This isn't an article about building a storage server. It's an article about building a kewl PC that's overloaded with disks. The author implies that two disks will run off one cable, which is death to performance in an IDE storage system. But that might not be a problem, since:
Something else to consider is that a RAID array would use more power than our current setup because the drives are run simultaneously.
So, only one drive will be accessed at once? Sounds like a single-user PC, not a storage server. And then:
There is an SPDIF header on board, but the cable is not included. Inaddition, more and more manufacturers are providing comprehensive software bundles and high performance accessories, such as rounded hard drive cables. These are all accessories that we would need to buy anyways.
and:
There were a few things we didnt like though, for one, it didnt include any front USB, firewire, or audio ports. These ports really do improve a usability of a system especially when the back of the case is hard to reach.
Sound like a storage server?
The article doesn't start by defining what it wants to accomplish. 1 TB of capacity, OK, but what levels of data transfer and downtime? There are mixed signals:
In our case, we dont even need the bandwidth of the SCSI drives quantity rather than blistering speed was important.
... and later:
This allows for a full gigabit of Ethernet bandwidth without taking away from PCI bandwidth, with a maximum rate of 266 MB/s which is over 2000 megabits/sec, enough for full-duplex gigabit. We see this as an essential feature for a storage server.
Since the author doesn't know what level of performance he's aiming for, he has no basis for choosing the amount of RAM, CPU, and disk formatting strategy.
So it should be important to note that simply turning on the RAID option does not guarantee the safety of your data. Our solution is to run the drives in a standard configuration.
IOW, RAID isn't perfect so I didn't use it. What's a "standard configuration"? Separate mount points for all disks? That can work, but ripples through to the applications.
This is also the reason Linux was not a good choice for our system -- it doesn't make sense to put XFS/ext3/ReiserFS drives into a USB2.0/Firewire external box.
Why? Since the three FS's mentioned are journalling, my first thought is that you're concerned about saturating the link bandwidth with journalling data. Is that it? Tough to analyze without some throughput numbers. If that's it, why not use ext2?
So what OS will you use? Let me guess - Windows XP. Because it has the most games.
A lot of posters are acting outraged, like this is deliberate sabotage in the reactor control computer or something. But really, this is a consumer appliance and the vast majority of consumers won't care. The only question is, how deferential will Belkin be to the angry geeks, who probably weren't buying the product anyway? Will they pull it, or just make soothing noises? My guess is that they'll get a very low conversion rate and then pull it.
The obvious next step is to sell interstitial ads. I wonder how many commercials a normal internet user would put up with? That leads to the thought of modified routers which avoid playing the ads, which leads to DMCA-letters - but based on what? Could the advertisers claim that the router controlled access to protected content (the ads)?
In five years, companies like Belkin might be getting 80% of their revenue from advertisers. And think of the market data they can sell - they can intercept all your web and email traffic. If you send email to a friend about selling your old car, you might immediately face a full-page, blocking ad for a charity which would like you to donate the car. And the amount of the tax deduction (tailored to your car and tax bracket) could be in giant blinking nubers in the middle.
I don't think you'll fend off such a future by accusing Belkin of "violating the HTTP RFC" or something, because those words mean nothing to normal people. The idea that a router's responsibilities are sacred and cast in stone may be a good idea, but we haven't managed to explain it to normal people.
First, I was thinking of the books, not the movie[s]. I should have been more explicit, but when I think of the Lord of the Rings, the existence of the movies doesn't even cross my mind.
When I say that these books are well thought out, I am not claiming that they are full of subtexts, metaphors, commentaries, etc. Tolkien, and most Tolkien fans, explicitly disclaim such subtextual content. (Although that disclaimer is somewhat questionable.) What I mean is that the trilogy is internally consistent. Such consistency is hard to achieve, and rarely found in fiction. Tolkien achieved it by many years of thinking about Middle Earth before he ever considered writing fiction in that setting. The Silmarillion (which I find unreadable) is the back-story to the Ring Trilogy. If you wanted to write a Matrix-style narrative with a level of cohesion comparable to the Ring Trilogy's, you'd need to first write something like the Silmarillion - a kind of factual history of the Matrix. I'm not sure how much of that effort could shine through in a movie.
I saw the first LOTR movie. I can't evaluate it fairly because I am familiar with the books. Inevitably, it lacked some of the depth and sensitivity of the books.
I don't know how much the movie's ending will be warped by commercial pressures, but the books' ending is not triumphalist, but bittersweet.
Given that voters are never removed from the registration roles in California (I believe this is true even after death, but the cases I've heard of may have been glitches), I'm suspecting there's a lot more vote fraud caused by folks voting who aren't legally entitled to than there is by subversive maniuplation of machines in that particular case.
Your implication is chilling. These putrescent cadavers, unmindful of any legislative bar on their activities, boldly sashay into the polling place and drain the very life from our democracy.
Not that anyone will read this. Sometimes you need a lame sequel to realize how lame the whole thing was.
I liked the first movie until the detour into martial arts. The bit where Neo gets the cell phone package was cool. The lameness started to increase rapidly however. The whole karate motif was clearly grafted on to create neat visuals, and can't be supported logically. And the agents - it makes sense that the Matrix would have scripts (personify them if you must) scanning for and killing "bad" processes. Lots of web hosts do that - kill any process running too long or consuming too much resource. It makes no sense that these scripts would be limited by any of the rules that apply to "people" in the Matrix. If this were real, the agents would be invisible, would not need to travel from place to place, and would kill by simply deleting. But in order to drag this film down to the action-movie level, we have to have guys with dark glasses and big guns running around.
So I emerged from the first movie with mixed feelings. The second movie was much simpler - it just flat out sucked. That got me thinking that the good parts in the first movie were pretty small. Apparently the third sucks even worse.
And reading the fans quarrel over the fine points of Matrix philosophy and universe is embarassing. It's not embarassing to argue like that over Tolkien, because he really put some thought into his works. But this batch of movies is so obviously thrown together with little thought that it's incongruous to debate the half-baked details.
Savvy Strauss won't fall for those phony free deals. He knows that sooner or later you have to pay the piper. How ironic, then, that these offers float seductively below his essay:
Free Custom Content Management Demo!: See your site being managed by a CMS
Free Evaluation!: Engage and Support Your Faculty and Students Online!
Motion Tablet PCs: improve the classroom experience, no compromises!
Mitsubishi Projectors: Get a FREE Replacement Lamp (A $499 Value)
I'm not sure you understood the question. Windows personal firewalls can be configured to only let certain apps access the network. Let's say, only IE. So when ActiveSpyWare++ Gold decides to phone home, it gets blocked. I don't see how to do that with an external Linux firewall.
The patent covers a way of moving user settings from an old computer to a new computer. So it's a logical fit for a company that upgrades desktop PC's in an organization.
Once again, slashdot posts an utterly wrong synopsis of a patent.
...covering the Internet installation of any software or settings on new computers.
Wrong. Since very few slashdot readers can be bothered to actually read the patent before complaining about it, it covers this process:
Upload user settings from an old computer to a server.
Download those settings to a new computer.
This little game of outright lying about the content of a patent and then joining in a chorus of ignorant bleating about the awful, awful patent system accomplishes nothing. I hope that no legislator or decision-maker ever reads this drivel, as he'll be convinced that patent reform is championed only by cretins.
Sounds like popfile would work for you. Unfortunately, as far as I can tell it must still download the spam messages before discarding them. At least that can occur silently in the background.
First, how do you know the sender is "using a Road Runner account"? Are you going off the From: header? That's usually forged. Look at the Received headers, and see if the last server before yours is constant. If so, you have an easy criterion for blocking.
Second, you didn't specify how you're getting your mail. Do you run your own mail server? If so, add an ipchains/iptables rule to block the offending IP. If you're getting your mail via POP/IMAP from someone else's server, I don't see how you can do server-side blocking.
I think it's tragic that we buy devices that are inherently programmable, and yet the programmability is sealed off from us. When the device runs Linux, it's ironic as well.
Remember Richard Stallman and the printer with proprietary interface codes? Some future Richard Stallman will be in the exact same position, and the printer will be running Linux.
I thought that J2ME meant the same code could run on different mobile devices. A developer working on that stuff told me otherwise. The code he's writing is full of device-specific hooks. "So why Java?" "It's a marketing gimmick."
Opening up the mobile platforms would have two sets of benefits. First, it would benefit the techies who could automate certain repetitive work. Maybe you'd program your phone to use its GPS or signal strength indicators to know when you're driving. If so, it would answer some originating numbers with a recording.
Second, it would allow the best software, rather than the incumbent software, to win with consumers. Imagine how stagnant the PC world would be if IBM had locked down the PC platform to only run IBM's code. And yet that is what many mobile platform owners are doing - even requiring code signing!
I think that's cheating. You're assuming the existence of a "traffic network" that's open and standards-based enough to piggyback a new kind of message. I'm pretty sure this 3m system is a self-contained system that can be sold to governments with various kinds of traffic hardware, whether or not they have data links.
The funny thing is, the RFC's are written in plain text (or nroff translated to plain text) and communicate some pretty substantial ideas. They use ascii art where necessary. The more I see of the ever-changing world of rich documents, the more I think everything should be plain ascii. It has more chance of being readable 50 years from now.
When national security is at stake you don't hand any advantages to the enemy. Military cryptosystems are kept secret for lots of good reasons. It doesn't mean they're vulnerable. The NSA's ciphers probably receive at least as good peer review within the huge agency as any public cipher receives from academic peers.
John Walker supplied the Soviets with keylists for the KW-7 crypto machine, but until they got a KW-7 (via the North Korean capture of the Pueblo) they couldn't use those keys. I think the NSA/military follows this rule: make the algorithm strong, then treat it as if it were weak. Defense in depth.
I think you're wrong. I think you selectively noticed people who drove more aggressively than you and didn't get far. Remember, the motorists you saw may in fact achieve better trip times than you on the average. They probably experiment more and learn more.
Weaving only works in some kinds of traffic. It can be successful in moderately heavy traffic if the driver is skilled at timing the relative shifts of the lanes. But to maintain that skill, the driver has to experiment, and sometimes fail.
I don't drive very aggressively any more, both because I'm older and because such driving is not really accepted in California.
Usually it's the impatient people that create traffic in the first place. The more that people obey speed limits the better the timing of intersections gets.
I don't get it. You seem to be saying that Fast Frank races ahead and skids to a stop at the red light. Slow Sam glides up behind him as the light turns green. How is Fast Frank "causing traffic"? Isn't traffic caused by the volume of vehicles on the road?
I don't know how much you can blame the motorist. You're sending a pretty mixed message by coming to a stop when he's expecting you to go. I understand your rationale for doing so, but I can certainly see how it's confusing for the other motorist.
A friend of mine was driving in a grocery store parking lot. He waited for some pedestrians to cross in front of the vehicle, and didn't understand why they refused to. He didn't seem to notice that he was slowly rolling forward. A similar case of mixed messages.
Change a light with lots of people around and you just pissed off everyone in the opposite flow of traffic.
Why? The light's going to change at some point; nobody was expecting it to stay green forever. You could be right if I see the light turn green a block ahead of me, then turn yellow almost immediately, then red, and a lone car whizzes by. But normally when a light turns red I don't start wondering which car caused it.
Good point. Sometimes there's a tradeoff between aesthetics and practicality. The HMMV is incredibly cool, and incredibly impractical for anything non-military. The H2 is incredibly pathetic - nothing worse than a phony version of a something cool.
The fact that this was modded down points to a problem with slashdot. I think the offtopic mod should be reserved for material that seriously disrupts the conversation. And likewise, 'troll' should be used for page-widening posts and the like, not for subtle sarcasm. It bothers me that someone took a few minutes to post something coherent and interesting and got modded down so it sinks below tons of 5-second posts saying "lol linux rules omg" or something.
I see a few problems. First, the bitrate here is probably pretty low. I read somewhere that they're using 40-60 hz strobe flashes. I don't know how many milliseconds of guarranteed exposure there are. Second, the receiver currently doesn't transmit. So this mod adds a transmitter to the receiver, and a receiver to the transmitter. Big increase both in cost and in probability of not working. For this application, a roling code or time-based code is a way better fit. These are both authentication techniques used when there is no channel back from the verifier to the prover.
This technique is relatively simple. It's big fault is a shared secret key.
It could be a problem. What if someone breaks open a traffic control box and steals the device? What if someone breaks open the box and nothing is missing? Do you assume that the key is compromised? If the verifier module breaks, how do we destroy it? Will all the maintainers know not to toss it in the dumpster?
Maybe the best thing is to use time-windowed keys, as you suggested, and have someone physically open the box and rekey at least once a year.
Well, police departments, or rather their vendors, already face these issues for encrypted radios. Typically a fixed key is used for a large group of radios for a period of time, loaded via a key loader box. Some systems allow OTAR - over the air rekey, where the new key is encrypted with (I think) the old key and broadcast.
This problem is a little harder because an attacker could steal either the traffic light equipment or the vehicle equipment. The traffic light equipment is probably more numerous and vulnerable. Public Key crypto would be very useful here, because then stealing the receiver doesn't help you impersonate the transmitter. Unfortunately, it probably needs too many bits for the IR strobe carrier. So unless we "cheat" with a supplemental data channel or something, there's no good solution.
Although - you could have an operational key (symmetric) and a rekey key (public key). The rekey box is guarded like the crown jewels. Once a month, or if the operational key is known to be compromised, you bolt the rekey box to the roof of a police car and drive slowly around town, pausing 60 seconds or so at each light. Enough time for the slow public-key transmission. Then you rekey all the vehicle units off the rekey box. Could work.
On castro, that was wrong and it was an act of war (not terrorism).
My point is not whether it was right or wrong, but that it shows the ambiguity in your definition. Consider the criteria:
The unlawful use or threatened use of force or violence... Was it unlawful? I assume that under Cuban law it is unlawful to assassinate the head of state. But under US law, it might be lawful to commit such an action in service of your country. Which law applies?
...by a person or an organized group... Yes, the CIA is an organized group. Even if it wasn't, the actual killing must be committed by a person, making this clause redundant.
...against people or property...
Yes, Castro is a person. Again, this clause is redundant - what targets was this clause meant to exclude?
...with the intention of intimidating or coercing societies or governments, often for ideological or political reasons. Yes, the action was intended to help topple the Communist government, allowing the return of the exiles and a more US-friendly government.
So that action is or isn't terrorism depending solely on whether it was "unlawful". This becomes a subjective criterion. Each "side" will claim that the other side's actions are unlawful.
...an act of war (not terrorism).
I think you're saying that the two are mutually exclusive. But your definition did not make any provision for this. Maybe you want to change a person or an organized group to a person or an organized group other than military personnel. But if you think about it, that definition is still full of holes. Special Forces? Special Forces without uniforms or dogtags? Resistance fighters in occupied territory?
The same points apply to Rachel Corrie. Whether she was right, wrong, smart, stupid, intentionally or accidentally killed, her killing could meet your definition of terrorism, or could fail to meet it, depending entirely on whether her killing was unlawful. That makes the definition essentially useless.
I am trying to persuade you that properly defining terrorism is a very hard task.
Slashdot Mirror
So, only one drive will be accessed at once? Sounds like a single-user PC, not a storage server.
And then:
and: Sound like a storage server?
The article doesn't start by defining what it wants to accomplish. 1 TB of capacity, OK, but what levels of data transfer and downtime? There are mixed signals:
Since the author doesn't know what level of performance he's aiming for, he has no basis for choosing the amount of RAM, CPU, and disk formatting strategy.
IOW, RAID isn't perfect so I didn't use it. What's a "standard configuration"? Separate mount points for all disks? That can work, but ripples through to the applications.
Why? Since the three FS's mentioned are journalling, my first thought is that you're concerned about saturating the link bandwidth with journalling data. Is that it? Tough to analyze without some throughput numbers. If that's it, why not use ext2?
So what OS will you use? Let me guess - Windows XP. Because it has the most games.
A lot of posters are acting outraged, like this is deliberate sabotage in the reactor control computer or something. But really, this is a consumer appliance and the vast majority of consumers won't care. The only question is, how deferential will Belkin be to the angry geeks, who probably weren't buying the product anyway? Will they pull it, or just make soothing noises? My guess is that they'll get a very low conversion rate and then pull it.
The obvious next step is to sell interstitial ads. I wonder how many commercials a normal internet user would put up with? That leads to the thought of modified routers which avoid playing the ads, which leads to DMCA-letters - but based on what? Could the advertisers claim that the router controlled access to protected content (the ads)?
In five years, companies like Belkin might be getting 80% of their revenue from advertisers. And think of the market data they can sell - they can intercept all your web and email traffic. If you send email to a friend about selling your old car, you might immediately face a full-page, blocking ad for a charity which would like you to donate the car. And the amount of the tax deduction (tailored to your car and tax bracket) could be in giant blinking nubers in the middle.
I don't think you'll fend off such a future by accusing Belkin of "violating the HTTP RFC" or something, because those words mean nothing to normal people. The idea that a router's responsibilities are sacred and cast in stone may be a good idea, but we haven't managed to explain it to normal people.
Clearly, you're into the whole brevity thing. Glad to hear the Almighty intervened and you didn't drown after all.
First, I was thinking of the books, not the movie[s]. I should have been more explicit, but when I think of the Lord of the Rings, the existence of the movies doesn't even cross my mind.
When I say that these books are well thought out, I am not claiming that they are full of subtexts, metaphors, commentaries, etc. Tolkien, and most Tolkien fans, explicitly disclaim such subtextual content. (Although that disclaimer is somewhat questionable.) What I mean is that the trilogy is internally consistent.
Such consistency is hard to achieve, and rarely found in fiction. Tolkien achieved it by many years of thinking about Middle Earth before he ever considered writing fiction in that setting. The Silmarillion (which I find unreadable) is the back-story to the Ring Trilogy. If you wanted to write a Matrix-style narrative with a level of cohesion comparable to the Ring Trilogy's, you'd need to first write something like the Silmarillion - a kind of factual history of the Matrix. I'm not sure how much of that effort could shine through in a movie.
I saw the first LOTR movie. I can't evaluate it fairly because I am familiar with the books. Inevitably, it lacked some of the depth and sensitivity of the books.
I don't know how much the movie's ending will be warped by commercial pressures, but the books' ending is not triumphalist, but bittersweet.
Your implication is chilling. These putrescent cadavers, unmindful of any legislative bar on their activities, boldly sashay into the polling place and drain the very life from our democracy.
Not that anyone will read this. Sometimes you need a lame sequel to realize how lame the whole thing was.
I liked the first movie until the detour into martial arts. The bit where Neo gets the cell phone package was cool. The lameness started to increase rapidly however. The whole karate motif was clearly grafted on to create neat visuals, and can't be supported logically. And the agents - it makes sense that the Matrix would have scripts (personify them if you must) scanning for and killing "bad" processes. Lots of web hosts do that - kill any process running too long or consuming too much resource. It makes no sense that these scripts would be limited by any of the rules that apply to "people" in the Matrix. If this were real, the agents would be invisible, would not need to travel from place to place, and would kill by simply deleting. But in order to drag this film down to the action-movie level, we have to have guys with dark glasses and big guns running around.
So I emerged from the first movie with mixed feelings. The second movie was much simpler - it just flat out sucked. That got me thinking that the good parts in the first movie were pretty small. Apparently the third sucks even worse.
And reading the fans quarrel over the fine points of Matrix philosophy and universe is embarassing. It's not embarassing to argue like that over Tolkien, because he really put some thought into his works. But this batch of movies is so obviously thrown together with little thought that it's incongruous to debate the half-baked details.
Does a whiff of Nigerian snake oil tickle the nostrils?
I'm not sure you understood the question. Windows personal firewalls can be configured to only let certain apps access the network. Let's say, only IE. So when ActiveSpyWare++ Gold decides to phone home, it gets blocked. I don't see how to do that with an external Linux firewall.
The patent covers a way of moving user settings from an old computer to a new computer. So it's a logical fit for a company that upgrades desktop PC's in an organization.
Wrong. Since very few slashdot readers can be bothered to actually read the patent before complaining about it, it covers this process:
This little game of outright lying about the content of a patent and then joining in a chorus of ignorant bleating about the awful, awful patent system accomplishes nothing. I hope that no legislator or decision-maker ever reads this drivel, as he'll be convinced that patent reform is championed only by cretins.
Sounds like popfile would work for you. Unfortunately, as far as I can tell it must still download the spam messages before discarding them. At least that can occur silently in the background.
First, how do you know the sender is "using a Road Runner account"? Are you going off the From: header? That's usually forged. Look at the Received headers, and see if the last server before yours is constant. If so, you have an easy criterion for blocking.
Second, you didn't specify how you're getting your mail. Do you run your own mail server? If so, add an ipchains/iptables rule to block the offending IP. If you're getting your mail via POP/IMAP from someone else's server, I don't see how you can do server-side blocking.
I think it's tragic that we buy devices that are inherently programmable, and yet the programmability is sealed off from us. When the device runs Linux, it's ironic as well.
Remember Richard Stallman and the printer with proprietary interface codes? Some future Richard Stallman will be in the exact same position, and the printer will be running Linux.
I thought that J2ME meant the same code could run on different mobile devices. A developer working on that stuff told me otherwise. The code he's writing is full of device-specific hooks. "So why Java?" "It's a marketing gimmick."
Opening up the mobile platforms would have two sets of benefits. First, it would benefit the techies who could automate certain repetitive work. Maybe you'd program your phone to use its GPS or signal strength indicators to know when you're driving. If so, it would answer some originating numbers with a recording.
Second, it would allow the best software, rather than the incumbent software, to win with consumers. Imagine how stagnant the PC world would be if IBM had locked down the PC platform to only run IBM's code. And yet that is what many mobile platform owners are doing - even requiring code signing!
Good post. I think posts like this are the biggest reason to read slashdot.
I think that's cheating. You're assuming the existence of a "traffic network" that's open and standards-based enough to piggyback a new kind of message. I'm pretty sure this 3m system is a self-contained system that can be sold to governments with various kinds of traffic hardware, whether or not they have data links.
The funny thing is, the RFC's are written in plain text (or nroff translated to plain text) and communicate some pretty substantial ideas. They use ascii art where necessary. The more I see of the ever-changing world of rich documents, the more I think everything should be plain ascii. It has more chance of being readable 50 years from now.
When national security is at stake you don't hand any advantages to the enemy. Military cryptosystems are kept secret for lots of good reasons. It doesn't mean they're vulnerable. The NSA's ciphers probably receive at least as good peer review within the huge agency as any public cipher receives from academic peers.
John Walker supplied the Soviets with keylists for the KW-7 crypto machine, but until they got a KW-7 (via the North Korean capture of the Pueblo) they couldn't use those keys. I think the NSA/military follows this rule: make the algorithm strong, then treat it as if it were weak. Defense in depth.
Ah, was wondering if you had one-way streets. Manhattan also has this system - it's great.
Weaving only works in some kinds of traffic. It can be successful in moderately heavy traffic if the driver is skilled at timing the relative shifts of the lanes. But to maintain that skill, the driver has to experiment, and sometimes fail.
I don't drive very aggressively any more, both because I'm older and because such driving is not really accepted in California.
I don't get it. You seem to be saying that Fast Frank races ahead and skids to a stop at the red light. Slow Sam glides up behind him as the light turns green. How is Fast Frank "causing traffic"? Isn't traffic caused by the volume of vehicles on the road?
I don't know how much you can blame the motorist. You're sending a pretty mixed message by coming to a stop when he's expecting you to go. I understand your rationale for doing so, but I can certainly see how it's confusing for the other motorist.
A friend of mine was driving in a grocery store parking lot. He waited for some pedestrians to cross in front of the vehicle, and didn't understand why they refused to. He didn't seem to notice that he was slowly rolling forward. A similar case of mixed messages.
Why? The light's going to change at some point; nobody was expecting it to stay green forever. You could be right if I see the light turn green a block ahead of me, then turn yellow almost immediately, then red, and a lone car whizzes by. But normally when a light turns red I don't start wondering which car caused it.
Good point. Sometimes there's a tradeoff between aesthetics and practicality. The HMMV is incredibly cool, and incredibly impractical for anything non-military. The H2 is incredibly pathetic - nothing worse than a phony version of a something cool.
The fact that this was modded down points to a problem with slashdot. I think the offtopic mod should be reserved for material that seriously disrupts the conversation. And likewise, 'troll' should be used for page-widening posts and the like, not for subtle sarcasm. It bothers me that someone took a few minutes to post something coherent and interesting and got modded down so it sinks below tons of 5-second posts saying "lol linux rules omg" or something.
For this application, a roling code or time-based code is a way better fit. These are both authentication techniques used when there is no channel back from the verifier to the prover.
It could be a problem. What if someone breaks open a traffic control box and steals the device? What if someone breaks open the box and nothing is missing? Do you assume that the key is compromised? If the verifier module breaks, how do we destroy it? Will all the maintainers know not to toss it in the dumpster?
Maybe the best thing is to use time-windowed keys, as you suggested, and have someone physically open the box and rekey at least once a year.
Well, police departments, or rather their vendors, already face these issues for encrypted radios. Typically a fixed key is used for a large group of radios for a period of time, loaded via a key loader box. Some systems allow OTAR - over the air rekey, where the new key is encrypted with (I think) the old key and broadcast.
This problem is a little harder because an attacker could steal either the traffic light equipment or the vehicle equipment. The traffic light equipment is probably more numerous and vulnerable. Public Key crypto would be very useful here, because then stealing the receiver doesn't help you impersonate the transmitter. Unfortunately, it probably needs too many bits for the IR strobe carrier. So unless we "cheat" with a supplemental data channel or something, there's no good solution.
Although - you could have an operational key (symmetric) and a rekey key (public key). The rekey box is guarded like the crown jewels. Once a month, or if the operational key is known to be compromised, you bolt the rekey box to the roof of a police car and drive slowly around town, pausing 60 seconds or so at each light. Enough time for the slow public-key transmission. Then you rekey all the vehicle units off the rekey box. Could work.
My point is not whether it was right or wrong, but that it shows the ambiguity in your definition. Consider the criteria:
So that action is or isn't terrorism depending solely on whether it was "unlawful". This becomes a subjective criterion. Each "side" will claim that the other side's actions are unlawful.
I think you're saying that the two are mutually exclusive. But your definition did not make any provision for this. Maybe you want to change a person or an organized group to a person or an organized group other than military personnel. But if you think about it, that definition is still full of holes. Special Forces? Special Forces without uniforms or dogtags? Resistance fighters in occupied territory?
The same points apply to Rachel Corrie. Whether she was right, wrong, smart, stupid, intentionally or accidentally killed, her killing could meet your definition of terrorism, or could fail to meet it, depending entirely on whether her killing was unlawful. That makes the definition essentially useless.
I am trying to persuade you that properly defining terrorism is a very hard task.