Sort of. They (which means pretty much all big content producers) would rather have a market in which they controlled 100% of the available product, rather than a market 10 times bigger in which they only controlled 50% of it. Maybe their profits aren't so big, but then neither are their risks or the amount of effort they have to put into maintaining their hold. It's not in the public interest, and our governments should not be helping them maintain their dominance, but it's predictable that the companies themselves do anything in their power to maintain it.
Why can't you put it back in the ocean? If you didn't, you'd be slowly desalinating the Pacific! (The water you pull out eventually makes it back there, so if the salt doesn't...)
Hey, you've already bought their kit, what do they care anymore? Sony support has been getting a lot worse recently - check out the current state of the club-vaio and related websites and you get the distinct impression they're designed deliberately to get you to go away rather than answer any questions you might have.
Phoenix seems to deal very badly with focus being in different places. I find often that scroll-wheel functionality, or keyboard shortcuts (Ctrl-T, Ctrl-N ) just don't work, but if you click on a part of the page's normal HTML background (not a form control, link, embedding or anything else) it all starts working again. Opera has much the same problem but slightly worse.
No, pick any one. Security is hardly ever easy, and for a project of any size efficiency will take some thought too, and at least a small hit if you're trying to be secure. And if the one you pick is Secure, then you *will* make mistakes in that, so you might have to settle for none of the above.
Please enlighten me how you propose to unplug the sun.
By persuading it to sign an international treaty setting a reduced energy output level phased in over the next 10 years. Just like we did with America.
Re:sigh .. there is no such thing as "macroevoluti
on
Shapes of Time
·
· Score: 1
I wonder if there have been any "good" mutations observed (meaning mutations that don't kill an organism).
The vast majority of mutations do not kill an organism, or stop it from reproducing (which is more important in evolutionary terms.) The susceptibility of different genes to mutation is quite variable, and those genes which are 'core' to the viability of an organism (say, genes coding for DNA transcription itself) have developed to be very well protected against change.
A more restrictive definition of "good mutation" might be one which gives the organism and its own offspring a competitive boost. Unfortunately this depends on the very specific selective pressures the organism finds itself under at its own time and place in the ecosystem as a whole. Observation in the lab would be a very artificial ecosystem indeed, and in the field it would be very difficult to observe measurable effects.
I've also heard that mutations observed today are caused by the deteroration of genetic information. Anyone care to refute that?
Mutation is certainly the result of replication failing to produce an absolutely faithful copy of the original DNA strands, however whether you call that deterioration or not is rather subjective! A certain rate of "transcription errors" is built in (by natural selection) because it allows for the generation of genetic diversity that natural selection can work with.
So if they've got any one segment (and they can store them up for offline processing, and presumably they have to get the very first key to the other side before they can do any encryption at all), they know the key to decrypt the next segment, therefore the one after... Nice try.
The way this is done in practice is to use a dedicated "key exchange" mechanism such as Diffie-Hellman. This allows two parties to agree on a (pretty much has to be a symetric) key to use such that knowing all the communication between them won't let you figure out the key they've agreed to. (It has other nice features such as neither side being able to force the other to agree to any particular pre-chosen key.)
If this key is fed into a decent cipher (128 bit keys or more is easily enough for now) then you only really need the one key, though if you're really paranoid you could use hash-chaining of the old key to produce new keys for subsequent segments, since that's faster than DH and wouldn't need two-way communication in the keygen.
I guess rms should target God as the largest producer of closed-source software in the Universe?
His work is fully open-source (very much in the GPL spirit - you're free to make modifications and redistribute them), but unfortunately the source language is rather complicated and we're only just starting to really understand it. Just wait until the O'Reilly book comes out!
Making an event handler reentrant is a piece of cake.
Sometimes. Usually it requires more thought. There are many environments where the lock strategy you suggest would be a guaranteed deadlock. Other common strategies include: forget the second event because you know the first will drain the whole work-queue by the time it exits; set a flag which the first will check on exit and re-run itself if necessary; post a message back to the head of the application message queue to do the work later; make the handler extract tasks from the work queue atomically so it really doesn't matter if they two instances "simultaneously"; make the handlers stubs that simply trigger later execution of the work function in a way that can't happen more than once at a time; panic because the rest of the system was supposed to prevent this happening at all costs.
Without examining the system in detail it's certainly not safe to assert that it's a "piece of cake"!
If you detect a scan, why would you want to send back MORE syn/acks? If anything, I'd want to start playing dead by sending RSTs at you.
There are several things you can do to discourage an attempted scan. One, as you say, you can send an RST to force the other side to abandon the connection. Another is to just drop incoming packets on the floor - the other end won't know whether you're doing this or the network between is flaky, so they have potentially long time-out delay before they can move on to the next port/machine. (You can scan multiple targets in parallel of course, but if they all black-hole you you're still going to get wedged at some point.)
A more involved one is to tar-pit (or teergrube) them. In SMTP this is used against spammers, but the principal is quite general. The mechanism you would use in this case is to ack their syn, then drop completely any further packets from them. If you drop or reset immediately then they can wait for a fairly short timeout period and/or close their end of the connection immediately with no penalty to them. If they are using anything remotely like a normal TCP implementation then by acking them you are forcing them to make a more long-lasting entry in their socket table, and wait for a much longer timeout period (longer than SYN-SENT retransmission timeouts) causing their socket table to fill up, therefore slowing them down as they try to scan yours or other peoples' networks.
Since you're only ever sending valid protocol responses or ignoring them, rather then actively attacking them, it is hoped that they will decide to give up rather than DOS you.
In W2k and later, the entire network stack is completely pluggable. You can insert any layer you want to that sits between NDIS and a protocol driver, and you can create other layers as required. I'd be very surprised if they couldn't do everything required with windows exactly as it sits today.
Sure, you can plug anything you want into the stack. But then your otherwise loyal citizens could 'accidentally' unplug those components, thereby potentially exposing themselves to extremely damaging subversive ideas.
I suspect what they would rather do is *remove* the pluggability from the stack and hardwire their constraints so deep into the OS that a domestic build couldn't possibly be used to access any unapproved information.
And yeah, I wouldn't expect them to be getting their PR flaks to do the coding either.
Just today Oracle is taking some real heat for dropping support on old applications software. The reason is simple. Corporations want to upgrade when they want to upgrade and not at any other time.
OTOH it's unrealistic to expect any business to support an old product indefinitely - the cost of providing support rises continuously, whereas the revenue from new product will fall unless customers can be persuaded to upgrade. In a commercial environment a large part of that burdon has to be born by customers and third party support shops, or the vendor dies and everybody is worse off. (As much damage as Microsoft have done and will no doubt continue to do, a Microsoft forced to be competetive would be better than no Microsoft at all.)
I still don't think I'd pay $3k for a 2gig replacement.
Especially when you can buy 4x512Mb CompactFlash cards, 2x2-port PCI->PCMCIA cards, and 4xCF->PCMCIA adapters for a third of the price. Less if you go for 256Mb CF cards and 4x2-port adapters (if you have PCI slots to burn).
The academies, Dr. Colglazier said, "have recognized that it makes sense to restrict public access to some areas of sensitive information that is unclassified," like information about national infrastructures that could be disrupted by terrorist attacks.
"But the concern," he said, "is that there should be clear guidance on what information would fall into this category."
"We recognize that it makes sense to allow the Government to sodomize us whenever they like, we're just concerned that there's no clear guidance on what types of lubricant they should use.
One final thought, why should you pay the same for insurance as someone who is a walking time bomb if you have no increased genetic risks?
Because that's what insurance is. By smearing the probabilities over a larger population you make things more fair in general for a larger number of people. The logical endpoint of your argument is that everyone pays for exactly what they, individually, need (insurance companies don't give you money, they only redistribute it, taking their percentage.) In other words, insurance as such no longer exists.
When the top judges are appointed by politicians, of course there's a limit to how divided they can be. However in this case the courts are working exactly as they should. A court can only act when there is a conflict between two parties - it is an absolute requirement that someone be prepared and capable of standing up and challenging wrongdoers. If the current administration lacks the will to do so, the courts are not allowed to make pre-emptive judgements against any party.
What about all the rental stores that buy only a few copies and rent them out to people.
Firstly, the cost of any media (whether VHS, DVD, XBox or whatever) for rental is typically several times the normal retail price for individual home use. Secondly, end consumer purchases are relatively rare - as and when you see something you like and can afford it. To stay in business a rental store has to keep up to date with the latest releases - if it can't offer the newest films or games people would rapidly stop going there. That means it has to keep buying more and more original product at frequent intervals. The rental market is *good* for content producers.
Slashdot Mirror
Sort of. They (which means pretty much all big content producers) would rather have a market in which they controlled 100% of the available product, rather than a market 10 times bigger in which they only controlled 50% of it. Maybe their profits aren't so big, but then neither are their risks or the amount of effort they have to put into maintaining their hold. It's not in the public interest, and our governments should not be helping them maintain their dominance, but it's predictable that the companies themselves do anything in their power to maintain it.
Why can't you put it back in the ocean? If you didn't, you'd be slowly desalinating the Pacific! (The water you pull out eventually makes it back there, so if the salt doesn't...)
Nope, the mode is 90. The median is 135.
If you were to post an apology (or even a normal comment) as yourself to this story, your mod-in-error would be automatically discarded?
Hey, you've already bought their kit, what do they care anymore? Sony support has been getting a lot worse recently - check out the current state of the club-vaio and related websites and you get the distinct impression they're designed deliberately to get you to go away rather than answer any questions you might have.
Phoenix seems to deal very badly with focus being in different places. I find often that scroll-wheel functionality, or keyboard shortcuts (Ctrl-T, Ctrl-N ) just don't work, but if you click on a part of the page's normal HTML background (not a form control, link, embedding or anything else) it all starts working again. Opera has much the same problem but slightly worse.
There appear to be a great many people crapping over the internet too...
No, pick any one. Security is hardly ever easy, and for a project of any size efficiency will take some thought too, and at least a small hit if you're trying to be secure. And if the one you pick is Secure, then you *will* make mistakes in that, so you might have to settle for none of the above.
(But that doesn't mean you shouldn't try.)
By persuading it to sign an international treaty setting a reduced energy output level phased in over the next 10 years. Just like we did with America.
The vast majority of mutations do not kill an organism, or stop it from reproducing (which is more important in evolutionary terms.) The susceptibility of different genes to mutation is quite variable, and those genes which are 'core' to the viability of an organism (say, genes coding for DNA transcription itself) have developed to be very well protected against change.
A more restrictive definition of "good mutation" might be one which gives the organism and its own offspring a competitive boost. Unfortunately this depends on the very specific selective pressures the organism finds itself under at its own time and place in the ecosystem as a whole. Observation in the lab would be a very artificial ecosystem indeed, and in the field it would be very difficult to observe measurable effects.
Mutation is certainly the result of replication failing to produce an absolutely faithful copy of the original DNA strands, however whether you call that deterioration or not is rather subjective! A certain rate of "transcription errors" is built in (by natural selection) because it allows for the generation of genetic diversity that natural selection can work with.
So if they've got any one segment (and they can store them up for offline processing, and presumably they have to get the very first key to the other side before they can do any encryption at all), they know the key to decrypt the next segment, therefore the one after... Nice try.
The way this is done in practice is to use a dedicated "key exchange" mechanism such as Diffie-Hellman. This allows two parties to agree on a (pretty much has to be a symetric) key to use such that knowing all the communication between them won't let you figure out the key they've agreed to. (It has other nice features such as neither side being able to force the other to agree to any particular pre-chosen key.)
If this key is fed into a decent cipher (128 bit keys or more is easily enough for now) then you only really need the one key, though if you're really paranoid you could use hash-chaining of the old key to produce new keys for subsequent segments, since that's faster than DH and wouldn't need two-way communication in the keygen.
His work is fully open-source (very much in the GPL spirit - you're free to make modifications and redistribute them), but unfortunately the source language is rather complicated and we're only just starting to really understand it. Just wait until the O'Reilly book comes out!
This would be the Henry Ford definition of choice then? "You can choose any supplier you like, so long as it's us."
Except Lloyds.
Sometimes. Usually it requires more thought. There are many environments where the lock strategy you suggest would be a guaranteed deadlock. Other common strategies include: forget the second event because you know the first will drain the whole work-queue by the time it exits; set a flag which the first will check on exit and re-run itself if necessary; post a message back to the head of the application message queue to do the work later; make the handler extract tasks from the work queue atomically so it really doesn't matter if they two instances "simultaneously"; make the handlers stubs that simply trigger later execution of the work function in a way that can't happen more than once at a time; panic because the rest of the system was supposed to prevent this happening at all costs.
Without examining the system in detail it's certainly not safe to assert that it's a "piece of cake"!
There are several things you can do to discourage an attempted scan. One, as you say, you can send an RST to force the other side to abandon the connection. Another is to just drop incoming packets on the floor - the other end won't know whether you're doing this or the network between is flaky, so they have potentially long time-out delay before they can move on to the next port/machine. (You can scan multiple targets in parallel of course, but if they all black-hole you you're still going to get wedged at some point.)
A more involved one is to tar-pit (or teergrube) them. In SMTP this is used against spammers, but the principal is quite general. The mechanism you would use in this case is to ack their syn, then drop completely any further packets from them. If you drop or reset immediately then they can wait for a fairly short timeout period and/or close their end of the connection immediately with no penalty to them. If they are using anything remotely like a normal TCP implementation then by acking them you are forcing them to make a more long-lasting entry in their socket table, and wait for a much longer timeout period (longer than SYN-SENT retransmission timeouts) causing their socket table to fill up, therefore slowing them down as they try to scan yours or other peoples' networks.
Since you're only ever sending valid protocol responses or ignoring them, rather then actively attacking them, it is hoped that they will decide to give up rather than DOS you.
Sure, you can plug anything you want into the stack. But then your otherwise loyal citizens could 'accidentally' unplug those components, thereby potentially exposing themselves to extremely damaging subversive ideas.
I suspect what they would rather do is *remove* the pluggability from the stack and hardwire their constraints so deep into the OS that a domestic build couldn't possibly be used to access any unapproved information.
And yeah, I wouldn't expect them to be getting their PR flaks to do the coding either.
OTOH it's unrealistic to expect any business to support an old product indefinitely - the cost of providing support rises continuously, whereas the revenue from new product will fall unless customers can be persuaded to upgrade. In a commercial environment a large part of that burdon has to be born by customers and third party support shops, or the vendor dies and everybody is worse off. (As much damage as Microsoft have done and will no doubt continue to do, a Microsoft forced to be competetive would be better than no Microsoft at all.)
Especially when you can buy 4x512Mb CompactFlash cards, 2x2-port PCI->PCMCIA cards, and 4xCF->PCMCIA adapters for a third of the price. Less if you go for 256Mb CF cards and 4x2-port adapters (if you have PCI slots to burn).
Easy - you just increase your count by 33,093,474,372,000.
"We recognize that it makes sense to allow the Government to sodomize us whenever they like, we're just concerned that there's no clear guidance on what types of lubricant they should use.
Because that's what insurance is. By smearing the probabilities over a larger population you make things more fair in general for a larger number of people. The logical endpoint of your argument is that everyone pays for exactly what they, individually, need (insurance companies don't give you money, they only redistribute it, taking their percentage.) In other words, insurance as such no longer exists.
When the top judges are appointed by politicians, of course there's a limit to how divided they can be. However in this case the courts are working exactly as they should. A court can only act when there is a conflict between two parties - it is an absolute requirement that someone be prepared and capable of standing up and challenging wrongdoers. If the current administration lacks the will to do so, the courts are not allowed to make pre-emptive judgements against any party.
Ban memcpy() now! It's an infringement device!
(ITYM ROM or BIOS.)
Firstly, the cost of any media (whether VHS, DVD, XBox or whatever) for rental is typically several times the normal retail price for individual home use. Secondly, end consumer purchases are relatively rare - as and when you see something you like and can afford it. To stay in business a rental store has to keep up to date with the latest releases - if it can't offer the newest films or games people would rapidly stop going there. That means it has to keep buying more and more original product at frequent intervals. The rental market is *good* for content producers.