private contractor (non-government) in the south of the *state*, and is 80% engineers. But for all that, almost everyone I know is instantly removed from the jury pool: saying "engineer" is like announcing that you have the Monopoly "Get Out of Jury Duty Free" card. Oh my GOD, they might consider the evidence! They might be capable of understanding complicated matters and not voting based on what they had for lunch!
This happens because, for lawyers, there is no incentive not to game and fuck the system. No, no one expects them to act like humans and simply consider what is proper and correct for justice's sake! You can't have the defense and prosecution teams get together and, like DECENT GODDAMN HUMANS, simply say, "Unfit, fit, unfit, fit" based purely on intellectual, etc. merits. No, it has to be this goddamn caricature of a process where each looks for the person mostly likely to be swayed to their point of view, regardless of the consequencs.
And there is no oversight to say, "Oh, bullshit, you picked 4 illiterate hicks from the pool of 60 people in the profession under discussion?" NOTE: you could probably easily stick the 4 illiterate hicks in various criminal cases where their judgment, if unlikely to be considered "as good", should be "sufficient". But we
This doesn't happen in engineering. Here is Feynman on the Challenger Shuttle debacle: For a successful technology, reality must take precedence over public relations, for nature cannot be fooled. And it is rarely in the financial (and thus, corrosive) interests of engineers to fuck people. Definitely it happens, but ask if institutions run by engineers are as fucking TOXIC as organizations run by lawyers. Anyone disagree?
And I have worked with engineers I personally loathed and whose work and careers it would absolutely tickle me to destroy, but I don't, because the end result is a bad product, and you cannot HIDE that. Also, I have a soul. Also, there are consequences. Can you say the same about the law? I don't think so.
My job is produce things that work. A lawyer's job should be to produce justice. But we've decided that only lawyers can be in charge of the incentives to behave for lawyers, and that's simply a receipe for disaster for everyone except lawyers.
Work with me, dude: we have these wacky things called batteries. Or supercapacitors. Or super-bitchin' flywheels (my personal favorite). Or heating a fluid in an insulated vessel. Or separating water.
There are lots of ways to store juice.
While I agree that the market may not work for everything, I think a lot of that is because of the incentives we have set up in the law really distort things.
Note, however, that corrective forces are at work. There are a bunch of insurance companies that are forcing heavy payments onto companies in at-risk areas. The problem is, the government heavily subsidizes citizens in places like Florida or New Orleans. If they didn't, then we would see (a) fewer [non-wealthy] people living on the beach (b) more pressure to reduce things that cause environmental problems.
The problem currently is one of information: how do we include *all* the costs of burning petroleum into its price, including environmental effects? I think the EU (European Union) has made a brave start with their carbon market (and California, now, too): now we need to figure out at what level to set carbon emissions targets (which is in progress now planning targets for the next few years, I believe). So far that's been gamed, though: the allowed level of emissions was *higher* than the combined emissions of industry for 2005 and 2006 (sayeth the Economist).
So, the market can do it (at least more ably), but only if you are able to fully include the information as to cost. It's still a problem, of course: how do you keep the current generation from deferring the cost to those who have no voice? But there are better ways to do it then heavy-handed regulation by bureaucrats who have never had a real job.
Uhm, hey. You should let the PS2 play it for you. I didn't figure this out myself until after driving a damn Miata (MX-5) around and around. You can do this either at the beginning of the race, or at a pit stop. You want to switch to "B-Spec". This makes you more of a manager, and changes the points you get or whatever, if you care. In this mode, you can accelerate the flow of time up to 3x, and thus finish in only (!) 8 hours. However, time reverts back to 1x afer every pit stop, and I think I usually got about 3-4 laps per set of tires in the Nurburgring, so that's about once every 6-10 minutes (remember, 3x) you have to come back and nudge things. Or just wander away and the PS2 will run things at 1x all the way through.
However, as you know driver intelligence is miserable, so you may have to be careful. Usually I build up a big lead in the first few laps, or choose the most inequitable car matching I can, and only then hand over my machine to the computer retard.
Now you know. And knowing is half the battle.
He's talking about O.J. Simpson, you knucklehead.
As for security matters: If anyone wanted to create havoc, they'd take one glance at the report and burn down the sites responsible for the largest outages listed. "National infrastructure" is described in painstaking detail. It wouldn't take a criminal mastermind - only a couple of drunk high school kids.
Might you then have to consider (scary thought ahead!) ACTUALLY FUCKING PROTECTING THE NATIONAL INFRASTRUCTURE?!
Oh noes, it costs us money to do that and we would far rather censor the reports than attempt to protect against "a couple of drunk high school kids".
Thank god that people like you, sir, are here to Secure by Obscurity(tm) the pathetic risk reduction and safety habits of our corporate overlords!
That stuff sounds kinda neat, but after reading the SLAM summary paper, it looks pretty similar to the Stanford Checker extensions to gcc, which eventually turned into Coverity. Seems like there are some differences: I don't think that the Checker has to do the rewrite to the Boolean language thing, but the whole idea of rules like "match lock with unlock" seems analagous. Recent Linux kernels have featured some reasonably neat gcc stuff that warns if you don't check return codes and the like.
The Vulcan thing sounds pretty cool, though. Wonder what the overhead is?
Er, how about not making everything such a gigantic, monolithic, unwieldy piece-o'-crap that you have subtle bugs propogate through your system? How about using the magic of 'make' and not rebuilding everything? I know that there's value in a 'make clean ; make', but I'm guessing that if my stuff took hours to build, I'd invest a lot of effort to minimize those needs.
And how 'bout distributed compilation? I mean, I'm basically my own administrator and chief developer, and I setup icecream (http://en.opensuse.org/Icecream) on about 30 processors in only a few minutes. I know there's other stuff with distcc and some caching junk to speed things up even more, but I only have about 14k lines of code in my main project (well, soon to be at least an order of magnitude greater once I begin integration), and it takes only seconds to compile. Surely this is a job for oodles of $400 blades with Athlons and dual Gig-E NICs that you can channel bond for bitchin' throughput.
To me it sounds like the whole build process needs a hell of a rethink by some people who know what they're about. I mean, I think it took an hour to bootstrap gcc 3.4 on a P4 2.5GHz a couple of years ago. With 30 processors and perfect parallelization, I should be able to compile 30x that much source in an hour. I'll just say that processor improvements cancel out less-than-linear scaling and that the same is true today. Just how much bigger is your tool than gcc, and *why*? Seriously. I'm not building Linux from scratch each time I want to upgrade gcc. Libraries and modularization are good.
Okay, here's the only specifics I saw in the article:
This includes not just software for computers and networks but, in some cases, programs for military aircraft, missile guidance, and battlefield management systems.
Okay, I can believe that "battlefield management systems" could have some commercial junk that came from somewhere, but otherwise I find large parts of this less than convincing.
About the missile guidance part I say: bullshit. Hell, for a lot of missiles, particularly older ones, the processor is custom, and so is the operating system: I repeat, no part of the missile was developed by people without citizenship and clearances. A more modern missile might run VxWorks (for example), but so what? Everything under the operating system (the board support package necessary to make VxWorks run on custom hardware), and everything on top of the operating system is largely custom, classified, and written by clearance-bearing people. Certainly you don't just buy Missile Guidance v3.2 (now with support for your uncooled IR focal plane array!) from Habeeb's Software Hut. And missile software has what might be considered a, you know, reasonably thorough testing process, where some care is taken to verify inputs and not just take things on faith.
And how do you subvert these systems? What, you DoS a Tomahawk? Or it accepts your logon attempt and you then your run your buffer overrun 'sploit and become root? There are a rather finite set of inputs to the system and access to them is controlled. I think missiles are pretty safe. Most are too single purpose to care about things that cause more general purpose computers problems.
So, yes, maybe software systems like battlefield control systems or radar control that use commercial databases or the like -- but one does not put these on the 'Net. And I'd hope that the NSA is providing some pretty rigorous guidance like "use NSA-Linux".
Then why does Norway have the highest standard of living in the world?
Oil.
Let's see how it goes after that North Sea oil runs out in ~10 years or so.
http://thomas.loc.gov/cgi-bin/query/C?c109:./temp/ ~c109i6ly2s
Signed on October 17th. Look in Section 212, which has this:
(A) by striking `Director of the Defense Advanced Research Projects Agency' and inserting `Director of Defense Research and Engineering and the service acquisition executive for each military department'; and
Emphasis mine. You can see that now they have to add a dude (assuming that Director of Defense of DARPA is now "Director of Defense Research and Engineering", otherwise it's out of DARPA's hands all together). Maybe it's just a matter of signatures, but I can see how they have been forced to put the award on hold until they can, you know, obey the law as Congress has fiddled with it. And I confess that I haven't looked at the legislation that this section amends, which is:
Subsection (a) of section 2374a of title 10, United States Code
It's in German...
i d=c3zrJD0TPekC&printsec=frontcover&dq=Enzyklopaedi e+des+Maerchens
i d=pN__NqrfGTcC&pg=PA9&lpg=PA9&dq=%22Encyclopedia+o f+Fairy+Tales%22+Walter+de+Gruyter&sig=6gbeiG569yu cQ6TwYIpMwSdIIU8&hl=en
http://books.google.com/books?vid=ISBN3110082012&
The subtitle is "Handbook of historical and comparative narrative research" according to:
http://books.google.com/books?vid=ISBN0422809403&
Circumference = 2*Pi*r
r = 1 AU
You're off by a factor of 2. Other points as noted.
It's easier when you don't have to fight the Sun, isn't it.
Weight changes, momentum does not. That could be ugly.
I can't see the referenced webpage, but the only SBIRS (I think there's usually an 'S' on the end to indicate Satellite) I know of is the planned system of missile plume detection satellites. These satellites are looking down for the (apparently) distinctive plumes of a missile launch in order to allow for the launch of an interceptor like EKV (Exo-atmospheric Kill Vehicle), KEI (Kinetic Energy Interceptor), SM3 (the Navy's Standard Missile 3 for use with Aegis), or even potential future programs like NFIRE (Near-field Infrared Experiment).
Last I heard SBIRS was ridiculously over-budget ($10 billion vs originally planned $4 or something), so this is probably the world's most expensive CCD by an order of magnitude or two.
Oh, bullshit. Sure, a lot of standard gear is from U.S. companies, but I doubt that much of it is made here, or inspected with any kind of rigor. Where I work we mostly use HP machines, but I'm putting a bunch of machines from IWill into a closed area with nary a peep. I put serial cards from Axxon into some computers, and though it was annoying to get them, it was mostly due to it being an "International Purchase Order" than because they are Fiendish Canadians.
Yeah, U.S. companies are preferred and are largely required for software, but the inspection process is anything but rigorous -- don't make me laugh. Those clowns at DSS wouldn't know an exploit if it lept up and bit them in the ass. Maybe there are clued folks at the NSA helping in the background (not that NSA seems particularly clued-in to me anymore), but I imagine that Suse Linux, which was verboten, is probably now okay since it's Novell. Think the codebase has changed? Does any of that matter for Linux anyway?
And how does one transmit information since by definition classified machines are not connected to the outside world in any way? Is it N-rays?
It's all ridiculous paranoia by people who all type with two fingers while staring at the keyboard, or who have their email printed by their secretaries and don't understand that there are other means to edit text besides Word (and I don't mean Notepad).
I love you.
It might get exciting if they fly over/around the White Sands Missile Range (WSMR). They shoot some pretty impressive missiles there, including THAAD, which is exoatmospheric. THAAD has to pull some pretty crazy maneuvers (tight spirals) right after launch to burn off enough fuel to stay within the range. It has a flight termination device (self-destruct), too, I'm sure. But accidents, they do happen.
Also fired there: ESSM, Standard Missile, various other explosive critters. I didn't look to see exactly where the spaceport will be, but WSMR is near Las Cruces: Alamagordo is on the far side.
Burning appendages indeed!
Whippersnapper (though, uhm, I'm 28), it's from 'advent':
u re#xyzzy
http://en.wikipedia.org/wiki/Colossal_Cave_Advent
Learn ye the etymology of the Significant Words.
I make HILs.
You don't need the whole missile. You need the guidance electronics and the seeker. I've seen THAAD's HIL.
Yes, you absolutely can project IR imagery. Look for the word "mirror". You can make mirrors that are RADAR transparent. I've seen them.
I don't the details of THAAD, but for the lower stages (since they simulate from launch), you intercept fin commands and use that information within a simulation to calculate updated attitude and position information. This is then used in returning data to the missile (altering IR imagery, etc.).
For the upper stage, they can detect the commands to set off the little "poppers", and they jitter the image with a high-speed mirror. They adjust the position, too.
For my missile, we don't feel a HIL is necessary: we will inject everything, including the video, and fully simulate the inertial measurement unit (IMU). But what if the Army wants to take a missile out of its container and test it? You can't tell them to take the whole damn thing apart, but you can usually get them to take the motor off. Then it goes in the table. It's not strictly necessary, since you can provide modes and still simulate everything, but they have millions of dollars and it gives them warm fuzzies, and they are the customer.
For THAAD, they may have rate sensors on their seeker that need to be fooled by putting them on a table. They may want to combine real and synthesized IMU data (real data for rates, synthesized data for unsimulatable acceleration (gravity, thrust)). Javelin and AIM-9x have to do this for various reasons: I'm not sure of them all. I think Javelin has *no* IMU: because it's lock-on before launch, you can figure out how you're moving based on how the seeker gimbals are pointed. Something like that.
You do HIL tests to make sure things work. For the early models, you want to make damn sure your seeker is good before you shoot off a $1 million missle in a test that costs $30 million (range time is *damn* expensive, and THAAD can only use White Sands cuz they shoot so far). In general, including as much of the flight hardware as you can is good. If you can't mount the CAS on the 3-axis (or whatever: I've seen up to 5-axis (3 for missile, 2 for target (no roll)) table, then you can put it on the floor in a test fixture and apply torques to the fins to simulate pressures and stresses. Depends upon what your requirements are.
In general, though, you use what we call a CIL: Computer-in-the-loop. No seeker: imagery is all injected, no IMU: all simulated, etc. This is purely for flight software and electronics hardware check out. I make these, too.
I have a bunch of SGI machines that I use where I work:
2x 8 processor Onyx2s
1x 8 processor Origin 300
1x 8 processor Origin 2200
1x 32 processor Origin 350
1x 4 processor Prism
3x 1 processor Octane2s
and I hate them all with a passion. I've been fighting with software installation on the older Origin 2200 (8 400MHz processors, 6GB of RAM). SGI's crap compiler can't bootstrap gcc 4.0.2, their versions of common Unix tools like grep, etc., suck (forcing you to upgrade to the GNU versions, if their stupid compiler can build them), and IRIX has been at release 6.5 since 1998 or something. Sure, they want you to move to their new Linux-based Prism machines, and I've got one of those, too. Yippee, Itanics! What a super swell processor! I have an 8 processor Origin 300 where the total power consumption of all 8 processors is less than the consumption of 1 of the Itanics! See also, the poor code produced by gcc for this processor.
So, anyway. Upgrading SGIs sucks, their hardware is immensely fragile, its very persnickety about its environment (god forbid the temperature in the room not be in the 60s), licensing all their tools is hellish, their debugger is ancient and decrepit, my tech is a retard who tried to cable together the Origin 300 incorrectly and I had to fix it for him, and get this -- 8GB of RAM for an Origin 300 cost $25,000. That's right: $25k. You know what it is: it's PC3200 with some goddamn proprietary bullshit thrown in so you have to order your parts from SGI.
I'm glad you're dying. You've made every misstep possible: lets sell Windows NT machines! You sell Fuels in regular ATX cases with rockin' 800MHz processors that start at something like $10k. Your video offerings, once your strong suit, suck -- all you offer is older ATI cards in crap configurations -- $40k for two cards since I needed a new node (didn't buy it, duh).
The only reason to buy an SGI in the last five years or so is because of the good realtime performance of IRIX: I can sustain 16us interrupt times pretty much forever. But that's it. I'm not paying $130k for another slow-ass computer without even a damn video card for a console. And I don't need to: Ingo Molnar's realtime patches are coming along, and my quad Opteron box wipes the floor with the Origin and cost, oh yeah: $19,992 including shipping, and $7k of that is pimpin' SCSI disks.
Yay for your death! Ding dong, bitches.
And your information comes from...?
Some contracts are "cost plus", where there is a guaranteed level of profit: all extra costs are eaten by the government. This makes okay sense, since usually the government jerks companies around (oh, now we want it in purple and to cure cancer...). This guaranteed fee usually results in profits of less than 10% (the government is looking to bump this somewhat since they can't find takers for some of the work they want done). Take a look at commercial companies and see what kinds of margins they make.
Other contracts are "fixed fee": build it to the defined price or you take it in the keister. Government still jerks you around.
As far as competition goes: it's there for some things, and not for others. But how many people can you expect to make fighters when you only ask for a new one every 10-20 years? Between F-22 and JSF is about 15 years or so. Lockheed Martin and Boeing have had to combine their satellite launch operations because they can't make money off military launches.
And EADS (European Aerospace something xyz) is now competing in the U.S.
Frankly, I'm tired of hearing about this "subsidized this, blah that" from people who know jackshit about how the industry works. The government can sign contracts, then just as quickly obviate them with no repercussions (see the F-22 contract, or DDX (the next gen Navy destroyer), or just about every other damn thing). The government defines who things can be sold to and under what conditions, and requires a fuck ton of paperwork to sell even things like radios (ITAR regulations).
Do your homework before you spout off. There's a difference between AirBus saying "gimme money" and the EU saying "how much?" and Boeing fighting off LockMart, Northrop Grumman, Raytheon, EADS, ATK, etc. for a chance at a contract. I can't speak to the subsidies from various states (Washington, etc.), but I notice Alabama is offering substantial financial incentives to AirBus to built a plant in Mobile, so it looks like that assistance is basically equivalent.
Trust me, I know how these things work.
No. The defense is metal garbage can lids.
This is relatively authoritative.