Re:Ok, so HURD is a microkernal os...
on
Hurd: H2 CD Images
·
· Score: 1
Notable microkernel based OSes include Windows NT, MacOS X, and of course HURD.
I'm not sure you could call NT a microkernel design anymore. That was the original intention, but over the years more and more stuff has been pulled into a monolithic kernel in the interests of performance.
It is true that microkernels have some of the advantages you cite, and on paper they certainly seem more sensible. However, the reality is that without proper hardware support, they just don't offer the same level of performance as can be achieved by a monolithic design.
I'm not sure exactly what the capabilities of the GC are (perhaps someone else here can share?) but what you describe is at least partly it. A vanilla "dependent texture read" does exactly what you say (this capability was actually available in the Matrox G400 graphics cards in the DX 6 era). A more sophisticated one, such as exposed in DX 8.0 and available in Xbox can perform some additional arithmetic on the sampled texel before using it as a texture address - for example, you can sample a normal from a normal map, rotate by a per-pixel iterated matrix, compute a reflection vector based on the iterated eye-vector and the rotated normal, and use that result to look up into a cube-map for a specular environment mapping term, all per-pixel. Maybe the GC can do this too - I don't know the details. I do know that the PS2 can't (heck, the PS2 can't even do a lot of pretty basic alpha blending modes). I was more responding to the assertion that DX 8.0 pixel shaders are really just simple combiners, because that is missing half the story.
I'm guessing most games are not going to use this though, instead opting to put the power on the eye candy.
In Xbox (and I think GC too, but I'm not sure) the Dolby encoding is handled by a separate audio processor, and so is essentially free, the game just has to set the right parameters. There is no tradeoff between eye-candy and audio fidelity on this score (although of course there may be the usual memory constraints for sound samples versus textures).
If you buy an extra add-on thing for an undisclosed price that Sony have not actually released yet and that no current games will take advantage of.
2. PS2 can have 8 controllers max using a Multi-tap (not 2).
Again, this supposes that you buy an extra add-on (although at least this time the add-on in question is more than vapor). 8 players doesn't seem like that big of a deal to me - sure, it's great for a few games, but I think the usefulness for most tops out at around 4 (if only because split-screen becomes silly after that point).
3. I'm pretty sure that they meant Dolby Digital decoding, not encoding. Because what is there to encode? Is there a line-in jack somewhere? All of the data on the disc is ENcoded, so it needs to be DEcoded when played back.
And I'm certain they meant encoding. There is plenty to encode - the game produces sound effects that can be spatialized and encoded as Dolby Digital, so that if you have a nice surround sound setup you hear the sound effects coming from the appropriate spot. Your receiver does the decoding, the console does the encoding.
4. PS2 does NOT require a seperate remote for DVD Video.
The PS2 only supports two out of the box, however, which I think was the point. A multitap is a pretty ugly mass of cables, and you do have to pay extra for it. Yes, you can play up to 8 players with two multitaps, but I don't see any reason why the Xbox or GC couldn't support similar multiway extensions in principle (Xbox should just be a USB hub), it's just that they generally don't need to because wanting more than 4 players is rare (more than 4 player split-screen just gets silly) and they already provided sufficient ports for that in the first place.
Colour combiners and pixel shaders are (more often than not) exactly equivalent. Have you actually used DX8? I've written a shitload of pixel shaders, and frankly, there's very little that you can't do with a suitably powerful colour combiner.
This is broadly true of the 'combiner' or 'arithmetic' portion of a DX 8.0 pixel shader, but leaves out the texture sampling portion of the shader, which is arguably the most powerful. This portion drives the texture sampling units, and can do simple operations whereby the addresses of texel fetches can depend on texels fetched by other texture units ("dependent texture reads"). This extends as far as being able to do a 3x3 matrix multiply and a simple reflection calculation into a cubic-environment map per-pixel, which is how you get some of the groovy bump-mapping effects. There are ways to get bump-mapping effects without this but they generally don't look as good.
I think there may be some confusion about the value of N here. Is N the actual size of the number in question, or the number of bits used to represent the number? In the first case, I think there are O(N) algorithms, but that's not terribly useful since it's easy to make N exponentially large (N is O(2^K) if K is number of bits). In the second case, then it's a trickier problem, and I believe the best known solutions are superpolynomial.
"Natural market forces" was your description, not mine - hence the quotation marks around it when I used the term. I also notice that you didn't address my point that you are asserting that freedom of speech is a "natural" or "fundamental" right but that ownership of one's intellectual ideas is somehow less "natural" or "fundamental" without really giving any justification for that point of view. I don't think it's at all obvious that one is a more "fundamental" right than the other - you simply assert that it is true, and of course, hey presto, that proves your case.
As for signing away your freedom of speech, thousands of people do that every day. In fact, if you're employed, you've probably done it yourself. Or are NDAs a gross violation of some "fundamental" or "natural" (however you were planning to define it) right too?
It seems fairly obvious that regardless of the philosophical purity or otherwise of freedom of speech reigning supreme, there are excellent pragmatic reasons why society sometimes puts a check on it. Blind following of a philosophical point of view (however noble you may believe it to be) without regard to the practical implications for society is called religion - and from my observance of religous governments around the world, most of them don't really look like fun places to be.
A moment ago it was a natural right, and now you are back to utilitarianism.
I don't think I ever described it as a "natural right". I'm pointing out that copyright serves a very valuable purpose in society. And what is a "natural right" anyway? How do you define the difference between a "natural" right and one that society grants people for some other reason? Sounds pretty vague and hand-wavy to me that one right is somehow "natural" and another one isn't. The right to free speech? Sounds natural, but is it? Many countries don't have it. What about it makes it "natural"? What about the "right" to bear arms? Is that natural? If so, why? If not, why not? Point is that you're describing a particular right as "natural" as though that makes it superior, but I'm arguing that every right we have is really just a construction of society (unless you want to invoke some sort of quasi-religious justification) and therefore the term "natural" is meaningless.
Actually, selling proprietary software doesn't even need copyright - just contracts (you do support contracts, right?). I can make you sign a contract agreeing not to redistribute the software before I give it to you. If you copy it, I sue you for breach. The only piece that copyright adds is that it prevents you from laundering away the contract via intermediary (i.e. a third party who didn't sign the contract that somehow got ahold of it would be able to copy it without limitation were it not for copyright).
My personal opinion is that if copyright were done away with, there'd be far less rubbish produced, but the best stuff would remain, since the best stuff tends to be done regardless of the profit motive.
Erm, no. For example, how are video games going to get made? It takes a whole team of people, including programmers, artists, testers, musicians and designers several man-years of effort to make a game. Those people are not going to build the quality of games with the depth of content you see today without being compensated for their work - because it's a full time job. Without copyright, the only games you'd see would be those that one or two guys can produce in their garage in their spare time - sure, they're fun, but they just aren't the same as polished modern productions. (And if you're going to argue that those modern polished productions are crap anyway, then all I can say is that it's a point of view, but not one shared by millions - and Final Fantasy VII could not have been produced by a couple of guys in their garage).
The fact is that copyright enables a whole class of work to be produced that otherwise would not have been because it would not be economically viable.
Furthermore, copyright is precisely what enables Free (as in speech) software. Without copyright, I could take a piece of Free software, modify it, and then distribute the binary without redistributing the source. The thing that stops me doing it is the license agreement. So if you support Free software, then you already, by implication, support the idea of software licensing. The only question is which software licenses do you allow. I would argue that developers should be free to choose whatever license they want and let the market decide which licenses people find attractive.
There is only one segment of the industry (small in number of programmers, but highly visible) that counts on copyright for survival: The off-the-shelf software producers. Most of these companies are producing bad code at the expense of users. I won't shed a tear if the Microsofts or the Adobes of the world fail to make a profit.
You may not shed a tear if those particular companies didn't make a profit, but you might shed a tear if off-the-shelf software disappeared completely. Why shouldn't it be possible to run a business producing off-the-shelf software? You may not like some of the companies currently doing that, but it doesn't follow that going into business to produce off-the-shelf software is in and of itself a bad thing.
while people like you argue otherwise on the grounds that people have some a basic right to a sallary even when the natural market forces do not support it.
But "natural market forces" do support it. Witness the many software companies in existence that exist purely by selling software, and not from selling services or hardware. Stallman is bemoaning that fact. The market currently supports proprietary software. If at some time in the future the market does not support proprietary software, then that's fine too, but I just don't think it's going to happen.
A world where software had to be free (e.g. by dictatorial fiat, constitutional amendment, or whatever) would be a distortion of that market. To take your nose-picking analogy - I'm all in favour of people charging for nose-picking if the market supported it, but I'm against a world where it was mandated that nose picking had to be provided free.
The freedom to be able to charge for my work if I like, and to let the market decide is a freedom. If you don't like the conditions I attach to my work, and someone else offers conditions you prefer, then get your software from them. Let the market decide who wins.
Free software - no problem. Probably a great idea for somethings. All software being free - not so great. The problem is that it means you can't have a software business. You can have a service or hardware business that happens to employ some programmers. But you can't have a business that exists to develop software. That just doesn't seem right. Why is it okay to make money from selling hardware or services, but not okay to make money from developing software?
In Stallman's universe, software companies just wouldn't exist. It would be impossible for a bunch of programmers to get together and support themselves by developing great software. They'd have to find some other thing they could sell along with it. But suppose they didn't want to do that. Suppose they just wanted to write software - they're screwed. Those people are no longer free to just write software!
The freedom to decide to charge for some of your software is a freedom, because it allows you to choose your career. Without the ability for anyone anywhere to ever charge for any software, the freedom for programmers to just be programmers disappears.
I'm not saying that Free software is a bad thing. But it has to co-exist with proprietary software for software development as a whole to remain viable.
You can only do the same thing with Java public member variables if the property being set is just that - a simple variable. C# get and set methods are actual methods - the whole point being that you can encapsulate a property which might be stored in some complex way, or that triggers processing (like logging) when it is changed, but still look syntactically as though it were a simple member variable.
And you can't seriously be touting Java's lack of enumerated types as an advantage.
Except that MS has to certify all titles before they can ship, so developers can't just put out shovelware without MS approval. And I can't see MS approving broken titles as it would damage the reputation of the console.
The US has the highest carbon emission per head of population in the world, more than any other industrialized country (indeed, more than any country).
At any rate, even if the US were no worse than other industrialized nations (which isn't true, but for the sake of argument, suppose it was), that still doesn't invalidate the need for a treaty. The point is that industrialized nations in general are emitting too much, and everybody has to reduce. The Kyoto treaty calls for everyone to meet certain targets, including other industrialized countries such as those in Europe. It's not as though the US is being asked to stick to a tougher target than anyone else.
The only way it's possible to argue that the US shouldn't sign up to the Kyoto treaty is if you don't buy into global warming as problem and carbon emissions as a cause of that problem. If that's the argument you want to make, then fine, make it. But don't try to pretend that the US doesn't really emit too much carbon, because on any reasonable measure it emits the most in the world.
[*] Before ranting that USA is world's biggest polluter, be sure to divide the pollution by the population before comparing and also remember that the "cleanest" nations are below dirt poor too.
Okay, I know this is probably a troll, but I'll bite. A quick glace at the EPA website shows that the US is currently responsible for 24% of the world's carbon emissions. The US has far less than that proportion of the world population. Per head of population, the US is the world's biggest polluter.
The constitution curtails the power of Congress to make laws which impede your freedom of speech. You can still sign away your rights to free speech in a contract - and this is indeed what many employees do when they sign an NDA as part of their employment contract. This would certainly not allow them to talk to the press about what's going on in the company, without proper authorisation.
So, it's quite likely that you can be fired for talking to the press.
I believe MSSQL doesn't have a one process per connection model (which would bring NT/2K to its knees pretty quickly), or even a one thread per connection model. Instead it uses a thread pooling model, which is generally more efficient (on NT/2K kernel architecture).
One advantage I saw is that they are going to be upgradeable.
In the world of console gaming, that's a disadvantage and is merely evidence that the Indrema folks just don't get it. The key advantage that consoles have over PCs from a game developers' perspective is that they are all identical (modulo a couple of country specific variations for TV standards - which are well known in advance). The moment you introduce upgradeability that goes out of the window, and all you really have is a PC in a different form factor.
Or rather, all such demonstrations must rely on assertions saying 'XYZ is a slow process'. Such assertions can never be proven.
Depends on exactly what you mean. It most certainly is possible to prove that some problems will take a certain minimum number of operations to solve in the worst case. For example, it is possible to prove that the general sorting problem for N keys requires at least NlogN operations in the worst case.
On the other hand, such proofs are often for general cases. It may be that there will be domain specific knowledge that can be brought to bear to build faster algorithms for specific cases. (For example, the sorting problem where the keys are known to be integers in a given finite range can be solved in near linear time.)
Also, it's possible that technological advances such as quantum computing might allow us to build machines where the individual 'operations' are of a higher class, thus reducing the time to solve these problems.
And after the tenth straight usability tester yells in frustration "Yes! Just render the damn thing already! And stop asking me dumb questions!", that option will be set to default to doing just that...
Slashdot Mirror
I'm not sure you could call NT a microkernel design anymore. That was the original intention, but over the years more and more stuff has been pulled into a monolithic kernel in the interests of performance.
It is true that microkernels have some of the advantages you cite, and on paper they certainly seem more sensible. However, the reality is that without proper hardware support, they just don't offer the same level of performance as can be achieved by a monolithic design.
I'm not sure exactly what the capabilities of the GC are (perhaps someone else here can share?) but what you describe is at least partly it. A vanilla "dependent texture read" does exactly what you say (this capability was actually available in the Matrox G400 graphics cards in the DX 6 era). A more sophisticated one, such as exposed in DX 8.0 and available in Xbox can perform some additional arithmetic on the sampled texel before using it as a texture address - for example, you can sample a normal from a normal map, rotate by a per-pixel iterated matrix, compute a reflection vector based on the iterated eye-vector and the rotated normal, and use that result to look up into a cube-map for a specular environment mapping term, all per-pixel. Maybe the GC can do this too - I don't know the details. I do know that the PS2 can't (heck, the PS2 can't even do a lot of pretty basic alpha blending modes). I was more responding to the assertion that DX 8.0 pixel shaders are really just simple combiners, because that is missing half the story.
In Xbox (and I think GC too, but I'm not sure) the Dolby encoding is handled by a separate audio processor, and so is essentially free, the game just has to set the right parameters. There is no tradeoff between eye-candy and audio fidelity on this score (although of course there may be the usual memory constraints for sound samples versus textures).
If you buy an extra add-on thing for an undisclosed price that Sony have not actually released yet and that no current games will take advantage of.
2. PS2 can have 8 controllers max using a Multi-tap (not 2).
Again, this supposes that you buy an extra add-on (although at least this time the add-on in question is more than vapor). 8 players doesn't seem like that big of a deal to me - sure, it's great for a few games, but I think the usefulness for most tops out at around 4 (if only because split-screen becomes silly after that point).
3. I'm pretty sure that they meant Dolby Digital decoding, not encoding. Because what is there to encode? Is there a line-in jack somewhere? All of the data on the disc is ENcoded, so it needs to be DEcoded when played back.
And I'm certain they meant encoding. There is plenty to encode - the game produces sound effects that can be spatialized and encoded as Dolby Digital, so that if you have a nice surround sound setup you hear the sound effects coming from the appropriate spot. Your receiver does the decoding, the console does the encoding.
4. PS2 does NOT require a seperate remote for DVD Video.
Ok, I guess I'll give you that one.
The PS2 only supports two out of the box, however, which I think was the point. A multitap is a pretty ugly mass of cables, and you do have to pay extra for it. Yes, you can play up to 8 players with two multitaps, but I don't see any reason why the Xbox or GC couldn't support similar multiway extensions in principle (Xbox should just be a USB hub), it's just that they generally don't need to because wanting more than 4 players is rare (more than 4 player split-screen just gets silly) and they already provided sufficient ports for that in the first place.
This is broadly true of the 'combiner' or 'arithmetic' portion of a DX 8.0 pixel shader, but leaves out the texture sampling portion of the shader, which is arguably the most powerful. This portion drives the texture sampling units, and can do simple operations whereby the addresses of texel fetches can depend on texels fetched by other texture units ("dependent texture reads"). This extends as far as being able to do a 3x3 matrix multiply and a simple reflection calculation into a cubic-environment map per-pixel, which is how you get some of the groovy bump-mapping effects. There are ways to get bump-mapping effects without this but they generally don't look as good.
I think there may be some confusion about the value of N here. Is N the actual size of the number in question, or the number of bits used to represent the number? In the first case, I think there are O(N) algorithms, but that's not terribly useful since it's easy to make N exponentially large (N is O(2^K) if K is number of bits). In the second case, then it's a trickier problem, and I believe the best known solutions are superpolynomial.
As for signing away your freedom of speech, thousands of people do that every day. In fact, if you're employed, you've probably done it yourself. Or are NDAs a gross violation of some "fundamental" or "natural" (however you were planning to define it) right too?
It seems fairly obvious that regardless of the philosophical purity or otherwise of freedom of speech reigning supreme, there are excellent pragmatic reasons why society sometimes puts a check on it. Blind following of a philosophical point of view (however noble you may believe it to be) without regard to the practical implications for society is called religion - and from my observance of religous governments around the world, most of them don't really look like fun places to be.
I don't think I ever described it as a "natural right". I'm pointing out that copyright serves a very valuable purpose in society. And what is a "natural right" anyway? How do you define the difference between a "natural" right and one that society grants people for some other reason? Sounds pretty vague and hand-wavy to me that one right is somehow "natural" and another one isn't. The right to free speech? Sounds natural, but is it? Many countries don't have it. What about it makes it "natural"? What about the "right" to bear arms? Is that natural? If so, why? If not, why not? Point is that you're describing a particular right as "natural" as though that makes it superior, but I'm arguing that every right we have is really just a construction of society (unless you want to invoke some sort of quasi-religious justification) and therefore the term "natural" is meaningless.
Actually, selling proprietary software doesn't even need copyright - just contracts (you do support contracts, right?). I can make you sign a contract agreeing not to redistribute the software before I give it to you. If you copy it, I sue you for breach. The only piece that copyright adds is that it prevents you from laundering away the contract via intermediary (i.e. a third party who didn't sign the contract that somehow got ahold of it would be able to copy it without limitation were it not for copyright).
Erm, no. For example, how are video games going to get made? It takes a whole team of people, including programmers, artists, testers, musicians and designers several man-years of effort to make a game. Those people are not going to build the quality of games with the depth of content you see today without being compensated for their work - because it's a full time job. Without copyright, the only games you'd see would be those that one or two guys can produce in their garage in their spare time - sure, they're fun, but they just aren't the same as polished modern productions. (And if you're going to argue that those modern polished productions are crap anyway, then all I can say is that it's a point of view, but not one shared by millions - and Final Fantasy VII could not have been produced by a couple of guys in their garage).
The fact is that copyright enables a whole class of work to be produced that otherwise would not have been because it would not be economically viable.
Furthermore, copyright is precisely what enables Free (as in speech) software. Without copyright, I could take a piece of Free software, modify it, and then distribute the binary without redistributing the source. The thing that stops me doing it is the license agreement. So if you support Free software, then you already, by implication, support the idea of software licensing. The only question is which software licenses do you allow. I would argue that developers should be free to choose whatever license they want and let the market decide which licenses people find attractive.
I'm guessing you're not a gamer then.
You may not shed a tear if those particular companies didn't make a profit, but you might shed a tear if off-the-shelf software disappeared completely. Why shouldn't it be possible to run a business producing off-the-shelf software? You may not like some of the companies currently doing that, but it doesn't follow that going into business to produce off-the-shelf software is in and of itself a bad thing.
But "natural market forces" do support it. Witness the many software companies in existence that exist purely by selling software, and not from selling services or hardware. Stallman is bemoaning that fact. The market currently supports proprietary software. If at some time in the future the market does not support proprietary software, then that's fine too, but I just don't think it's going to happen.
A world where software had to be free (e.g. by dictatorial fiat, constitutional amendment, or whatever) would be a distortion of that market. To take your nose-picking analogy - I'm all in favour of people charging for nose-picking if the market supported it, but I'm against a world where it was mandated that nose picking had to be provided free.
The freedom to be able to charge for my work if I like, and to let the market decide is a freedom. If you don't like the conditions I attach to my work, and someone else offers conditions you prefer, then get your software from them. Let the market decide who wins.
In Stallman's universe, software companies just wouldn't exist. It would be impossible for a bunch of programmers to get together and support themselves by developing great software. They'd have to find some other thing they could sell along with it. But suppose they didn't want to do that. Suppose they just wanted to write software - they're screwed. Those people are no longer free to just write software!
The freedom to decide to charge for some of your software is a freedom, because it allows you to choose your career. Without the ability for anyone anywhere to ever charge for any software, the freedom for programmers to just be programmers disappears.
I'm not saying that Free software is a bad thing. But it has to co-exist with proprietary software for software development as a whole to remain viable.
And you can't seriously be touting Java's lack of enumerated types as an advantage.
Except that MS has to certify all titles before they can ship, so developers can't just put out shovelware without MS approval. And I can't see MS approving broken titles as it would damage the reputation of the console.
Because they made Mac games, which pretty much qualifies them as a non-profit organization :)
At any rate, even if the US were no worse than other industrialized nations (which isn't true, but for the sake of argument, suppose it was), that still doesn't invalidate the need for a treaty. The point is that industrialized nations in general are emitting too much, and everybody has to reduce. The Kyoto treaty calls for everyone to meet certain targets, including other industrialized countries such as those in Europe. It's not as though the US is being asked to stick to a tougher target than anyone else.
The only way it's possible to argue that the US shouldn't sign up to the Kyoto treaty is if you don't buy into global warming as problem and carbon emissions as a cause of that problem. If that's the argument you want to make, then fine, make it. But don't try to pretend that the US doesn't really emit too much carbon, because on any reasonable measure it emits the most in the world.
Okay, I know this is probably a troll, but I'll bite. A quick glace at the EPA website shows that the US is currently responsible for 24% of the world's carbon emissions. The US has far less than that proportion of the world population. Per head of population, the US is the world's biggest polluter.
So, it's quite likely that you can be fired for talking to the press.
I believe MSSQL doesn't have a one process per connection model (which would bring NT/2K to its knees pretty quickly), or even a one thread per connection model. Instead it uses a thread pooling model, which is generally more efficient (on NT/2K kernel architecture).
Yes, you're absolutely right, history has shown that when it comes to making money, Microsoft have no clue. Erm, wait a minute...
In the world of console gaming, that's a disadvantage and is merely evidence that the Indrema folks just don't get it. The key advantage that consoles have over PCs from a game developers' perspective is that they are all identical (modulo a couple of country specific variations for TV standards - which are well known in advance). The moment you introduce upgradeability that goes out of the window, and all you really have is a PC in a different form factor.
Depends on exactly what you mean. It most certainly is possible to prove that some problems will take a certain minimum number of operations to solve in the worst case. For example, it is possible to prove that the general sorting problem for N keys requires at least NlogN operations in the worst case.
On the other hand, such proofs are often for general cases. It may be that there will be domain specific knowledge that can be brought to bear to build faster algorithms for specific cases. (For example, the sorting problem where the keys are known to be integers in a given finite range can be solved in near linear time.)
Also, it's possible that technological advances such as quantum computing might allow us to build machines where the individual 'operations' are of a higher class, thus reducing the time to solve these problems.
And after the tenth straight usability tester yells in frustration "Yes! Just render the damn thing already! And stop asking me dumb questions!", that option will be set to default to doing just that...