Yeah, there needs to be some authentication. You should be able to prove your age and your sex, and then other clients will give out their information only if you can provide a valid certificate saying that you are twelve years old. Such certificates could be created by a PGP-style web of trust (but this is probably too complicated and too likely to go wrong) or by a central authority, like the government. (When you apply for a passport you're also given a diskette containing digital certs signed by the government saying your name, age, sex, possibly address - it's then your choice whether to use these certificates or just destroy the diskette.)
This wouldn't avoid a 12-year-old using the software while a child molester looks over their shoulder, but at least it's a start.
But why should you have to specify the routing as part of the address? Surely it's better just to say _where_ the letter should go to and let the postal system work out _how_. Like the change from UUCP decvax!host1!host2!user email addresses to the Internet style which specifies a destination host and lets the network (and MX records) do the routing.
My objection to this plan is why invent some new alphanumeric coding? Why not just use latitude and longitude?
But 'not techical' people are exactly the target market of Linux desktop distributions or BeOS-Lazarus or Mac OS X. I think OSNews does a good job of pointing out things that people don't want to hear - that often, things are broken or unintuitive out of the box, and that saying 'just run vi and edit this file in/etc' isn't an acceptable answer.
For real end-user testing you need people a lot less technical and a lot more stupid than the average developer.
It's Eugenia Loli-Queru! For years now, new distributions have been reviewed on OSNews, a story gets posted to Slashdot, and one of the first comments is that 'Eugenia hates anything that isn't BeOS'. But now we have an official BeOS, or as close as possible, reviewed on her site and she still runs through a long list of complaints. Which seems to demonstrate that she's not that biased after all.
Can some explain this 'bada-bing' catchphrase and why it is thought to be associated with Americans of Italian extraction or the Mafia? Do real mobsters go around saying it? Does it mean anything?
They already got $1G from Microsoft. Why not return that money to shareholders, wind up the company and go home?
Last time I looked SCO's market capitalization was only $40M or so, which reflects the market's judgement that SCO's management is not going to make good use of the cash windfall, so its value when tied up in SCO is a lot less than the cash value.
Unless Novell is getting 95% of the payment Microsoft made? Which would explain why the share price is so low, and deflate conspiracy theories about Microsoft funding a lawsuit. (Although $50M is still a handy sum to throw at the legal system.)
If your hardware has caching you do not 'gain from using a smaller address space' because that is nonsensical. Code running on most processors doesn't have a choice of what address space to use, and even processors that have a choice of modes (eg 32 bit / 64 bit) don't make memory access any faster if you choose the smaller address space.
But I accept that memory accesses may on average be faster if you use a smaller amount of memory, since it will fit better in the various caches.
Random access to an n-bit address space takes O(n) time
This is not relevant when deciding what algorithm to use, since you already know that your code will be running on 32-bit or 64-bit hardware with 32 or 40 or whatever number of address lines. On such a machine, memory access takes constant time whether your program uses a one kilobyte array or one gigabyte. I don't know of any modern computer which is able to do faster memory access if the program restricts itself to a smaller address space, so for all practical purposes you can say that a memory lookup takes constant time. This is not a bogus assumption, at least not for a programmer.
Random access to an n-bit address space takes O(log n) time.
Doesn't this depend on your particular hardware? On a machine with L1 cache, L2 cache, maybe some kind of NUMA, and swap space your statement may be a good approximation of the truth, but a machine with no processor cache and no swap file (they did exist, not so long ago) would have constant time access to any part of memory. Or is this not what you meant?
Who is to say that the universe - the 'real' universe inhabited by the top-level civilization running simulations of all the others - is not infinite? They could build a computer which builds new parts of itself as it runs, and so have unbounded (if not infinite) computing power.
If the restriction is on speed rather than memory size, you can run as many simulations as you want, they just get slower and slower.
I always had a slight feeling of unease about hash tables. Like in C++, the standard library provides std::map which has guaranteed O(log N) time for insertion and lookup, where N is the number of elements in the map. But std::map is too slow for many people, who use instead a non-standard hash_map class. Such a hash table can only guarantee O(N) time, but 'usually' insertion and extraction take constant time.
I mean, you'd have to be really unlucky to get a large number of hash collisions, wouldn't you? Apparently more than just luck is involved, if the person generating input knows the hash function you're using. So maybe the standards committee had the right idea after all in choosing the container with the best worst-case performance rather than the best average-case.
I'm interested by what the authors say about 'modern universal hash functions' and how they are immune from attack. Does this mean you pick a random string and XOR against it before computing each hash value, or something like that? It would seem there could still be some way to discover the exact hash function being used by means of timing attacks (you can tell when an input has caused a hash collision because the response time might be slightly slower). But I need to read the paper referred to.
Surely by the same argument the people running a simulation of us are themselves being simulated along with their simulation, and that by another group of people who are themselves living in a simulation created by an even more advanced civilization...
I hear that Sealand makes a profit... would it be possible to stick an aircraft carrier in the middle of the ocean and stick web servers on it? Perhaps the jurisdictional issues would be more complex than with Sealand.
That may be true... but if people calculate cost per week why isn't the unit of electricity billing the kilowatt-week? Every other electric device gives power consumption (or power output, if you prefer) in watts or kilowatts. Better to say 'power consumption N kilowatts, which means 168 * N kilowatt hours per week'. Then at least you are giving the sane units as well.
Engine power is often given in horsepower but that's a crazy and obsolete unit - move to kW now! </metric-troll>
I think that in practice it is going the other way - the various free languages which compile to Java bytecode are _more_ featureful than Sun's efforts, not less.
The story contains a typo - the power output is about 2.6kW, not 2.6kWh (which would be a measure of energy).
Of course, if you are like the freezer manufacturer who quoted their product's power consumption in 'kilowatt hours per week', you could say that the scooter produces 2.6kWh per hour...
Of course it brings them a revenue stream. Semi-legal file sharing networks are one of the main drivers behind broadband. The 'Time Warner' part of the company may not like it, but the 'AOL' half might think that if piracy is going to happen anyway, you might as well try to make a buck off it.
I think that any serious Internet user would not use AOL broadband however... that's the problem.
AFAIK all the changes in Java 1.5 are in the compiler - the bytecode has the same format as before. (At least this is the impression I get from the Slashdot comments.) So it's not a question of whether to change the JVM.
What's interesting is that many of the new features were already available in other languages that compile down to JVM bytecode, such as Generic Java and Nice. But people seem reluctant to use new features until they're in an officially blessed Sun release, even if the generated code is 100% Java-compatible.
Lord of the Rings has now officially jumped the shark - musical episode. Next expect to see Frodo's long-lost cousin Bodo and his hip new catchprase 'it's bodocious'!. They tell me that Will Smith will be making a guest appearance as Sauron's evil twin.
Re:IT/Clustering made cheap
on
ClusterKnoppix
·
· Score: 1
Can people still log in and do their work while the cluster is running jobs?
Slashdot Mirror
Yeah, there needs to be some authentication. You should be able to prove your age and your sex, and then other clients will give out their information only if you can provide a valid certificate saying that you are twelve years old. Such certificates could be created by a PGP-style web of trust (but this is probably too complicated and too likely to go wrong) or by a central authority, like the government. (When you apply for a passport you're also given a diskette containing digital certs signed by the government saying your name, age, sex, possibly address - it's then your choice whether to use these certificates or just destroy the diskette.)
This wouldn't avoid a 12-year-old using the software while a child molester looks over their shoulder, but at least it's a start.
But why should you have to specify the routing as part of the address? Surely it's better just to say _where_ the letter should go to and let the postal system work out _how_. Like the change from UUCP decvax!host1!host2!user email addresses to the Internet style which specifies a destination host and lets the network (and MX records) do the routing.
My objection to this plan is why invent some new alphanumeric coding? Why not just use latitude and longitude?
But 'not techical' people are exactly the target market of Linux desktop distributions or BeOS-Lazarus or Mac OS X. I think OSNews does a good job of pointing out things that people don't want to hear - that often, things are broken or unintuitive out of the box, and that saying 'just run vi and edit this file in /etc' isn't an acceptable answer.
For real end-user testing you need people a lot less technical and a lot more stupid than the average developer.
It's Eugenia Loli-Queru! For years now, new distributions have been reviewed on OSNews, a story gets posted to Slashdot, and one of the first comments is that 'Eugenia hates anything that isn't BeOS'. But now we have an official BeOS, or as close as possible, reviewed on her site and she still runs through a long list of complaints. Which seems to demonstrate that she's not that biased after all.
7) ???
8) Profit!
(or not, in the case of Be, Inc.)
Good to see that Sinclair Research has made another comeback with the Zeta. I wonder if they will port it to the QL or to tricycles?
It might be wasted energy for a human to spin round quickly, but Yoda is not human. Perhaps his species has specially adapted spinning muscles.
Can some explain this 'bada-bing' catchphrase and why it is thought to be associated with Americans of Italian extraction or the Mafia? Do real mobsters go around saying it? Does it mean anything?
They already got $1G from Microsoft. Why not return that money to shareholders, wind up the company and go home?
Last time I looked SCO's market capitalization was only $40M or so, which reflects the market's judgement that SCO's management is not going to make good use of the cash windfall, so its value when tied up in SCO is a lot less than the cash value.
Unless Novell is getting 95% of the payment Microsoft made? Which would explain why the share price is so low, and deflate conspiracy theories about Microsoft funding a lawsuit. (Although $50M is still a handy sum to throw at the legal system.)
If your hardware has caching you do not 'gain from using a smaller address space' because that is nonsensical. Code running on most processors doesn't have a choice of what address space to use, and even processors that have a choice of modes (eg 32 bit / 64 bit) don't make memory access any faster if you choose the smaller address space.
But I accept that memory accesses may on average be faster if you use a smaller amount of memory, since it will fit better in the various caches.
Doesn't this depend on your particular hardware? On a machine with L1 cache, L2 cache, maybe some kind of NUMA, and swap space your statement may be a good approximation of the truth, but a machine with no processor cache and no swap file (they did exist, not so long ago) would have constant time access to any part of memory. Or is this not what you meant?
Who is to say that the universe - the 'real' universe inhabited by the top-level civilization running simulations of all the others - is not infinite? They could build a computer which builds new parts of itself as it runs, and so have unbounded (if not infinite) computing power.
If the restriction is on speed rather than memory size, you can run as many simulations as you want, they just get slower and slower.
I always had a slight feeling of unease about hash tables. Like in C++, the standard library provides std::map which has guaranteed O(log N) time for insertion and lookup, where N is the number of elements in the map. But std::map is too slow for many people, who use instead a non-standard hash_map class. Such a hash table can only guarantee O(N) time, but 'usually' insertion and extraction take constant time.
I mean, you'd have to be really unlucky to get a large number of hash collisions, wouldn't you? Apparently more than just luck is involved, if the person generating input knows the hash function you're using. So maybe the standards committee had the right idea after all in choosing the container with the best worst-case performance rather than the best average-case.
I'm interested by what the authors say about 'modern universal hash functions' and how they are immune from attack. Does this mean you pick a random string and XOR against it before computing each hash value, or something like that? It would seem there could still be some way to discover the exact hash function being used by means of timing attacks (you can tell when an input has caused a hash collision because the response time might be slightly slower). But I need to read the paper referred to.
Surely by the same argument the people running a simulation of us are themselves being simulated along with their simulation, and that by another group of people who are themselves living in a simulation created by an even more advanced civilization...
It's turtles all the way down!
I hear that Sealand makes a profit... would it be possible to stick an aircraft carrier in the middle of the ocean and stick web servers on it? Perhaps the jurisdictional issues would be more complex than with Sealand.
That may be true... but if people calculate cost per week why isn't the unit of electricity billing the kilowatt-week? Every other electric device gives power consumption (or power output, if you prefer) in watts or kilowatts. Better to say 'power consumption N kilowatts, which means 168 * N kilowatt hours per week'. Then at least you are giving the sane units as well.
Engine power is often given in horsepower but that's a crazy and obsolete unit - move to kW now! </metric-troll>
I think that in practice it is going the other way - the various free languages which compile to Java bytecode are _more_ featureful than Sun's efforts, not less.
The story contains a typo - the power output is about 2.6kW, not 2.6kWh (which would be a measure of energy).
Of course, if you are like the freezer manufacturer who quoted their product's power consumption in 'kilowatt hours per week', you could say that the scooter produces 2.6kWh per hour...
Of course it brings them a revenue stream. Semi-legal file sharing networks are one of the main drivers behind broadband. The 'Time Warner' part of the company may not like it, but the 'AOL' half might think that if piracy is going to happen anyway, you might as well try to make a buck off it.
I think that any serious Internet user would not use AOL broadband however... that's the problem.
AFAIK all the changes in Java 1.5 are in the compiler - the bytecode has the same format as before. (At least this is the impression I get from the Slashdot comments.) So it's not a question of whether to change the JVM.
What's interesting is that many of the new features were already available in other languages that compile down to JVM bytecode, such as Generic Java and Nice. But people seem reluctant to use new features until they're in an officially blessed Sun release, even if the generated code is 100% Java-compatible.
First rule of computer journalism: no technology exists until Microsoft invents it.
Corollary: most language features are too obscure, complex and dangerous, until Java adopts them.
Lord of the Rings has now officially jumped the shark - musical episode. Next expect to see Frodo's long-lost cousin Bodo and his hip new catchprase 'it's bodocious'!. They tell me that Will Smith will be making a guest appearance as Sauron's evil twin.
Can people still log in and do their work while the cluster is running jobs?
Yes but you say that trucks, cars and vans are vehicles. You do not call them 'vehicle products'.