Shush. People think these new apps are all better, and that keeps the attention on them. Meanwhile we can quietly get everything off gnutella and laugh at them.
It's apparently so much faster (I never find BT downloads any better than gnutella or fasttrack, but apparently it's very fast) that all the copyright infringers now want to use it.
It's more of a protocol than a network, each swarm is effectively a separate network, but it is distinct from http. After all, gnutella does the actual downloading over http.
Or why not gnunet? IME gnunet works better than ANTs, it has decent download speeds and a proper search function unlike freenet's goofy keys system. The selection of files isn't that wide, but that will improve if more people use it.
You don't need a non-python core. Write a python core, use wxpython or pykde or anything you like to do the gui. You can't seriously claim writing widgets in swing is easier than the toolkits you can use from python, swing is goofy, probably because it has (next to) no competition. Plugins for python are very easy, certainly no harder than java, yes security could do with some work, but as you say the plugin can do anything the app can, so plugins have to be trusted.
Java UIs are not useable on my system, 800mhz from 3 years ago (but it wasn't top of the range then, yes). The fact is gui java is much slower than gui python. It's not java that's the problem, it's swing. In fact it's not even swing that's the problem, it's sun's implementation of it. But that doesn't matter; practically, all gui java apps are dog slow. (CLI java is no better or worse than any other language, but cli apps are generally not desktop programs) Graphical python runs fast, because pygtk or pyqt is just a thin wrapper on the fast c funtions. Wheras swing does every pixel in java. And I'm pretty sure python is JIT, and if you're convinced the java interpreter is faster then you can compile python to java bytecode with jython and run it in there. Most people don't, because the JVM is a huge memory hog and no faster performance-wise.
On my system, no speed demon but works fine, azureus takes 20-30 seconds to start up, and 1-2 seconds to display a menu after you click on it, with no other programs running. That's why.
I think he wants it to increase, so there's a simple way to do that. Upload is capped at 2x the download you get from the same peer. You proxy stuff they're uploading if there's no file you actually want to get from them. When requesting a file you simply upload a few blocks of something popular to get it started - it doesn't need to be something they need, it just needs to show that you're willing to upload. Going more sophisticated, you could have a credit system for uploading based on digital signatures. You send a signature every time you receive a block, saying who you received it from. Not sending signatures gets a big mark against you from the person who sent you the block. Sending too many signatures shows you as downloading far more than you upload, so people no longer upload to you. I think it should be possible to arrange for such signatures to propagate over a gnutella-like network, gaining the advantages of bittorrent's enforced share ratios without the necessity of centralisation.
1. Have you ever tried to implement gnutella2? The reference implementation link on the site doesn't work, and there are precious few implementations around, and even fewer non-windows ones, none of them well documented. So you basically have to do it from scratch. The protocol is nicely documented, yes, but it's still a lot harder than with bittorrent.
2. The whole point your parent post was making is that advanced swarming like bittorrent does relies on centralisation. You can only have a proper retribution system if you have a centralised tracker to track it. The only way to decentralise bittorrent is to remove that swarming payback ability - and if you do that, it becomes just another gnutella clone.
3. It's odd that you complain about people making new protocols when you've just been recommending gnutella2, which is Mike going off on his own because he's fed up with real gnutella rather than working on it to make it better.
4. Merging protocols is really hard. It's easier to write a network from scratch, really. There are decentralised networks that try to have payback working, most of the f2f systems depend on it, and also have the advantage of anonymity. But they're not popular yet.
Oh yeah, and the intention of that was to, erm, look pretty or something. The intent of having that lump of metal move at high velocity is to kill people. Claiming they're not designed to kill people is like claiming http isn't designed to transfer web pages, it's designed to copy bits from one computer to another.
I appreciate attempts to increase the popularity of science - but I wonder if the choice of categories is rather shortsighted. Are these the areas that are important long-term, or simply the trendiest parts of science at the moment? I wonder if some more traditional areas would benefit more from a new award - it's quite easy to get people excited about nanotech, less so for some other areas.
The basic idea is there are two measurements you can carry out, but making the wrong one will destroy the information. The sender generates a random bitstream and encodes it using random choices of which method. The receiver makes one measurement at random for each photon they get. Then, after the measurements have been made and the photons destroyed, the sender tells the receiver which encodings were used and the receiver tells the sender which bits are correct. These bits are random - the sender cannot choose which they are - but it doesn't matter, since they're only going to be used as a key. Once the key has been exchanged like this, then the sender encrypts. This is a practical delay like with diffie-hellman-merkle - the receiver has to be online to recieve the key before the message itself can be sent.
I'm stating that objects do not make people better. The topic was Intelligent Design is Psuedoscience and you seemed to be claiming (though perhaps not) that inventions make "life" better, possibly implying that humans have evolved due to applied science.
I'm claiming that humanity is better as a result of science and inventions. We have easier lives, yes, but I also think we are better as a race - more civilised if you will, able to achieve more.
I say that people have not and will not change for milennia. Likewise the abolition of slavery had no universal effect on humanity either. It was a benefit for slaves and would-be slaves. No one else. As glad as you and I may be about it, that doesn't make us better.
I disagree. I think that no longer having slaves makes humanity better as a species. It may not make you or me better off ourselves, but it makes humanity as a whole better.
Not once did I suggest or imply that murdering is acceptable. But wre we better if we exist indefinitely?
You implied that it is better not to live. I hold that human life is fundamentally good and increasing it is generally good. I don't think there can be any reasonable moral system which doesn't believe this.
The common thread of these is that not mine, not yours, not anyone's emotional state has any effect on humanity as a species. Emotional state can be modified. What if murdering people at a young age made me happy?
Then it would be good, apart from the fact that by killing people you would cause more suffering for others. I can accept the view that making people happier doesn't make humanity any better, but I think things like the abolition of slavery go beyond merely making people happy, and do make us better.
By what standard can we claim that slavery was bad? What is our basis for morality?
That's a different debate. I can't think of any moral viewpoint that would imply slavery was good. Yes that's an appeal to the majority, but I doubt we have the same basis for morality anyway, and I can't defend mine as any better than yours.
Whilst a free reference implementation would be nice, that's enough. Implementing a 1236 page specification is a helluva lot easier than working out the document format by reverse engineering like has to be done with.doc.
This page is still appearing in a horrible gold/brown shade, so I think slashdot just ignores its readers completely. After all, we keep reading and commenting, giving them their ad views, so why should they change?
By that logic how did we ever get to the relational model? After all, flat-file is good enough for separating things from each other and finding them by their properties
No, that's not what he said. If you're child's in a big room, let's say swimming or something, they're safer by you removing all the non-former-molestors from the room than you removing all the former molestors. I can easily believe that.
Quite simple solution to that. All cops carry the readers, and can stop anyone with a gun and check it. In fact, if it's RFID they might not have to even stop you. Just if you seem to be carrying one and their reader doesn't trigger, they stop you. Sure, it's easy to remove the tag - but if you do, as soon as you walk by a policeman you're going to get stopped. This would mean most criminals who did remove the tags would be arrested before they used their untraceable guns.
Well, I hold that I'm better off because I have a far better working model of the world I live in. I know to a much greater extent what the universe is like. But that's much harder to claim it makes one better, wheras I can't believe anyone would seriously argue against the facts of the improvements to life.
There's no question of perspective, unless your "perspective" is so screwed up you think living is a bad thing. Do you really think it's better for people to die young, that someone looking to improve the world should go out and kill people?
A minute ago you were saying to look at the betterment of humanity rather than the comfort of individuals. I hold it to be self-evident that humanity is better without slavery. It might be natural, but that doesn't make it good. No, not all slaves were treated poorly, but on the whole slaves were.
WTF? I can't believe you're even trying to make this claim. You seem to have some romantic view that it wasn't that bad to be a slave. And tell me how any invention has made workers more slavelike. Go on, even one.
Why do you say we have no basis for morality? Religion and morality are completely orthogonal. It's hard to try and claim something's better or worse for humanity on anything other than moral grounds.
No, but there were slaves in almost every nation prior to industrialisation, and you can trace a pretty close relationship between when a machine became available to do the largest job in a country and when it abolished slavery.
Yes, but what they're doing is dropping from 1080 to 540 and then upscaling to 720, which loses far more data than straight downscaling to 720. I can understand doing it because downscaling is far more cpu intensive than upscaling, but it will lose a lot of data.
Slashdot Mirror
Shush. People think these new apps are all better, and that keeps the attention on them. Meanwhile we can quietly get everything off gnutella and laugh at them.
It's apparently so much faster (I never find BT downloads any better than gnutella or fasttrack, but apparently it's very fast) that all the copyright infringers now want to use it.
It's more of a protocol than a network, each swarm is effectively a separate network, but it is distinct from http. After all, gnutella does the actual downloading over http.
Or why not gnunet? IME gnunet works better than ANTs, it has decent download speeds and a proper search function unlike freenet's goofy keys system. The selection of files isn't that wide, but that will improve if more people use it.
You don't need a non-python core. Write a python core, use wxpython or pykde or anything you like to do the gui. You can't seriously claim writing widgets in swing is easier than the toolkits you can use from python, swing is goofy, probably because it has (next to) no competition. Plugins for python are very easy, certainly no harder than java, yes security could do with some work, but as you say the plugin can do anything the app can, so plugins have to be trusted.
Java UIs are not useable on my system, 800mhz from 3 years ago (but it wasn't top of the range then, yes). The fact is gui java is much slower than gui python. It's not java that's the problem, it's swing. In fact it's not even swing that's the problem, it's sun's implementation of it. But that doesn't matter; practically, all gui java apps are dog slow. (CLI java is no better or worse than any other language, but cli apps are generally not desktop programs) Graphical python runs fast, because pygtk or pyqt is just a thin wrapper on the fast c funtions. Wheras swing does every pixel in java. And I'm pretty sure python is JIT, and if you're convinced the java interpreter is faster then you can compile python to java bytecode with jython and run it in there. Most people don't, because the JVM is a huge memory hog and no faster performance-wise.
On my system, no speed demon but works fine, azureus takes 20-30 seconds to start up, and 1-2 seconds to display a menu after you click on it, with no other programs running. That's why.
I think he wants it to increase, so there's a simple way to do that. Upload is capped at 2x the download you get from the same peer. You proxy stuff they're uploading if there's no file you actually want to get from them. When requesting a file you simply upload a few blocks of something popular to get it started - it doesn't need to be something they need, it just needs to show that you're willing to upload. Going more sophisticated, you could have a credit system for uploading based on digital signatures. You send a signature every time you receive a block, saying who you received it from. Not sending signatures gets a big mark against you from the person who sent you the block. Sending too many signatures shows you as downloading far more than you upload, so people no longer upload to you. I think it should be possible to arrange for such signatures to propagate over a gnutella-like network, gaining the advantages of bittorrent's enforced share ratios without the necessity of centralisation.
2. The whole point your parent post was making is that advanced swarming like bittorrent does relies on centralisation. You can only have a proper retribution system if you have a centralised tracker to track it. The only way to decentralise bittorrent is to remove that swarming payback ability - and if you do that, it becomes just another gnutella clone.
3. It's odd that you complain about people making new protocols when you've just been recommending gnutella2, which is Mike going off on his own because he's fed up with real gnutella rather than working on it to make it better.
4. Merging protocols is really hard. It's easier to write a network from scratch, really. There are decentralised networks that try to have payback working, most of the f2f systems depend on it, and also have the advantage of anonymity. But they're not popular yet.
Oh yeah, and the intention of that was to, erm, look pretty or something. The intent of having that lump of metal move at high velocity is to kill people. Claiming they're not designed to kill people is like claiming http isn't designed to transfer web pages, it's designed to copy bits from one computer to another.
I appreciate attempts to increase the popularity of science - but I wonder if the choice of categories is rather shortsighted. Are these the areas that are important long-term, or simply the trendiest parts of science at the moment? I wonder if some more traditional areas would benefit more from a new award - it's quite easy to get people excited about nanotech, less so for some other areas.
The basic idea is there are two measurements you can carry out, but making the wrong one will destroy the information. The sender generates a random bitstream and encodes it using random choices of which method. The receiver makes one measurement at random for each photon they get. Then, after the measurements have been made and the photons destroyed, the sender tells the receiver which encodings were used and the receiver tells the sender which bits are correct. These bits are random - the sender cannot choose which they are - but it doesn't matter, since they're only going to be used as a key. Once the key has been exchanged like this, then the sender encrypts. This is a practical delay like with diffie-hellman-merkle - the receiver has to be online to recieve the key before the message itself can be sent.
So? X11 is part of GNU/Linux, it says as much in the manifesto, yet X11 is not GNU.
I'm claiming that humanity is better as a result of science and inventions. We have easier lives, yes, but I also think we are better as a race - more civilised if you will, able to achieve more.
I say that people have not and will not change for milennia. Likewise the abolition of slavery had no universal effect on humanity either. It was a benefit for slaves and would-be slaves. No one else. As glad as you and I may be about it, that doesn't make us better.
I disagree. I think that no longer having slaves makes humanity better as a species. It may not make you or me better off ourselves, but it makes humanity as a whole better.
Not once did I suggest or imply that murdering is acceptable. But wre we better if we exist indefinitely?
You implied that it is better not to live. I hold that human life is fundamentally good and increasing it is generally good. I don't think there can be any reasonable moral system which doesn't believe this.
The common thread of these is that not mine, not yours, not anyone's emotional state has any effect on humanity as a species. Emotional state can be modified. What if murdering people at a young age made me happy?
Then it would be good, apart from the fact that by killing people you would cause more suffering for others. I can accept the view that making people happier doesn't make humanity any better, but I think things like the abolition of slavery go beyond merely making people happy, and do make us better.
By what standard can we claim that slavery was bad? What is our basis for morality?
That's a different debate. I can't think of any moral viewpoint that would imply slavery was good. Yes that's an appeal to the majority, but I doubt we have the same basis for morality anyway, and I can't defend mine as any better than yours.
Whilst a free reference implementation would be nice, that's enough. Implementing a 1236 page specification is a helluva lot easier than working out the document format by reverse engineering like has to be done with .doc.
This page is still appearing in a horrible gold/brown shade, so I think slashdot just ignores its readers completely. After all, we keep reading and commenting, giving them their ad views, so why should they change?
Nope. If it's going to be completely secure, you need as much key as message - a 100 bit key can only encrypt 100 bits. So this will be useful.
By that logic how did we ever get to the relational model? After all, flat-file is good enough for separating things from each other and finding them by their properties
No, that's not what he said. If you're child's in a big room, let's say swimming or something, they're safer by you removing all the non-former-molestors from the room than you removing all the former molestors. I can easily believe that.
Quite simple solution to that. All cops carry the readers, and can stop anyone with a gun and check it. In fact, if it's RFID they might not have to even stop you. Just if you seem to be carrying one and their reader doesn't trigger, they stop you. Sure, it's easy to remove the tag - but if you do, as soon as you walk by a policeman you're going to get stopped. This would mean most criminals who did remove the tags would be arrested before they used their untraceable guns.
AIUI quantum fluctuations allow particles to appear at random. Given long enough that would result in the appearance of a universe.
But I always fail them
There's no question of perspective, unless your "perspective" is so screwed up you think living is a bad thing. Do you really think it's better for people to die young, that someone looking to improve the world should go out and kill people?
A minute ago you were saying to look at the betterment of humanity rather than the comfort of individuals. I hold it to be self-evident that humanity is better without slavery. It might be natural, but that doesn't make it good. No, not all slaves were treated poorly, but on the whole slaves were.
WTF? I can't believe you're even trying to make this claim. You seem to have some romantic view that it wasn't that bad to be a slave. And tell me how any invention has made workers more slavelike. Go on, even one.
Why do you say we have no basis for morality? Religion and morality are completely orthogonal. It's hard to try and claim something's better or worse for humanity on anything other than moral grounds.
No, but there were slaves in almost every nation prior to industrialisation, and you can trace a pretty close relationship between when a machine became available to do the largest job in a country and when it abolished slavery.
Yep. And that's also true.
Yes, but what they're doing is dropping from 1080 to 540 and then upscaling to 720, which loses far more data than straight downscaling to 720. I can understand doing it because downscaling is far more cpu intensive than upscaling, but it will lose a lot of data.