First, Be never really solved the "fragile base class" problem, they just worked around it, with a set of library guidelines designed to minimize the impact. I assume that the paper you were referring to was this: http://www.2f.ru/holy-wars/fbc.html
It's true that the intent was that application developers wouldn't have to recompile for future OS releases (though even that's not guaranteed, in the above statement). I was thinking more about the pre-1.0 timeframe, where by DR8 or so, people were already getting pretty sick of the whole process. Yeah, I know - "pre-1.0, you'd expect compatibility problems". But still, the lack of compatibility release-to-release was much worse than on OSs with a more robust interface to the OS and system libraries.
Those workarounds for the FBC problem don't really address the other ABI issues, either. The layout of the vtable, and various other aspects of C++ object layout, have a distressing tendency to change with newer releases of the compiler. This would break compatibility in weird and wonderful ways, as well as making it *really* difficult to use any other compiler to build BeOS apps.
When I asked engineers at Be about why they were using a then relatively-old version of Metrowerks C++ (on PowerPC), the response I was given was that they couldn't upgrade, because of ABI issues. They were certainly aware of the issues, and were actively working on addressing some of them, but there were (and are) substantial problems with using C++ classes as a library interface.
According to what I could find online, the Haiku project intends to solve the FBC by implementing a new, incompatible ABI for a future release. I also note from other comments on this story that Haiku only has binary compatibility with BeOS apps if it's compiled with a specific version of GCC. That doesn't sound like the ABI compatibility problem has been "solved", in any useful sense.
See also: http://www.bebits.com/app/4011 A page for GCC on BeOS, which states that gcc-3.x for BeOS can't be used to build applications that link to the Be libraries.
The dependency on the C++ ABI was one of the strangest things about BeOS. Having to recompile all of your applications for incremental OS releases is a real drag for third-party developers.
Somehow, the idea that they'd be totally at the mercy of their compiler vendor didn't seem like a problem to them. I don't know of any other OS that tied itself so closely to one programming environment, especially not one controlled by a different company.
If Haiku doesn't resolve that issue, then it'll only ever see open-source applications. Commercial software developers won't want to put up with that kind of instability.
Some folks don't like the terms of Apple's APSL license, and that'd be a factor. But notice that the major Linux distributions is agreeing on any other single system, either.
I think mostly they're going off on their own because it's a fun little problem to work on, but obviously it'd be to the advantage of sysadmins everywhere to have a standard, maintainable services startup system. Maybe a clear winner will emerge, and at least keep Linux from fragmenting too much.
"Of course it's legal - they wouldn't have gotten approval to conduct the experiment otherwise, and they certainly wouldn't have published the results."
Not that this will necessarily make you feel any better about the ethics of the situation, but the embryo they used was a reject from in-vitro fertilization. It was going to be destroyed anyway, regardless of whether the experiment was performed. It's a fact that the majority of embryos produced for IVF are never brought to term.
contemporaneous with the C64 in the Apple stable was the IIGS.
Uh, no - the C64 was released in 1982, and the Apple IIGS wasn't released until 1986 (that's right - two years after the first Macintosh). The Apple contemporaries to the C64 were the Apple II+ and Apple IIe.
Radiosity rendering and ray tracing are very complementary, in that one does a really good job with diffuse/global lighting effects, and the other does a much better job with local lighting, refraction, and reflection. Neither one alone gives what I'd consider really realistic images, at least without "cheating" in one way or another. Some combination of techniques is probably the best approach for realistic rendering.
As it was explained to me when I worked there, the Legal team at Apple feels that they'd be vulnerable to shareholder lawsuits if they gave away something that customers would be willing to pay for. This is traditionally attributed to the Sarbanes-Oxley Act, but I think the intent of the policy is more general than that.
In general, updates to existing products are not allowed to be free if they add new features, only if they fix bugs. There are a bunch of exceptions, including for products that are given away, like iTunes. I expect that the iPhone is actually being "sold" a bit at a time over the course of the mandatory 2-year contract, and so since customers are still technically paying for them, it's okay to add new features in a software update.
I didn't much like this explanation the first time I heard it, but given the number of shareholder lawsuits Apple already gets every year, they definitely have reason to be cautious. As long as the prices for feature upgrades remain relatively low, it probably won't anger the customer base too much, and it'll hopefully keep the class-action lawyers at bay.
Allocated sectors will have approximately equal numbers of ones and zeroes, empty/erased sectors will be all zeroes (or all ones, depending on spin topology). So, you just count all the one bits (or zero bits), and divide by the total number of bits, then multiply by two.
The first US nuclear power reactor (EBR-1) was a liquid-metal cooled breeder reactor, as was the Fermi 1 reactor near Detroit, Michigan. The Fermi reactor had a minor meltdown accident in 1963. Overall, the safety record of liquid-metal reactors hasn't been particularly impressive, at least in the power-generation arena.
Yeah, the more I read about that system, the less I like it for screening carry-ons. Could still be useful for freight and checked baggage, though. It's really too bad that NMR requires such high magnetic fields, otherwise it'd probably be a good choice for an "explosives detector".
But I can only imagine the havoc that would result from passing travelers' luggage through the equivalent of an MRI machine before loading it on a plane. On the plus side, you wouldn't have to worry about not detecting knives in peoeple's luggage - when a knife shoots out the side of the bag into the magnet, you'll have a pretty good chance of noticing it:-)
X-ray spectroscopy, Neutron backscatter, and other techniques have all been proposed as additions to the baggage-screening procedure, for just this reason.
A computerized scanner that simply prints "WARNING: Contains >8 grams of reactive metal" will probably give a lot better results than a poorly-trained screener trying to puzzle out a contrast-enhanced X-ray image.
This is why it's so tiring arguing with Flat-Earthers like yourself. You apparently don't know much of anything about genetics or evolution, except for a few poorly-defined phrases like "mutations never add any information to the genome".
When you're faced with a counter-example, like Downs Syndrome, you just change the definition of your original statement to exclude that specific counter-example, and move on.
So, in what way does the integration of Endogenous Retroviruses not constitute "adding information to the genome"? Similarly, when Polyploidy creates a new species that then diverges from its parents, how is that not "adding information to the genome"? How about Transposons? Do those "add to the genome"?
I look forward to your answers with great expectation.
Okay, a couple of people report a problem on Apple's support site, one guy got what's probably bad advice from his local Mac Genius, and now it's on Slashdot?
I don't doubt that these people are upset, but there's nothing in the linked discussion that even validates the theory that the problem is Boot Camp related...
When the choice for the average consumer becomes, spend $50 on a Linux computer, or spend $800 on a Windows computer, I think that most people will begin to switch.
That argument only works if the two systems are otherwise equivalent. Linux is making progress on some of its various weaknesses relative to Windows (lack of first-tier applications, difficulty of setup, lack of an installed base), but for most people, Linux isn't even close to being a viable alternative to Windows.
my experience has been different. I get about as many errors on the iphone as i do typing on a real keyboard. But then again, my typing is awful... Oddly, the typo checker doesn't work in this textbox, which slows me down a lot.
Sorry, I was just responding to the issue of attribution. They seem to have done the right thing there, according to the requirements of the Creative Commons license.
I have to admit that I don't know much about Australian law, but under the (possibly flawed) assumption that it's similar to US and UK law, they really should have investigated the privacy issues some more. Given the inherent difficulty of suing someone in another country, they'll presumably get away with it, but it's fairly obviously incorrect behavior, nonetheless.
Which is an Australian solicitor's attempt to explain the legal rules covering candid photography in Australia. He actually addresses this particular case (search for "Virgin").
His conclusion is that, because the pictures Virgin Mobile used weren't taken in Australia, and the people pictured weren't Australians, they narrowly avoided falling under the jurisdiction of Australian law. Had those pictures been taken in Australia, or of Australian citizens, they absolutely would have needed model releases. There's no way they could have known ahead of time that none of the pictures were of Australians, so they basically got lucky.
Nevertheless, they really should have known better. Skating around the edges of the law like that is not very responsible behavior for a corporation to engage in. They've since abandoned that advertising campaign, presumably because someone with a clue looked into it and decided it was a really bad idea. Who knows? Maybe the bad publicity on the Internet contributed to the decision.
In the Flickr / Virgin Mobile case, the problem revolves around whether the photo had the appropriate attribution.
Actually, in the Virgin Mobile case, the issue at hand is that they didn't have a model release for the person in the picture. The picture was properly attributed.
It's well-established in the law (they even cited the code in the Slashdot summary) that you need someone's explicit permission to use their likeness for commercial purposes. Just having the photographer's permission isn't enough. How this relates to Facebook probably depends on the terms of service that the users agreed to when they signed up for the service. It's much less clearcut than the Virgin Mobile case.
Not exactly accurate...
on
Is SETI Worth It?
·
· Score: 4, Informative
SETI is looking for a signal from an advanced civilization that is deliberately using archaic methods to transmit. What I mean is that they're looking for a beacon signal that's designed to be easy to interpret, and that's transmitted at an extremely high power level.
On a practical level, that's the best they can do. Using the best receivers that we currently have, it'd just barely be possible to detect a megawatt-level signal from a few light years away, if it was aimed right at us. Detecting the equivalent of leakage from a TV transmission is a complete fantasy. Unless there's someone out there that's really desperate to be heard, we'll never find them.
And of course, we're not about to start a program of sending similar signals to all the nearest stars - that'd take real money. If we detect a signal, then we might respond back.
Unfortunately, the same argument holds in the other direction, too. Any alien civilizations out there would be foolish to waste the resources to send a signal we could detect, before they were sure we were there to hear it. When I think about SETI, I sometimes imagine thousands of intelligent species out there, all monitoring their antenna arrays, waiting for a signal that none of them have the funding to send...
If you list the designers, then you should list the developers. If you list the developers, then you should list the QA team. If the QA team gets credit, then you should also include Operations...
Eventually, to be "fair", you'd have to list the whole company. If you draw the line somewhere else other than all or none, then you'll be leaving someone off arbitrarily.
This was one of the arguments put forward at Apple to justify removing individual credits from Mac OS X.
In fact, the argument for American labor in this case feels a whole lot like the same luddite argument spurred by fears that "robots and computers in the workplace will cost American jobs!". That didn't really happen, either, did it?
Actually, it did. Maybe you're not from the USA, or maybe you live on one of the coasts, but there's no arguing with the fact that automation reduces the number of factory jobs, and directly leads to unemployment in manufacturing-heavy areas. The promise was that eventually the country would switch over to a services-based economy, and that'd take up the slack. That's happening, to some extent, but those guys that used to manufacture cars at $28/hour are now mostly working in service jobs that pay minimum wage, or slightly above. That's not exactly an equitable trade, and increasing automation is going to squeeze them out of some of those jobs, as well.
I actually worked at a company that produced automation for manufacturers, and the majority of my family has worked in/around the US car industry, so I know whereof I speak. The primary goal of all the "plant modernization" programs I was ever involved with was simply to reduce the labor cost of goods produced. That means reducing the number of workers. Anything else (increased quality, or increased flexibility) was, at best, a secondary goal.
Even on the non-manufacturing front, the news reports tell us that the productivity of American companies are up, but that the unemployment rate is holding steady. That implies that office workers are simply working longer hours to keep up with doing the work of more than one person, and in fact that's largely the case. E-mail, cellular phones, and other technological gadgets also keep employees "plugged in" to their jobs, even when they're nominally not at work.
While it's true that there's less of an emphasis on manufacturing than on design in the Valley, that doesn't mean that it's inevitable that things will be designed here, and made in China. As it turns out, there's even a car factory in the Bay Area, though it's not quite in "Silicon Valley", per se. Given that the initial market for practical electric cars is going to mostly be on the West coast anyway, it would make sense to manufacture the cars here, rather than transporting them from across the Pacific. That economic argument was sufficient for several Japanese and European car makers to build auto plants here, rather than shipping cars overseas.
1.5 million Kilometers is not very far, even in terms of the solar system. It's about 0.01 Astronomical units, or about 1% of the distance from Earth to the Sun, or 2.6% of the distance to Mars (at closest approach).
You'd need a much more powerful laser, or a more sensitive detector, to communicate over inteplanetary distances.
Slashdot Mirror
A lamp powered by happy thoughts.
First, Be never really solved the "fragile base class" problem, they just worked around it, with a set of library guidelines designed to minimize the impact. I assume that the paper you were referring to was this:
http://www.2f.ru/holy-wars/fbc.html
It's true that the intent was that application developers wouldn't have to recompile for future OS releases (though even that's not guaranteed, in the above statement). I was thinking more about the pre-1.0 timeframe, where by DR8 or so, people were already getting pretty sick of the whole process. Yeah, I know - "pre-1.0, you'd expect compatibility problems". But still, the lack of compatibility release-to-release was much worse than on OSs with a more robust interface to the OS and system libraries.
Those workarounds for the FBC problem don't really address the other ABI issues, either. The layout of the vtable, and various other aspects of C++ object layout, have a distressing tendency to change with newer releases of the compiler. This would break compatibility in weird and wonderful ways, as well as making it *really* difficult to use any other compiler to build BeOS apps.
When I asked engineers at Be about why they were using a then relatively-old version of Metrowerks C++ (on PowerPC), the response I was given was that they couldn't upgrade, because of ABI issues. They were certainly aware of the issues, and were actively working on addressing some of them, but there were (and are) substantial problems with using C++ classes as a library interface.
According to what I could find online, the Haiku project intends to solve the FBC by implementing a new, incompatible ABI for a future release. I also note from other comments on this story that Haiku only has binary compatibility with BeOS apps if it's compiled with a specific version of GCC. That doesn't sound like the ABI compatibility problem has been "solved", in any useful sense.
See also:
http://www.bebits.com/app/4011
A page for GCC on BeOS, which states that gcc-3.x for BeOS can't be used to build applications that link to the Be libraries.
The dependency on the C++ ABI was one of the strangest things about BeOS. Having to recompile all of your applications for incremental OS releases is a real drag for third-party developers.
Somehow, the idea that they'd be totally at the mercy of their compiler vendor didn't seem like a problem to them. I don't know of any other OS that tied itself so closely to one programming environment, especially not one controlled by a different company.
If Haiku doesn't resolve that issue, then it'll only ever see open-source applications. Commercial software developers won't want to put up with that kind of instability.
Some folks don't like the terms of Apple's APSL license, and that'd be a factor. But notice that the major Linux distributions is agreeing on any other single system, either.
I think mostly they're going off on their own because it's a fun little problem to work on, but obviously it'd be to the advantage of sysadmins everywhere to have a standard, maintainable services startup system. Maybe a clear winner will emerge, and at least keep Linux from fragmenting too much.
"Of course it's legal - they wouldn't have gotten approval to conduct the experiment otherwise, and they certainly wouldn't have published the results."
Not that this will necessarily make you feel any better about the ethics of the situation, but the embryo they used was a reject from in-vitro fertilization. It was going to be destroyed anyway, regardless of whether the experiment was performed. It's a fact that the majority of embryos produced for IVF are never brought to term.
Uh, no - the C64 was released in 1982, and the Apple IIGS wasn't released until 1986 (that's right - two years after the first Macintosh). The Apple contemporaries to the C64 were the Apple II+ and Apple IIe.
http://en.wikipedia.org/wiki/Apple_II
http://en.wikipedia.org/wiki/Commodore_64
Which is not to say that the Apple II wasn't an important gaming platform, but by the early 80's it was already showing its age.
Radiosity rendering and ray tracing are very complementary, in that one does a really good job with diffuse/global lighting effects, and the other does a much better job with local lighting, refraction, and reflection. Neither one alone gives what I'd consider really realistic images, at least without "cheating" in one way or another. Some combination of techniques is probably the best approach for realistic rendering.
As it was explained to me when I worked there, the Legal team at Apple feels that they'd be vulnerable to shareholder lawsuits if they gave away something that customers would be willing to pay for. This is traditionally attributed to the Sarbanes-Oxley Act, but I think the intent of the policy is more general than that.
In general, updates to existing products are not allowed to be free if they add new features, only if they fix bugs. There are a bunch of exceptions, including for products that are given away, like iTunes. I expect that the iPhone is actually being "sold" a bit at a time over the course of the mandatory 2-year contract, and so since customers are still technically paying for them, it's okay to add new features in a software update.
I didn't much like this explanation the first time I heard it, but given the number of shareholder lawsuits Apple already gets every year, they definitely have reason to be cautious. As long as the prices for feature upgrades remain relatively low, it probably won't anger the customer base too much, and it'll hopefully keep the class-action lawyers at bay.
Allocated sectors will have approximately equal numbers of ones and zeroes, empty/erased sectors will be all zeroes (or all ones, depending on spin topology). So, you just count all the one bits (or zero bits), and divide by the total number of bits, then multiply by two.
Easy as pie!
The first US nuclear power reactor (EBR-1) was a liquid-metal cooled breeder reactor, as was the Fermi 1 reactor near Detroit, Michigan. The Fermi reactor had a minor meltdown accident in 1963. Overall, the safety record of liquid-metal reactors hasn't been particularly impressive, at least in the power-generation arena.
Yeah, the more I read about that system, the less I like it for screening carry-ons. Could still be useful for freight and checked baggage, though. It's really too bad that NMR requires such high magnetic fields, otherwise it'd probably be a good choice for an "explosives detector".
:-)
But I can only imagine the havoc that would result from passing travelers' luggage through the equivalent of an MRI machine before loading it on a plane. On the plus side, you wouldn't have to worry about not detecting knives in peoeple's luggage - when a knife shoots out the side of the bag into the magnet, you'll have a pretty good chance of noticing it
X-ray spectroscopy, Neutron backscatter, and other techniques have all been proposed as additions to the baggage-screening procedure, for just this reason.
A computerized scanner that simply prints "WARNING: Contains >8 grams of reactive metal" will probably give a lot better results than a poorly-trained screener trying to puzzle out a contrast-enhanced X-ray image.
http://abcnews.go.com/Business/BusinessTravel/story?id=97746&page=1
It'll be a while before this tech hits your local airport, but I bet it's coming.
That's the price for an 11 inch diagonal OLED TV. I don't think I'd pay $1,750.00 for any 11-inch screen, no matter how nice it looks.
http://www.informationweek.com/news/showArticle.jhtml?articleID=202103858
This is why it's so tiring arguing with Flat-Earthers like yourself. You apparently don't know much of anything about genetics or evolution, except for a few poorly-defined phrases like "mutations never add any information to the genome".
When you're faced with a counter-example, like Downs Syndrome, you just change the definition of your original statement to exclude that specific counter-example, and move on.
So, in what way does the integration of Endogenous Retroviruses not constitute "adding information to the genome"? Similarly, when Polyploidy creates a new species that then diverges from its parents, how is that not "adding information to the genome"? How about Transposons? Do those "add to the genome"?
I look forward to your answers with great expectation.
Okay, a couple of people report a problem on Apple's support site, one guy got what's probably bad advice from his local Mac Genius, and now it's on Slashdot?
I don't doubt that these people are upset, but there's nothing in the linked discussion that even validates the theory that the problem is Boot Camp related...
That argument only works if the two systems are otherwise equivalent. Linux is making progress on some of its various weaknesses relative to Windows (lack of first-tier applications, difficulty of setup, lack of an installed base), but for most people, Linux isn't even close to being a viable alternative to Windows.
my experience has been different. I get about as many errors on the iphone as i do typing on a real keyboard. But then again, my typing is awful...
Oddly, the typo checker doesn't work in this textbox, which slows me down a lot.
Sorry, I was just responding to the issue of attribution. They seem to have done the right thing there, according to the requirements of the Creative Commons license.
I have to admit that I don't know much about Australian law, but under the (possibly flawed) assumption that it's similar to US and UK law, they really should have investigated the privacy issues some more. Given the inherent difficulty of suing someone in another country, they'll presumably get away with it, but it's fairly obviously incorrect behavior, nonetheless.
For those of you following along. There's a Flickr discussion about this issue here:
http://www.flickr.com/groups/central/discuss/72157600541608353/
One of the links currently lurking in that discussion is to:
http://www.4020.net/words/photorights.php
Which is an Australian solicitor's attempt to explain the legal rules covering candid photography in Australia. He actually addresses this particular case (search for "Virgin").
His conclusion is that, because the pictures Virgin Mobile used weren't taken in Australia, and the people pictured weren't Australians, they narrowly avoided falling under the jurisdiction of Australian law. Had those pictures been taken in Australia, or of Australian citizens, they absolutely would have needed model releases. There's no way they could have known ahead of time that none of the pictures were of Australians, so they basically got lucky.
Nevertheless, they really should have known better. Skating around the edges of the law like that is not very responsible behavior for a corporation to engage in. They've since abandoned that advertising campaign, presumably because someone with a clue looked into it and decided it was a really bad idea. Who knows? Maybe the bad publicity on the Internet contributed to the decision.
Actually, in the Virgin Mobile case, the issue at hand is that they didn't have a model release for the person in the picture. The picture was properly attributed.
It's well-established in the law (they even cited the code in the Slashdot summary) that you need someone's explicit permission to use their likeness for commercial purposes. Just having the photographer's permission isn't enough. How this relates to Facebook probably depends on the terms of service that the users agreed to when they signed up for the service. It's much less clearcut than the Virgin Mobile case.
SETI is looking for a signal from an advanced civilization that is deliberately using archaic methods to transmit. What I mean is that they're looking for a beacon signal that's designed to be easy to interpret, and that's transmitted at an extremely high power level.
On a practical level, that's the best they can do. Using the best receivers that we currently have, it'd just barely be possible to detect a megawatt-level signal from a few light years away, if it was aimed right at us. Detecting the equivalent of leakage from a TV transmission is a complete fantasy. Unless there's someone out there that's really desperate to be heard, we'll never find them.
And of course, we're not about to start a program of sending similar signals to all the nearest stars - that'd take real money. If we detect a signal, then we might respond back.
Unfortunately, the same argument holds in the other direction, too. Any alien civilizations out there would be foolish to waste the resources to send a signal we could detect, before they were sure we were there to hear it. When I think about SETI, I sometimes imagine thousands of intelligent species out there, all monitoring their antenna arrays, waiting for a signal that none of them have the funding to send...
If you list the designers, then you should list the developers. If you list the developers, then you should list the QA team. If the QA team gets credit, then you should also include Operations...
Eventually, to be "fair", you'd have to list the whole company. If you draw the line somewhere else other than all or none, then you'll be leaving someone off arbitrarily.
This was one of the arguments put forward at Apple to justify removing individual credits from Mac OS X.
Actually, it did. Maybe you're not from the USA, or maybe you live on one of the coasts, but there's no arguing with the fact that automation reduces the number of factory jobs, and directly leads to unemployment in manufacturing-heavy areas. The promise was that eventually the country would switch over to a services-based economy, and that'd take up the slack. That's happening, to some extent, but those guys that used to manufacture cars at $28/hour are now mostly working in service jobs that pay minimum wage, or slightly above. That's not exactly an equitable trade, and increasing automation is going to squeeze them out of some of those jobs, as well.
I actually worked at a company that produced automation for manufacturers, and the majority of my family has worked in/around the US car industry, so I know whereof I speak. The primary goal of all the "plant modernization" programs I was ever involved with was simply to reduce the labor cost of goods produced. That means reducing the number of workers. Anything else (increased quality, or increased flexibility) was, at best, a secondary goal.
Even on the non-manufacturing front, the news reports tell us that the productivity of American companies are up, but that the unemployment rate is holding steady. That implies that office workers are simply working longer hours to keep up with doing the work of more than one person, and in fact that's largely the case. E-mail, cellular phones, and other technological gadgets also keep employees "plugged in" to their jobs, even when they're nominally not at work.
While it's true that there's less of an emphasis on manufacturing than on design in the Valley, that doesn't mean that it's inevitable that things will be designed here, and made in China. As it turns out, there's even a car factory in the Bay Area, though it's not quite in "Silicon Valley", per se. Given that the initial market for practical electric cars is going to mostly be on the West coast anyway, it would make sense to manufacture the cars here, rather than transporting them from across the Pacific. That economic argument was sufficient for several Japanese and European car makers to build auto plants here, rather than shipping cars overseas.
1.5 million Kilometers is not very far, even in terms of the solar system. It's about 0.01 Astronomical units, or about 1% of the distance from Earth to the Sun, or 2.6% of the distance to Mars (at closest approach).
You'd need a much more powerful laser, or a more sensitive detector, to communicate over inteplanetary distances.