I can play that game too ya know. Linus' opinion changes from year to year. I can quote him saying that binary modules are something he'll sue you for. Then I can quote him saying that binary modules are ok, so long as they are distributed seperately. Then I can quote him saying that binary modules are evil and if you run them you're an idiot. Then I can quote him saying that binary modules are great, but he doesn't know what their legal standing is. Then I can quote him saying that he knows what their legal standing is, they're not legal but he doesn't care. Then I can quote him saying that he's looked into it and just because you've written a kernel module for Linux it doesn't mean it is a derived work of Linux. Then I can quote him saying the exact opposite of that. Then I can quote him saying that their evil again. Then I can quote him saying that they are a-ok.
Most recently (this year) he has been quoted as saying that if a binary module has a "life of its own beyond Linux" then it's ok and he morally doesn't feel it is a derivative work. Maybe next week he'll claim something else, but at the moment, that's his opinion.
It really isn't hard. Imagine Evil Inc develops a device that has an embedded signature checking chip that I can't get around. I bought the damn device, I just want to hack it so it doesn't have annoying-feature-X. I asked Evil Inc to fix annoying-feature-X but they just mwa-ha-ha-ed at me. What can I do? Well, it turns out that Evil Inc is running some GPL v3 software on this device. I know this because when I bought it I had to install the software myself. I thought this was kinda strange, after all, this is a consumer product, but I guess they needed to do that to get around the GPL v3 restrictions. If only I could make my own CD, I could insert whatever software I wanted.. hmm, but I'll need their private key so I can sign my binaries and make the device accept it. Right. No point asking Evil Inc nicely, they'll just mwa-ha-ha at me again. Who's the copyright owner for this GPL v3 work? That guy. Ok, I'll just get that guy to sue Evil Inc so they have to give up the key. He says he'll go along, just so long as I'm paying the legal costs (I really hate Evil Inc now, I'm in this to the death, pony up lawyer boy). Ok, so now that guy is telling me that Evil Inc had a cunning plan when they distributed those CDs.. turned out *they* didn't distribute it, Evil-Sub-Company distributed it and that's the only people I can sue. Right-o. Let's sue those bastards. The judge tells me that he can't order Evil-Sub-Company to hand over Evil Inc's private key, because it's not Evil-Sub-Company's private key to hand over. Fair enough. I've asked the judge to pass an injunction against Evil-Sub-Company and prevent them from distributing that guy's software at all. He says he'll do that. I've also asked the judge to award my legal costs and a nice big fat damages cheque. He's agreed to that too. Next time Evil Inc thinks they can subcontract their GPL-violations to a sub company I'll just sue them bastards too. Now we're off to buy a yaught and stock it with hookers. That guy is stoked.
Or do you seriously believe that Linus hasn't consulted with attorneys on this?
See, now that's funny. Linus is the guy who claims that a device driver isn't a derived work of the Linux kernel if it was originally developed for a different operating system and then ported to Linux. This, of course, is not based on any legal principle.. it's just his opinion, but it doesn't stop people from quoting Linus like they're referencing case law.
What's more funny is that when Linus added the "userland exception" to the Linux kernel he was absolutely clear about what he wanted.. he wanted people to be able to write proprietary apps that can run on the Linux kernel. He didn't want people to be able to write proprietary extensions to the Linux kernel. Now he's changed his mind because his "pragmatism" is telling him that graphics card manufacturers will never open source their drivers and he really wants all those pretty 3d games.
It's a trilogy, Manifold Time, Manifold Space and Manifold Origin. I thought the last one was the worst and the first one was the best, but you might think otherwise.
You mean other than every journalist who has refused to give up a source when ordered by a judge? Or the journalists who publish classified information?
I don't know where you people get your idea of laws from, but it isn't illegal for a private citizen to violate your privacy. If I read your diary without your permission you can't have me arrested. Jesus.
pollution is exactly the problem. When tracing hackers you can't trust that data. ISPs don't have good security. Hackers connect to things like wireless access points, jump through unsecured web servers, back through unsecured home PCs, etc, etc. Supposing you get ISPs to retain connection times for dialup users, what the hell does that tell you? Do you honestly think a hacker is connecting via a dialup? Connection times for DSL? Cable? Forget about it. So what are ISPs supposed to do? Run intrusion detection software? What do they do when they spot an attack? Block it? Tell the user? Of course, so what's the point of retaining the data again? If the network was secure you wouldn't need to retain the data. Great catch-22 there. Maybe you suspect that ISPs should log every packet in and out of their network? There isn't enough harddrives in a Google data centre to do that!
No, the only conceivable use for this stuff is to log what web pages the current government's political opponents are visiting, or to trace back naive file sharing users for the RIAA.
actually I just thought of the coolest way it could possibly work and went with that:) Nah. I obviously misread something 10 years ago when I was interested in this stuff and it's just festered in my mind since then. I can do a similar thing for video if you like. How I really think video compression works: every X number of frames emit a keyframe, find the differences between the next two consecutive frames, use the current motion vector model to predict those changes, calculate the error, modify the model to reduce the error, repeat for the next N frames, emit the motion vector model. How I wish video compression worked: detect the transitions between scenes in a film, break the film up into scenes, for each scene create a model of the objects in the scene, optimize the movement and representation of objects to reduce the error between frames, emit the models and scene transitions, reusing scene models if you happen to transistion back to a scene you were previously in (good for intercutting). Ultimately you'd just end up with a quake engine to playback the film on the client:)
if the people who make legislation actually had some idea about the problem the legislation was supposed to solve? Or, ya know, refused to vote for something they didn't understand? Just a simple "introduction to hacking" course would help so many of them recognise that data retention aint going to help you track a hacker. I hate to say it, but I honestly think the only way to "police the Internet" is to give policing powers to a police force. Those powers would include the right to enter systems without permission, install logging software, etc. Question is, who would you want to trust with that much power?
Extracting those discrete waveforms is called modelling, and how good your model is defines how much you can extract. So, for example, if you have a guitar model and you apply it to a track that has nothing but guitar on it, your model is going to match a hell of a lot better than if you have vocals on the track with the guitar. Even if you apply your guitar model, subtract the resulting approximation to get the error and then apply your vocals model, you're still not going to get as good results as you would if the tracks were seperate because of the noise accumulation. This is the kind of stuff FLAC does for lossless compression. I've not heard of people doing similar stuff for lossy compression, thus my original question, but there's probably a good reason.
You can't rationalize "dick squat" ok? It's an idiom. Much like when someone calls you a "troll" you can't object on the grounds that you are tall, it doesn't work that way.
All I'm saying, is that if you were to combine the mixing process and the compressing process, wouldn't the compressor have more information to play with? Mixing throws away information that the compressor could use, doesn't it?
Yeah, I guess all I was thinking was that mixing throws away information. If you were to get the mixer and the compressor to work together you'd get better results.
Call me crazy here, but wouldn't you wanna take advantage of *both* lossy and lossless compression techniques to get the smallest file with the best sound quality? It's not one or the other.. is it?
Thanks! Relating this back to my question, if you have three tracks and at time X there is a 700hz sound on track 1, a 703hz sound on track 2 and a 900hz sound on track 3 and track 1 is louder than track 2, then isn't it going to be really easy to determine that you can drop the bits from track 2? Easier, than say, trying to detect that a particular waveform in the mix of tracks 1, 2 and 3 was created from three seperate tracks and then determining that that you can drop the sound from track 2 and represent the whole frame as a 700hz signal and a 900hz signal? And, similarly, isn't it likely that a seperation of tracks into individual instruments (as it is usually done in making a recording) will be ameanable to this kind of analysis?
So, is there tools that already do this? Or is the guy who is trying to encode his own music (which is lying on his harddrive, uncompressed, in seperate instrument tracks) only ever going to get a compressed file which is as good as the d00d running Audio Grabber on a store bought CD?
Slashdot Mirror
Dude, go read the kernel mailing list. You obviously don't.
I can play that game too ya know. Linus' opinion changes from year to year. I can quote him saying that binary modules are something he'll sue you for. Then I can quote him saying that binary modules are ok, so long as they are distributed seperately. Then I can quote him saying that binary modules are evil and if you run them you're an idiot. Then I can quote him saying that binary modules are great, but he doesn't know what their legal standing is. Then I can quote him saying that he knows what their legal standing is, they're not legal but he doesn't care. Then I can quote him saying that he's looked into it and just because you've written a kernel module for Linux it doesn't mean it is a derived work of Linux. Then I can quote him saying the exact opposite of that. Then I can quote him saying that their evil again. Then I can quote him saying that they are a-ok.
Most recently (this year) he has been quoted as saying that if a binary module has a "life of its own beyond Linux" then it's ok and he morally doesn't feel it is a derivative work. Maybe next week he'll claim something else, but at the moment, that's his opinion.
in fact, they want a trusted computing world because they know how much raw, unaccountable Orwellian power it will give them.
Not to mention booz and hot chicks.
Sure. You can also prevent black people from using your software if you like. Oh, and fascists, let's stop them too.
Why? It's not like RMS is going to sue them.. that'd be too establishment, man.
It really isn't hard. Imagine Evil Inc develops a device that has an embedded signature checking chip that I can't get around. I bought the damn device, I just want to hack it so it doesn't have annoying-feature-X. I asked Evil Inc to fix annoying-feature-X but they just mwa-ha-ha-ed at me. What can I do? Well, it turns out that Evil Inc is running some GPL v3 software on this device. I know this because when I bought it I had to install the software myself. I thought this was kinda strange, after all, this is a consumer product, but I guess they needed to do that to get around the GPL v3 restrictions. If only I could make my own CD, I could insert whatever software I wanted.. hmm, but I'll need their private key so I can sign my binaries and make the device accept it. Right. No point asking Evil Inc nicely, they'll just mwa-ha-ha at me again. Who's the copyright owner for this GPL v3 work? That guy. Ok, I'll just get that guy to sue Evil Inc so they have to give up the key. He says he'll go along, just so long as I'm paying the legal costs (I really hate Evil Inc now, I'm in this to the death, pony up lawyer boy). Ok, so now that guy is telling me that Evil Inc had a cunning plan when they distributed those CDs.. turned out *they* didn't distribute it, Evil-Sub-Company distributed it and that's the only people I can sue. Right-o. Let's sue those bastards. The judge tells me that he can't order Evil-Sub-Company to hand over Evil Inc's private key, because it's not Evil-Sub-Company's private key to hand over. Fair enough. I've asked the judge to pass an injunction against Evil-Sub-Company and prevent them from distributing that guy's software at all. He says he'll do that. I've also asked the judge to award my legal costs and a nice big fat damages cheque. He's agreed to that too. Next time Evil Inc thinks they can subcontract their GPL-violations to a sub company I'll just sue them bastards too. Now we're off to buy a yaught and stock it with hookers. That guy is stoked.
Or do you seriously believe that Linus hasn't consulted with attorneys on this?
See, now that's funny. Linus is the guy who claims that a device driver isn't a derived work of the Linux kernel if it was originally developed for a different operating system and then ported to Linux. This, of course, is not based on any legal principle.. it's just his opinion, but it doesn't stop people from quoting Linus like they're referencing case law.
What's more funny is that when Linus added the "userland exception" to the Linux kernel he was absolutely clear about what he wanted.. he wanted people to be able to write proprietary apps that can run on the Linux kernel. He didn't want people to be able to write proprietary extensions to the Linux kernel. Now he's changed his mind because his "pragmatism" is telling him that graphics card manufacturers will never open source their drivers and he really wants all those pretty 3d games.
presumption of privacy restricts what governments can do. It doesn't restrict what I can do. Laws don't work the way you're told on tv they do ok?
Fuck off freak.
It's a trilogy, Manifold Time, Manifold Space and Manifold Origin. I thought the last one was the worst and the first one was the best, but you might think otherwise.
No and no. If the government does it, it's illegal, if citizens to do it to each other it's just bad manners.
You mean other than every journalist who has refused to give up a source when ordered by a judge? Or the journalists who publish classified information?
Yeah? It's called intent.
I don't know where you people get your idea of laws from, but it isn't illegal for a private citizen to violate your privacy. If I read your diary without your permission you can't have me arrested. Jesus.
Like, say, they believe an attack originated from that machine?
pollution is exactly the problem. When tracing hackers you can't trust that data. ISPs don't have good security. Hackers connect to things like wireless access points, jump through unsecured web servers, back through unsecured home PCs, etc, etc. Supposing you get ISPs to retain connection times for dialup users, what the hell does that tell you? Do you honestly think a hacker is connecting via a dialup? Connection times for DSL? Cable? Forget about it. So what are ISPs supposed to do? Run intrusion detection software? What do they do when they spot an attack? Block it? Tell the user? Of course, so what's the point of retaining the data again? If the network was secure you wouldn't need to retain the data. Great catch-22 there. Maybe you suspect that ISPs should log every packet in and out of their network? There isn't enough harddrives in a Google data centre to do that!
No, the only conceivable use for this stuff is to log what web pages the current government's political opponents are visiting, or to trace back naive file sharing users for the RIAA.
actually I just thought of the coolest way it could possibly work and went with that :) Nah. I obviously misread something 10 years ago when I was interested in this stuff and it's just festered in my mind since then. I can do a similar thing for video if you like. How I really think video compression works: every X number of frames emit a keyframe, find the differences between the next two consecutive frames, use the current motion vector model to predict those changes, calculate the error, modify the model to reduce the error, repeat for the next N frames, emit the motion vector model. How I wish video compression worked: detect the transitions between scenes in a film, break the film up into scenes, for each scene create a model of the objects in the scene, optimize the movement and representation of objects to reduce the error between frames, emit the models and scene transitions, reusing scene models if you happen to transistion back to a scene you were previously in (good for intercutting). Ultimately you'd just end up with a quake engine to playback the film on the client :)
if the people who make legislation actually had some idea about the problem the legislation was supposed to solve? Or, ya know, refused to vote for something they didn't understand? Just a simple "introduction to hacking" course would help so many of them recognise that data retention aint going to help you track a hacker. I hate to say it, but I honestly think the only way to "police the Internet" is to give policing powers to a police force. Those powers would include the right to enter systems without permission, install logging software, etc. Question is, who would you want to trust with that much power?
Extracting those discrete waveforms is called modelling, and how good your model is defines how much you can extract. So, for example, if you have a guitar model and you apply it to a track that has nothing but guitar on it, your model is going to match a hell of a lot better than if you have vocals on the track with the guitar. Even if you apply your guitar model, subtract the resulting approximation to get the error and then apply your vocals model, you're still not going to get as good results as you would if the tracks were seperate because of the noise accumulation. This is the kind of stuff FLAC does for lossless compression. I've not heard of people doing similar stuff for lossy compression, thus my original question, but there's probably a good reason.
You can't rationalize "dick squat" ok? It's an idiom. Much like when someone calls you a "troll" you can't object on the grounds that you are tall, it doesn't work that way.
All I'm saying, is that if you were to combine the mixing process and the compressing process, wouldn't the compressor have more information to play with? Mixing throws away information that the compressor could use, doesn't it?
Yeah, I guess all I was thinking was that mixing throws away information. If you were to get the mixer and the compressor to work together you'd get better results.
Call me crazy here, but wouldn't you wanna take advantage of *both* lossy and lossless compression techniques to get the smallest file with the best sound quality? It's not one or the other.. is it?
Thanks! Relating this back to my question, if you have three tracks and at time X there is a 700hz sound on track 1, a 703hz sound on track 2 and a 900hz sound on track 3 and track 1 is louder than track 2, then isn't it going to be really easy to determine that you can drop the bits from track 2? Easier, than say, trying to detect that a particular waveform in the mix of tracks 1, 2 and 3 was created from three seperate tracks and then determining that that you can drop the sound from track 2 and represent the whole frame as a 700hz signal and a 900hz signal? And, similarly, isn't it likely that a seperation of tracks into individual instruments (as it is usually done in making a recording) will be ameanable to this kind of analysis?
So, is there tools that already do this? Or is the guy who is trying to encode his own music (which is lying on his harddrive, uncompressed, in seperate instrument tracks) only ever going to get a compressed file which is as good as the d00d running Audio Grabber on a store bought CD?
True. I do know there's no "and" in "dick squat" though. Why didn't you choose an insult from your native language?