I'd only heard of the hot tub from Vic (unless we're misremembering her story), so I'd assumed that was what people used, having done similar (but less extreme) things myself.
The "boiling point" idea makes more sense between applications than with the first application. Once everyone uses at least one OSS program, they'll be much more likely to use more, just because the licenses and process will be familiar.
There's also a Brazillian company that sells a Linux distribution that's reasonably popular in the area. So Linux is also providing income to locals, which tends to earn political support.
It's a hot tub, not a sauna, that they have at the pole. 100 farenheit is warm for a hot tub, but cold for a sauna. A hot tub has more thermal conductivity, though, which means that your skin gets as hot as the water. (If your skin got to sauna temperatures, it would all die, because the water in the cells would boil)
Obviously, there's going to be a certain loss of quality...
Obviously, there's going to be a certain loss of fidelity if you make a DVD from a rip of a VHS tape. Chances are that they found a copy closer to the original than was previously available.
It has polymorphism via late binding of function calls, but the different implementations are independant of each other. I meant interfaces in the Java keyword sense (the set of methods an object of unknown exact type is known to support), not the more generic sense. That is, each type defines the v-table, and each implementation fills it in (subtypes are private and instances are generally produced by factory methods), using, if desired, some functions which give the "do nothing special" behavior.
Not necessarily. The kernel's use of objects is essentially to define interfaces and implement them. While there are generally provided defaults for many of the operations (so you don't have to write your own "return -ENOSYS;"), there isn't really inheritance. C++, so far as I know, doesn't have a way of specifying that your use of OOP follows this pattern.
That is to say, a language feature in one language may support more generality than the idiom you use in a language without this feature, and the may not have a way of specifying that the extra generality is not used. In this case, you may be using a complicated feature to get a simple effect, and you could just do something simple by hand and give the compiler the additional information that you are not doing (or supporting others doing) the complicated thing, making the optimizer work better.
C++ is not a well-designed language. There have been many occasions on which incompatible changes had to be made in the language standard. Unless you're actually doing language research, it's a good idea to avoid it.
On the other hand, a good exception mechanism, if implemented in C, would be very useful in the kernel for error-handling.
One of the main reasons for not supporting C++ in the kernel is that the kernel is object-oriented C, and C++ objects would confuse matters. Of course, C++ has other features as well, and it would be good if those, when useful, could also be supported in a standard idiom in C in the kernel.
You're wrong about SPF. It doesn't do anything with the RFC822 "From:" header. It verifies the SMTP "RCPT FROM" address, which appears (generally) as "Received: from " in the headers, and is not generally displayed. That is, it tells you about where you got the mail from, not who sent it. It's really more like a postmark than a sender, and lets you know that some guy with a red marker didn't draw some inaccurate postmark on the envelope.
For that matter, alumni.almamater.edu could check SPF records and let you relay outgoing mail through them as well, if it is authenticated as really coming from the address that your account forwards to. The only reason that forwarding services are asymmetrical this way is that there is no good way of having a relay which is not an open relay.
I see what GTA:SA has to do with AAA, and maybe Sims 2, but Doom 3 and Unreal Tournament 2004?
Re:Trying to answer the question that was asked...
on
How Cheap Can A PC Be?
·
· Score: 1
The tricky thing is actually that making a $100 PC requires chosing relatively obscure parts, which means that you're going to have to do your own motherboard. Once you factor in how much it costs to design the system, the $100 per-unit cost is going to be overwhelmed by design costs for quite a large number of units.
If you can ignore the issues for end users (your keyboard doesn't have the right connectors, it doesn't have a power supply, it's a choking hazard for small children, it'll get lost in the couch, it doesn't have a case, etc.), this site will sell you 1000 systems for $99 each.
But Gordon's love interest is vital to making a movie a reasonable representation of the game. I mean, the movie wouldn't work if Gordon himself were shredded by the giant fan, and it wouldn't be right if nobody you cared about was...
Actually, Gordon needs a whole bunch of them, so that they can get killed off throughout the movie.
"Creative Commons Licensed" is an annoying thing to say, because there were always a bunch of Creative Commons licenses. All it actually means is that you're allowed to do something you might not expect, and you can read a one-paragraph description and understand what you're allowed to do. A better analogy might be programs for the Mac: you don't know what the program will do, but you know it will be presented in a way familiar from other programs. Obviously, there are many different programs for the Mac.
Personally, I think that Creative Commons should also have proprietary licenses, simply for the benefit of users of proprietary content, who would be able to recognize the requirements without wading through legalese each time they got something (and also to show side-by-side comparisons of proprietary licenses and free ones). Probably starting with "Fair Use": you own a copy of the work and can use it yourself or sell it, but cannot duplicate it except to make a backup for your own use.
Offline games generally have NPCs who say and do a variety of offensive things. Often they also involve defeating these NPCs. Sometimes they involve just ignoring the NPCs. Why shouldn't online games have the same sorts of characters? The documentation and the game mechanics should be inoffensive, but the characters need not be.
Some of their items are really kind of strange. Does requiring each student to have their own computer mean that the campus is "more wired" or less? On the maximally wired campus, each student would have no use for a computer of their own, as there would be a campus-provided computer everywhere, with secure access to the student's data.
On the other hand, if everyone has a laptop with wifi, what would you need school-owned computers for? Currently, I doubt all of the public Athena workstations are ever in use at the same time. All of the workstations in one place often are, but there isn't a factor for distance to a workstation in the survey. Amusing note: the MIT museam has, in the Athena exhibit, a working quickstation (although I didn't check whether the root password is the usual one).
It's odd that offering courses online counts toward whether the campus is wired or not; an online course may be "wired", but has nothing to do with the campus. For that matter, streaming A/V is terrible for trying to learn anything (according to a friend who took a course that way when taking another course that conflicted).
MIT's registration system seemed to me mainly designed to encourage people to get advice from their advisors. Carrying the form across campus is a bit silly, but the face-to-face interaction is worthwhile, so that you can find out about prereqs you're ignoring or convince your advisor that you'll be fine.
For setting up web pages, there's also ssh/scp (to a dialup), actually writing them on a Athena machine, and AFS.
In order to do a proper study, you'd have to come up with a plausible model, so you know how to correct for such things as target size. The number of installations of a piece of software shouldn't have any effect on how many flaws there are in it. A larger piece of software would presumably have more flaws (since there are more chances to have flaws), but that might not be an effect you want to remove from the analysis.
On the other hand, the number of break-ins would presumably be related to the number of installations. But is it linear (picking a random machine, you're more likely to pick a windows box to atttack), or higher (the attacker is more likely to care, as well). Is there a saturation effect (an attacker isn't going to perform a second attack on a machine they already control)?
Makes me wonder whether the use of hydrogen gas for energy will lead to people getting newly-synthesized water (probably only as a luxury item, though). You can be sure that there's nothing but H2O in your water, because it was H2 and O2 until it was bottled.
The boards of US companies aren't legally obliged to increase shareholder value; they're obliged to do what they say they'll do in the prospectus. Most companies say that they're a good investment and they'll increase your value. Google says that shareholder value isn't a concern of theirs, and investors who want to affect company policy can kindly take their money elsewhere.
Companies normally become publically traded in order to raise investment money, which requires attacting investors with promises of future wealth. They also usually don't have well-known policies and aren't usually already successful. Google, on the other hand, didn't care particularly much about raising money, and people knew what kind of policies Google had, and that they were doing well with them. So Google promised nothing about investment returns and promised that they would continue to be non-evil, which was making them money.
It would, in fact, be deceiving investors if Google started doing evil things, and they could get in trouble for that.
It *is* Microsoft which has released the software that passes the data from one user's session to the next. Google's software just makes it accessible to people other than identity theives.
They're doing erasure correction with an error correcting code on the complete transfer. It's sort of a good idea, but it's only useful for transferring large files, because if there are dropped packets, you don't get that data until the end. This makes it useless for anything where you want to use the data progressively, like rendering a web page, image, etc., as you download it. So this isn't going to replace TCP except in special cases.
For that matter, it would be much easier and no less effective to send data by UDP and request replacements by TCP as the packets turn out to be missing, with the replacements sent at the end, rather than ending with a chunk of ECC information suitable for reconstructing the lost packets without any feedback from the receiver.
(Of course, it is an interesting math problem to figure out what information you should send after transferring a file to optimally correct a set of deletions that only the receiver knows; but in real life, you can just send the necessary information back)
It would be really amusing if they found out who leaked San Andreas and used his likeness for a big villian in the next game. Punish the pseudo-theft of a video game with a virtual lynching.
They're only counting the different sequences of DNA which are transcribed to give proteins. That's like counting the lines of C code in a program that abuses the preprocessor extensively. You're ignoring all the ifdefs, the macros, the set of makefiles and shell scripts stuffed into comments, and so forth.
Sure, you have a small set of proteins. But these proteins are expressed or not expressed based on binding sites for other molecules in the non-transcribed DNA around the portion for the proteins; this is sufficient to account for the differences between all of the types of cells in your body, which is clearly a major set of variations.
Slashdot Mirror
I'd only heard of the hot tub from Vic (unless we're misremembering her story), so I'd assumed that was what people used, having done similar (but less extreme) things myself.
The "boiling point" idea makes more sense between applications than with the first application. Once everyone uses at least one OSS program, they'll be much more likely to use more, just because the licenses and process will be familiar.
It took down the backend, not the ATMs. The ATMs were running fine, but obviously couldn't do much without any way to get transactions processed.
There's also a Brazillian company that sells a Linux distribution that's reasonably popular in the area. So Linux is also providing income to locals, which tends to earn political support.
You know a sauna is full of steam, right?
It's a hot tub, not a sauna, that they have at the pole. 100 farenheit is warm for a hot tub, but cold for a sauna. A hot tub has more thermal conductivity, though, which means that your skin gets as hot as the water. (If your skin got to sauna temperatures, it would all die, because the water in the cells would boil)
Obviously, there's going to be a certain loss of quality...
Obviously, there's going to be a certain loss of fidelity if you make a DVD from a rip of a VHS tape. Chances are that they found a copy closer to the original than was previously available.
It has polymorphism via late binding of function calls, but the different implementations are independant of each other. I meant interfaces in the Java keyword sense (the set of methods an object of unknown exact type is known to support), not the more generic sense. That is, each type defines the v-table, and each implementation fills it in (subtypes are private and instances are generally produced by factory methods), using, if desired, some functions which give the "do nothing special" behavior.
Not necessarily. The kernel's use of objects is essentially to define interfaces and implement them. While there are generally provided defaults for many of the operations (so you don't have to write your own "return -ENOSYS;"), there isn't really inheritance. C++, so far as I know, doesn't have a way of specifying that your use of OOP follows this pattern.
That is to say, a language feature in one language may support more generality than the idiom you use in a language without this feature, and the may not have a way of specifying that the extra generality is not used. In this case, you may be using a complicated feature to get a simple effect, and you could just do something simple by hand and give the compiler the additional information that you are not doing (or supporting others doing) the complicated thing, making the optimizer work better.
C++ is not a well-designed language. There have been many occasions on which incompatible changes had to be made in the language standard. Unless you're actually doing language research, it's a good idea to avoid it.
On the other hand, a good exception mechanism, if implemented in C, would be very useful in the kernel for error-handling.
One of the main reasons for not supporting C++ in the kernel is that the kernel is object-oriented C, and C++ objects would confuse matters. Of course, C++ has other features as well, and it would be good if those, when useful, could also be supported in a standard idiom in C in the kernel.
You're wrong about SPF. It doesn't do anything with the RFC822 "From:" header. It verifies the SMTP "RCPT FROM" address, which appears (generally) as "Received: from " in the headers, and is not generally displayed. That is, it tells you about where you got the mail from, not who sent it. It's really more like a postmark than a sender, and lets you know that some guy with a red marker didn't draw some inaccurate postmark on the envelope.
For that matter, alumni.almamater.edu could check SPF records and let you relay outgoing mail through them as well, if it is authenticated as really coming from the address that your account forwards to. The only reason that forwarding services are asymmetrical this way is that there is no good way of having a relay which is not an open relay.
"But where are we going to find rubber pants our size?" would make a great Google ad.
I see what GTA:SA has to do with AAA, and maybe Sims 2, but Doom 3 and Unreal Tournament 2004?
The tricky thing is actually that making a $100 PC requires chosing relatively obscure parts, which means that you're going to have to do your own motherboard. Once you factor in how much it costs to design the system, the $100 per-unit cost is going to be overwhelmed by design costs for quite a large number of units.
If you can ignore the issues for end users (your keyboard doesn't have the right connectors, it doesn't have a power supply, it's a choking hazard for small children, it'll get lost in the couch, it doesn't have a case, etc.), this site will sell you 1000 systems for $99 each.
But Gordon's love interest is vital to making a movie a reasonable representation of the game. I mean, the movie wouldn't work if Gordon himself were shredded by the giant fan, and it wouldn't be right if nobody you cared about was...
Actually, Gordon needs a whole bunch of them, so that they can get killed off throughout the movie.
"Creative Commons Licensed" is an annoying thing to say, because there were always a bunch of Creative Commons licenses. All it actually means is that you're allowed to do something you might not expect, and you can read a one-paragraph description and understand what you're allowed to do. A better analogy might be programs for the Mac: you don't know what the program will do, but you know it will be presented in a way familiar from other programs. Obviously, there are many different programs for the Mac.
Personally, I think that Creative Commons should also have proprietary licenses, simply for the benefit of users of proprietary content, who would be able to recognize the requirements without wading through legalese each time they got something (and also to show side-by-side comparisons of proprietary licenses and free ones). Probably starting with "Fair Use": you own a copy of the work and can use it yourself or sell it, but cannot duplicate it except to make a backup for your own use.
Offline games generally have NPCs who say and do a variety of offensive things. Often they also involve defeating these NPCs. Sometimes they involve just ignoring the NPCs. Why shouldn't online games have the same sorts of characters? The documentation and the game mechanics should be inoffensive, but the characters need not be.
Some of their items are really kind of strange. Does requiring each student to have their own computer mean that the campus is "more wired" or less? On the maximally wired campus, each student would have no use for a computer of their own, as there would be a campus-provided computer everywhere, with secure access to the student's data.
On the other hand, if everyone has a laptop with wifi, what would you need school-owned computers for? Currently, I doubt all of the public Athena workstations are ever in use at the same time. All of the workstations in one place often are, but there isn't a factor for distance to a workstation in the survey. Amusing note: the MIT museam has, in the Athena exhibit, a working quickstation (although I didn't check whether the root password is the usual one).
It's odd that offering courses online counts toward whether the campus is wired or not; an online course may be "wired", but has nothing to do with the campus. For that matter, streaming A/V is terrible for trying to learn anything (according to a friend who took a course that way when taking another course that conflicted).
MIT's registration system seemed to me mainly designed to encourage people to get advice from their advisors. Carrying the form across campus is a bit silly, but the face-to-face interaction is worthwhile, so that you can find out about prereqs you're ignoring or convince your advisor that you'll be fine.
For setting up web pages, there's also ssh/scp (to a dialup), actually writing them on a Athena machine, and AFS.
In order to do a proper study, you'd have to come up with a plausible model, so you know how to correct for such things as target size. The number of installations of a piece of software shouldn't have any effect on how many flaws there are in it. A larger piece of software would presumably have more flaws (since there are more chances to have flaws), but that might not be an effect you want to remove from the analysis.
On the other hand, the number of break-ins would presumably be related to the number of installations. But is it linear (picking a random machine, you're more likely to pick a windows box to atttack), or higher (the attacker is more likely to care, as well). Is there a saturation effect (an attacker isn't going to perform a second attack on a machine they already control)?
Makes me wonder whether the use of hydrogen gas for energy will lead to people getting newly-synthesized water (probably only as a luxury item, though). You can be sure that there's nothing but H2O in your water, because it was H2 and O2 until it was bottled.
The boards of US companies aren't legally obliged to increase shareholder value; they're obliged to do what they say they'll do in the prospectus. Most companies say that they're a good investment and they'll increase your value. Google says that shareholder value isn't a concern of theirs, and investors who want to affect company policy can kindly take their money elsewhere.
Companies normally become publically traded in order to raise investment money, which requires attacting investors with promises of future wealth. They also usually don't have well-known policies and aren't usually already successful. Google, on the other hand, didn't care particularly much about raising money, and people knew what kind of policies Google had, and that they were doing well with them. So Google promised nothing about investment returns and promised that they would continue to be non-evil, which was making them money.
It would, in fact, be deceiving investors if Google started doing evil things, and they could get in trouble for that.
It *is* Microsoft which has released the software that passes the data from one user's session to the next. Google's software just makes it accessible to people other than identity theives.
They're doing erasure correction with an error correcting code on the complete transfer. It's sort of a good idea, but it's only useful for transferring large files, because if there are dropped packets, you don't get that data until the end. This makes it useless for anything where you want to use the data progressively, like rendering a web page, image, etc., as you download it. So this isn't going to replace TCP except in special cases.
For that matter, it would be much easier and no less effective to send data by UDP and request replacements by TCP as the packets turn out to be missing, with the replacements sent at the end, rather than ending with a chunk of ECC information suitable for reconstructing the lost packets without any feedback from the receiver.
(Of course, it is an interesting math problem to figure out what information you should send after transferring a file to optimally correct a set of deletions that only the receiver knows; but in real life, you can just send the necessary information back)
It would be really amusing if they found out who leaked San Andreas and used his likeness for a big villian in the next game. Punish the pseudo-theft of a video game with a virtual lynching.
They're only counting the different sequences of DNA which are transcribed to give proteins. That's like counting the lines of C code in a program that abuses the preprocessor extensively. You're ignoring all the ifdefs, the macros, the set of makefiles and shell scripts stuffed into comments, and so forth.
Sure, you have a small set of proteins. But these proteins are expressed or not expressed based on binding sites for other molecules in the non-transcribed DNA around the portion for the proteins; this is sufficient to account for the differences between all of the types of cells in your body, which is clearly a major set of variations.