The success of Windows has depended on its nature as a bundle: you pay 100$ (or Dell pay ???$ for you) and get the whole shebang. The licenses from this release pay for development of the new items in the next version of the bundle.
This means that Windows customers expect everything to be included in the bundle that they need. The kind of services that TFA recommends MS sell (20$ a month for virual hard drive etc. like.Mac) cannot be funded from the Windows license fee, unless Windows costs 300$ a license. People expect not to have to pay extra, so it's hard to convince them to do so.
This bundling also affects the lifecycle of the product: 5-6 years between XP and Longhorn is required because they need to do a lot of work! (Could their 're-write' do to them what Netscape's did?). There is so much in the bundle, and MS want to add so much more, that it takes a long time.
This has an impact on EOLing too - MS is still supporting (to some extent) Windows 98(!), 2000, XP. The cost of having a rapid release cycle is supporting many different releases (unless you EOL these releases just as rapidly, cf. Redhat Linux).
Overall, the size of Windows counts against MS in several different ways. It will be difficult for them to move away from it. Perhaps all those companies killed by MS integrating their features into the OS will have the last laugh?
To turn that around: as long as Java remains "under strict control", it will be limited to what Sun can do with it. Want Java on a new platform? Wait for Sun to port it and support it. Want an urgent bug fix in JDK? Wait for Sun to do it.
"we'll have a dozen forks that won't play nice with each other"
We had a fork anyway (the MS 'Virtual Machine'), with closed source Java. Of all these divergent forks you predict, precisely NONE will be called Java (trademark). What will keep all these forks you predict going? If they are incompatible, they arn't Java! They won't be used. If they are compatible, but better, they become the standard, all try to follow. Meritocrasy.
This leaves runtime evaluation as the only way for the IDE (or any other piece of code) to discover the code structure.
What's wrong with that? It's a dynamic language, the IDE can do load and interpret the code as you write it, meaning that the full run-time environment (as much as you've written) is avavilable to the IDE. If you want to write a bit that depends on previous execution of other code, get the interpreter to interpret that bit, then write your code.
Eg.
f() creates a class and returns the class-name.
g() creates an instance of the class whose name is returned by f(), and executes method calls on it, depending on some other values.
Run g(), setting a break point, and continue writing after the call to f().
One of the best things about VB6 (from way back) IMHO was the 'immediate' window, which was basically an limited interpreter. With Python you get the whole system from an interpreter.
IMNHO, it is a horrible way to write a robust piece of software.
It depends. If you really need to do something like this, you'll do it, be it in C, C++, Java or whatever, and you'll make a horrible hack job of allowing dynamic class definition in those languages. Far better for the language to allow for it. (Eg. p249 More Effective C++ by Scott Meyers: 'Implementing Multiple Dispatch' vs. CLOS). All of these advanced features need to be used wisely, and I'd agree that injudicious use will create a mess, but would it be any worse than the hack-around in the static language?
At the risk of answering my own question, I suspect that a lot of the unpleasant idioms you get used to (in any language) will give ideas for what a better language should do automatically. No matter how used to the unpleasant idoms I get, I can still screw up one implemenation of them...
the flashplugin-nonfree package might be updated to use this version. the maintainer is listed here. You could email him/her/them or file a wish-list bug. Or you could hack it up yourself and do a NMU... Or just wait.
Exactly what I was thinking. If it baked the cold-blooded reptiles, why didn't it bake our warm-blooded ancestors? Just because they are smaller (like one replier said) doesn't cut it. If being hidden behind a rock suffices, then some dinosaurs would have been hidden behind mountains etc. and not all mice would have been in holes. TFA says that anything on the surface would have been broiled.
mate, that is an excellent idea. post a story about this. maybe an 'ask slashdot'. do some research on the voting habits of senators, congressmen etc. and point out, state-by-state and (whatever the area that a congressman is elected in is called*)-by-area who is voted worst and best for digital rights.
"China risks isolating itself if it creates standards that are incompatible with the rest of the world." ... perhaps the rest of the world risks cutting off China by charging to high patent licensing fees:)
If the government internalizes externalities by imposing fees for despoiling or consuming public goods (air, water) then this eliminates the "tragedy of the commons" problem and incentivizes business to reduce environmental impact. When resources cost money, the market will favor business which produce the most efficiently, that is, the most output for the least monetary (and therefore environmental) cost.
The "fee" will have to be very carefully calculated to match the risk. It should not be a linear relationship to mass of pollutant.
The quantity of a certain pollutant that is the point at which it starts killing people should be well above the foot of an exponential curve of cost. Factors like likelihood of continual exposure (eg. near homes/offices), the likely number of people exposed, the amount of 'nice green stuff' killed etc. should also be used. It should be prohibitively expensive to kill people (with probability more than existing risk), and it should be costly to increase risk non-negligibly.
One good thing that may come of this is that the densly populated slums would be much more costly to pollute, meaning that the poor, who are currently exposed to the most pollution, would get some protection.
good point. so my point is only relevant where portable music player makers want to ship software that can take audio from any source (eg. CD) and put it on the player. this will involve an encoder.
Grandparent noted the existence of 'cheapo' MP3 players. There is a good reason why OGG could gain mainstream acceptance: royalty-free use. The MPEG license will eventually be a significant part of the cost of the chips in a portable music player (if not a significant part of the overall cost of the player), but still a saving is a saving...
Since Windows is the dominant desktop OS by a significant margin, it is the only OS for which you expect all hardware to work. (This expectation will be violated from time to time, of course, and we all have anecdotes to show it). That much is pretty damned obvious.
The original article was annoying because he "didn't want to make this an issue about tech support", but it is just that. Not everything works straight off. Some people need to be told to turn the volume on. This can take a while for a tech support person to suggest.
Grrrr. This is just more of the same: mentioning a specific case, then arguing to the general. He is annoyed that people look at his particular problem and try to solve it for him! But his general point is either completely unjustified, or so painfully obvious that we don't need to be told (since he provides no evidence or argument to support anything more).
Stop posting this stuff!
Re: 1): (from TFA) M Bellare and C Namprempre "Authenticated encryption: Relations among notions and analysis of the generic composition paradigm" in T Okamato (ed.) Advances in cryptology - ASIACRYPT 2000, volume 1976 of Lecture notes in Computer Science pp 531-545. (Springer-Verlag, Berlin Germany Dec. 2000).
"The underlying AES-CTR-then-HMAC-SHA1 code is a provably secure authenticated encryption scheme per results by Bellare and Namprempre."
AFAIIA...
Two points: 1) there are proofs of security, I believe, but all such proofs are proofs of relative security. I.e. that cracking algorithm X with key length Y is at least as difficult as eg. factoring a prime of length Z. (I'm sorry not to provide a reference here. Anyone?).
2) The nature of the universe regarding determinism, is irrelevant to the existence of One Time Pads. The kind of randomness involved is not the type of whether some omniscient being could predict the values of the pad, but whether a (non-omniscient) attacker either a) can determine a pattern within the pad and usefully predict from that, or b) knowing the means of creation of the pad, can replicate the values of the pad.
In order to prevent (a), the pad creator must use a source of information that is sufficiently lacking in structure so that any structure that does exist does not exhibit itself within the lenght of the pad. Properties that depend on microscopic structure of materials here are commonly used, since the phase space is so incredibly vast (much larger than the length of any possible pad).
Again the chaotic nature (i.e. the sensitivity to initial conditions) of bodies about thermal equilibrium might make good sources for pad since even if an attacker replicates the pad creation equipment, no matter how close the initial conditions, there will be a point after which the replica diverges from the original, and only the first portion of the pad would be compromised.
Just 346 words remained on the list at this stage. The next stage is to involve the brain of the researcher. This eliminated all but seven words: Ugandan, Ukrainian, Egyptian, uninvited, incursive, indebted and unofficial. Naccache plumped for Egyptian, in this case.
Hmmm. So of the 346 words listed in the dictionary, the researcher had to guess? To a knowledgable attacker (eg. an analyst at the CIA who specializes in Egyptian terrorists) this step might be trivial, but in that case, they probably knew the answer anyway. On what grounds did the researcher choose Egyptian? Most likely because of the phrase "Egyptian Islamic Jihad". That is pure guesswork. It could have been an important point that the operative talked to Syrian intelligence or got captured by Mossad. The researcher simply didn't know.
The more general point is that algorithms cannot create information. Just like in image enhancement, if the data isn't there, in order to put it there you need other information to go on, and that is a non-algorithmic process (inference based on data), precisely because it is not guaranteed to be right.
"Half the time people insinuate that Bush is in bed with the Saudis (see Fahrenheit 911) and the other half of the time they insinuate that he hates the Muslims."
Bush hates *poor people*. The Saudi Royal family are rich, so Bush loves them.
The cost of war (just in terms of $$$ to fight) is almost unbelievable.
"Laser weapons are a relative bargain compared with existing long-range weapons: They're expected to cost $8,000 per shot versus up to hundreds of thousands for missiles."
8K$ is cheap for one shot!
What %age of GDP does the US spend on defense R&D? CIA factbook places it at 3.2% in 1999 (and it's gone up recently a lot). That is effectively the cost (along with the rest of defence spending) of staying as the most powerful country. This dominance is only the means to the true end of the govt. - the interests of the the country's citizens. Given the level of US defence spending lately, I wonder if the means has been mistaken for the end...
Strictly speaking the validation is only of the _implemenation_ of these algorithms. The NSA did invent SHA, but all these algorithms have stood up to academic attack (that we know of).
Well I'm still running my twin Voodoo2s SLIed. Doesn't run any games more recent than Deux Ex (1), but Rollcage runs smoothly. When Doom3 comes out then it'll be retired...
The article does make at least one fair point: that outsourcing information-based work overseas is bad for the economy, because 'Intellectual Capital' (i.e. mainly skilled people, but also organizations, contracts etc.) are moving out of the country. THIS IS NOTHING NEW! It's called Brain Drain, and the US has profited by for some fifty (50) years. Is this really a problem though? A lot of jobs moved overseas are not the IP-creating, IC (see above) -creating jobs. If a large proportion become highly skilled jobs, then there will be a problem of course...
"I think you misunderstand insurance. Either I am going to crash my car, or I am not going to crash my car. If I won't crash my car, then I don't need insurance."
my argument was that the above is fallacious not least because _we don't know_ who will and won't crash their cars. I admit I didn't make that clear.
my point was that we can't rule out the need for insurance because we don't know what the outcome of court cases will be.
Slashdot Mirror
The success of Windows has depended on its nature as a bundle: you pay 100$ (or Dell pay ???$ for you) and get the whole shebang. The licenses from this release pay for development of the new items in the next version of the bundle. .Mac) cannot be funded from the Windows license fee, unless Windows costs 300$ a license. People expect not to have to pay extra, so it's hard to convince them to do so.
This means that Windows customers expect everything to be included in the bundle that they need. The kind of services that TFA recommends MS sell (20$ a month for virual hard drive etc. like
This bundling also affects the lifecycle of the product: 5-6 years between XP and Longhorn is required because they need to do a lot of work! (Could their 're-write' do to them what Netscape's did?). There is so much in the bundle, and MS want to add so much more, that it takes a long time.
This has an impact on EOLing too - MS is still supporting (to some extent) Windows 98(!), 2000, XP. The cost of having a rapid release cycle is supporting many different releases (unless you EOL these releases just as rapidly, cf. Redhat Linux).
Overall, the size of Windows counts against MS in several different ways. It will be difficult for them to move away from it. Perhaps all those companies killed by MS integrating their features into the OS will have the last laugh?
To turn that around: as long as Java remains "under strict control", it will be limited to what Sun can do with it. Want Java on a new platform? Wait for Sun to port it and support it. Want an urgent bug fix in JDK? Wait for Sun to do it.
"we'll have a dozen forks that won't play nice with each other"
We had a fork anyway (the MS 'Virtual Machine'), with closed source Java. Of all these divergent forks you predict, precisely NONE will be called Java (trademark). What will keep all these forks you predict going? If they are incompatible, they arn't Java! They won't be used. If they are compatible, but better, they become the standard, all try to follow. Meritocrasy.
This leaves runtime evaluation as the only way for the IDE (or any other piece of code) to discover the code structure.
What's wrong with that? It's a dynamic language, the IDE can do load and interpret the code as you write it, meaning that the full run-time environment (as much as you've written) is avavilable to the IDE. If you want to write a bit that depends on previous execution of other code, get the interpreter to interpret that bit, then write your code.
Eg.
f() creates a class and returns the class-name.
g() creates an instance of the class whose name is returned by f(), and executes method calls on it, depending on some other values.
Run g(), setting a break point, and continue writing after the call to f().
One of the best things about VB6 (from way back) IMHO was the 'immediate' window, which was basically an limited interpreter. With Python you get the whole system from an interpreter.
IMNHO, it is a horrible way to write a robust piece of software.
It depends. If you really need to do something like this, you'll do it, be it in C, C++, Java or whatever, and you'll make a horrible hack job of allowing dynamic class definition in those languages. Far better for the language to allow for it. (Eg. p249 More Effective C++ by Scott Meyers: 'Implementing Multiple Dispatch' vs. CLOS). All of these advanced features need to be used wisely, and I'd agree that injudicious use will create a mess, but would it be any worse than the hack-around in the static language?
At the risk of answering my own question, I suspect that a lot of the unpleasant idioms you get used to (in any language) will give ideas for what a better language should do automatically. No matter how used to the unpleasant idoms I get, I can still screw up one implemenation of them...
ooops. someone got there first.
the flashplugin-nonfree package might be updated to use this version. the maintainer is listed here. You could email him/her/them or file a wish-list bug. Or you could hack it up yourself and do a NMU... Or just wait.
currently: the ability to build it with minGW (see yesterday's story for more info) and only using those parts of the Win32 API implemented in Wine.
Exactly what I was thinking. If it baked the cold-blooded reptiles, why didn't it bake our warm-blooded ancestors? Just because they are smaller (like one replier said) doesn't cut it. If being hidden behind a rock suffices, then some dinosaurs would have been hidden behind mountains etc. and not all mice would have been in holes. TFA says that anything on the surface would have been broiled.
mate, that is an excellent idea. post a story about this. maybe an 'ask slashdot'. do some research on the voting habits of senators, congressmen etc. and point out, state-by-state and (whatever the area that a congressman is elected in is called*)-by-area who is voted worst and best for digital rights.
*IANAAmerican
"China risks isolating itself if it creates standards that are incompatible with the rest of the world."
... perhaps the rest of the world risks cutting off China by charging to high patent licensing fees :)
If the government internalizes externalities by imposing fees for despoiling or consuming public goods (air, water) then this eliminates the "tragedy of the commons" problem and incentivizes business to reduce environmental impact. When resources cost money, the market will favor business which produce the most efficiently, that is, the most output for the least monetary (and therefore environmental) cost.
The "fee" will have to be very carefully calculated to match the risk. It should not be a linear relationship to mass of pollutant.
The quantity of a certain pollutant that is the point at which it starts killing people should be well above the foot of an exponential curve of cost. Factors like likelihood of continual exposure (eg. near homes/offices), the likely number of people exposed, the amount of 'nice green stuff' killed etc. should also be used. It should be prohibitively expensive to kill people (with probability more than existing risk), and it should be costly to increase risk non-negligibly.
One good thing that may come of this is that the densly populated slums would be much more costly to pollute, meaning that the poor, who are currently exposed to the most pollution, would get some protection.
good point.
so my point is only relevant where portable music player makers want to ship software that can take audio from any source (eg. CD) and put it on the player. this will involve an encoder.
Grandparent noted the existence of 'cheapo' MP3 players. There is a good reason why OGG could gain mainstream acceptance: royalty-free use. The MPEG license will eventually be a significant part of the cost of the chips in a portable music player (if not a significant part of the overall cost of the player), but still a saving is a saving...
so equilibrium rips it off a lot, then?
I can't actually find the homepage of Real Alternative, just a bunch of download sites. Is it like mplayer, i.e. using dlls of dubious legality?
Since Windows is the dominant desktop OS by a significant margin, it is the only OS for which you expect all hardware to work. (This expectation will be violated from time to time, of course, and we all have anecdotes to show it). That much is pretty damned obvious.
The original article was annoying because he "didn't want to make this an issue about tech support", but it is just that. Not everything works straight off. Some people need to be told to turn the volume on. This can take a while for a tech support person to suggest.
Grrrr. This is just more of the same: mentioning a specific case, then arguing to the general. He is annoyed that people look at his particular problem and try to solve it for him! But his general point is either completely unjustified, or so painfully obvious that we don't need to be told (since he provides no evidence or argument to support anything more).
Stop posting this stuff!
"the UK is part of the EU, but its intelligence services are among Echelon's sponsors."
It's kind of an open secret that the UK and the US together spy on Europe. In particular, there is evidence that the US used intelligence supplied from UK-based surveillance stations in order to give American companies advantages. One of those stations is at Menwith Hill. Mark Thomas did a stunt by flying over it IIRC in a balloon to see what would happen and had a party too.
Re: 1): (from TFA) M Bellare and C Namprempre "Authenticated encryption: Relations among notions and analysis of the generic composition paradigm" in T Okamato (ed.) Advances in cryptology - ASIACRYPT 2000, volume 1976 of Lecture notes in Computer Science pp 531-545. (Springer-Verlag, Berlin Germany Dec. 2000).
"The underlying AES-CTR-then-HMAC-SHA1 code is a provably secure authenticated encryption scheme per results by Bellare and Namprempre."
AFAIIA...
Two points: 1) there are proofs of security, I believe, but all such proofs are proofs of relative security. I.e. that cracking algorithm X with key length Y is at least as difficult as eg. factoring a prime of length Z. (I'm sorry not to provide a reference here. Anyone?).
2) The nature of the universe regarding determinism, is irrelevant to the existence of One Time Pads. The kind of randomness involved is not the type of whether some omniscient being could predict the values of the pad, but whether a (non-omniscient) attacker either a) can determine a pattern within the pad and usefully predict from that, or b) knowing the means of creation of the pad, can replicate the values of the pad.
In order to prevent (a), the pad creator must use a source of information that is sufficiently lacking in structure so that any structure that does exist does not exhibit itself within the lenght of the pad. Properties that depend on microscopic structure of materials here are commonly used, since the phase space is so incredibly vast (much larger than the length of any possible pad).
Again the chaotic nature (i.e. the sensitivity to initial conditions) of bodies about thermal equilibrium might make good sources for pad since even if an attacker replicates the pad creation equipment, no matter how close the initial conditions, there will be a point after which the replica diverges from the original, and only the first portion of the pad would be compromised.
Just 346 words remained on the list at this stage. The next stage is to involve the brain of the researcher. This eliminated all but seven words: Ugandan, Ukrainian, Egyptian, uninvited, incursive, indebted and unofficial. Naccache plumped for Egyptian, in this case.
Hmmm. So of the 346 words listed in the dictionary, the researcher had to guess? To a knowledgable attacker (eg. an analyst at the CIA who specializes in Egyptian terrorists) this step might be trivial, but in that case, they probably knew the answer anyway. On what grounds did the researcher choose Egyptian? Most likely because of the phrase "Egyptian Islamic Jihad". That is pure guesswork. It could have been an important point that the operative talked to Syrian intelligence or got captured by Mossad. The researcher simply didn't know.
The more general point is that algorithms cannot create information. Just like in image enhancement, if the data isn't there, in order to put it there you need other information to go on, and that is a non-algorithmic process (inference based on data), precisely because it is not guaranteed to be right.
"that our leadership hates ... muslims
"Half the time people insinuate that Bush is in bed with the Saudis (see Fahrenheit 911) and the other half of the time they insinuate that he hates the Muslims."
Bush hates *poor people*. The Saudi Royal family are rich, so Bush loves them.
The cost of war (just in terms of $$$ to fight) is almost unbelievable.
"Laser weapons are a relative bargain compared with existing long-range weapons: They're expected to cost $8,000 per shot versus up to hundreds of thousands for missiles."
8K$ is cheap for one shot!
What %age of GDP does the US spend on defense R&D? CIA factbook places it at 3.2% in 1999 (and it's gone up recently a lot). That is effectively the cost (along with the rest of defence spending) of staying as the most powerful country. This dominance is only the means to the true end of the govt. - the interests of the the country's citizens. Given the level of US defence spending lately, I wonder if the means has been mistaken for the end...
Strictly speaking the validation is only of the _implemenation_ of these algorithms. The NSA did invent SHA, but all these algorithms have stood up to academic attack (that we know of).
Well I'm still running my twin Voodoo2s SLIed. Doesn't run any games more recent than Deux Ex (1), but Rollcage runs smoothly. When Doom3 comes out then it'll be retired...
The article does make at least one fair point: that outsourcing information-based work overseas is bad for the economy, because 'Intellectual Capital' (i.e. mainly skilled people, but also organizations, contracts etc.) are moving out of the country. THIS IS NOTHING NEW! It's called Brain Drain, and the US has profited by for some fifty (50) years.
Is this really a problem though? A lot of jobs moved overseas are not the IP-creating, IC (see above) -creating jobs. If a large proportion become highly skilled jobs, then there will be a problem of course...
"I think you misunderstand insurance. Either I am going to crash my car, or I am not going to crash my car. If I won't crash my car, then I don't need insurance."
my argument was that the above is fallacious not least because _we don't know_ who will and won't crash their cars. I admit I didn't make that clear.
my point was that we can't rule out the need for insurance because we don't know what the outcome of court cases will be.