Note: This may look like a flame. But it's not. I'm right.
You're wrong. Very wrong. Why?
From Computer Architecture by Hennessey and Patterson, page 13:
Cost of die = Die Area ^ 4.
So, having 136 times more area per chip would give you a processor that costs about 136 ^ 4 = 300 million times more per chip. Umm, yeah. that's a lot of money.
Why does this happen? Because there are errors in your process, always. So you can't just not use those processors. If you want to know more, email me as slashdot at danbentley daught com.
-Dan
O(10^3)?!?! Being big o to a constant means any number. That's meaningless expression. Not to mention, I think it's understand that factoring 15 is still far away from useful application. Basically, what this poster says is, there are problems, because there are problems.
Insightful?
Their hardware is optimized to make the x86 the one that is emulated. They have special registers that makes the optimizations work best their. Comparing two pointers to know if they can be the same or not while dynamically recompiling, etc.
Wheels gave way to tires, and other things. How come we don't have pipes that are multi-way? How come I can't build distributed multi-way large programs from small chunks? Pipes aren't going anywhere.
Don't forget that NIL has significance in LISP, hence giving cool hackerness too it. Maybe even better, have it stand for NIL's a Consistent License. Hehe.
Dude, if you really want, make your own.
The first one should be easy. One chapter in "The Unix Programming Environment" makes an interpreter/compiler for a language almost as powerful as basic, just different.
It would be a great exercise to make your own. You'd learn a lot about compilers and languages. Also, because you maintain it, you have a lot of flexibility in what it does, and how it acts. You'd also, I'm sure, get mad props on slashdot and maybe after you posted about it later, win converts for people who wanted a quick language to do what you want to do.
Oh? Really? You think you could do that?
Learn about CRISP by Ditzel et al. from Bell Labs. Damn near 20 years ago, amazingly efficient stack based RISC processor.
Linux may not be hard to use, but it sure is DIFFERENT. In an emergency, there's no time to learn something different. I wasn't saying Linux was bad. Just not appropriate. Often, people paint Linux as a panacea on this site. It isn't. This is one of those places where it would be detrimental.
Well... This *may* be true of GNU/Linux. To paraphrase the Bard "were it so, 'twas a grievous fault, and grievously hath GNU/Linux answered it."
So PLEASE, STOP using GNU/Linux. Wait until GNU/Hurd comes out! RSN! It will solve problems.
You want a micro-kernel production level OS, you got it.
You want a better GUI, with a built-in Office program that's both monolithic and component-ized? You got it!
You want a girlfriend? Ummm, sure, we can build that in! Version 1.5!
Do you want someone killed? Well, yeah, that's what Free Software is all about! RMS the hit-man, version 2!
Supporting GNU/Linux hurts GNU/Hurd. Supporting any other OS is evil.
GNU/Hurd (in development) forever!
Note: If you do help...
Leave advocacy for later!
Do not push Linux. If people can't use it, it won't help right now.
There probably won't be 802.11b links. This will probably networking Pentium class machines so people can do data entry. They will almost certainly run Windows.
This is good, because this will make sense to the people who can directly help. If you go there, be prepared to deal with these systems and don't bitch about it.
Not to mention, you'll give Linux a bad name if you still push it in this time of crisis.
-Dan
The site is down, so pardon me if I'm thinking of the wrong thing. But if these are the same as the famous internet maps, they're damn cool.
I remember when Bill Cheswick started making them. At the time, both he and my dad were at Bell Labs. He even printed two different views out for me to hang on my dorm room hall. But these have interesting research aspects in small parts. The first is mapping the internet. Damn. He has daily logs going way back, and on his website has videos of the IP's of Bosnia blinking on and off during the days of bombing them. (Google search for him). It's incredible.
But the visualization has interesting problems. My dad did some interesting work on the computational geometry structure that allowed for these things to be visualized. They have various springiness between all the connections that eventually reach the state that's displayed. The colors can be assigned in various ways (the one I remember is that each different part of the IP address is a component of RGB). It's an amazing effort that's a lot less hype and a lot more science than we might think.
For more info, the book Programming Pearls by Jon Bentley has mention of it (my copy is packed away, otherwise I would cite it), if obscure. But if you want to do cool stuff like this, it's important to remember that it's not just scripts, cs theory can help everywhere! A useful lesson to take to heart.
The bash on Slashdot in the article, that responses are flames and Linux-centric is unfairly leveled. Yes, people are flamed if they don't like tha tLinux isn't easy to use. In fact,/. isn't agnostic. It's a bunch of bigoted assholes who want everyone to use Linux. Or at least, you can hear that. It's very intimidating to newbies.
BUT, that's what Slashdot, THE COMMUNITY, has decided to be. Those AREN'T journalists. It's not CmdrTaco who's coming down and flaming people. There even exists many legitimate criticisms of Slashdot and Slashdot's journalism. But this guy, in confusing the whole issue, just comes off as stupid.
If you're going to say Slashdot is harsh, say it in an article about the environment of weblog.
If you're going to say journalism is bad, get on them for the all the times they've been had by hoaxes and post press releases for companies submitted by people with the same username as the company.
But if you're going to criticize/., at least do it fairly and in the right forum. Otherwise, you come off seeming like an idiot who doesn't understand what, exactly, he's writing about or what his subject is.
This is an accomplishment, don't get me wrong. But 10 years is not, no matter what the dot-commers would have you think, a huge milestone in computing. This is just the continued maturation of a product getting better all the time.
Congrats to everybody who's changing the way software is created!
Interesting idea, but it doesn't quite work. True, either sentence could be false, but as long as one and only one is true, the whole statement is self consistent. I believe the idea you're looking for is:
the following sentence is a lie. the preceding sentence is true.
I happen to be a researcher in steganography at the moment. I fear that all this work, while "practical" is not as comprehensive as you might make it sound. If you read IEEE Transactions on Information Theory, you might remember the article a couple months ago: "Quantization Index Modulation" by B. Chen and... umm... Cornwell? Sorry, I don't have it in front of me.
The point is, this article and others have been doning some amazing work on provably good steganography and making some strides in really making stego fit to the information theory model in good ways.
A lot of the papers cited are less "practical" experiments in steganography but rather information theory which has similar issues. The two most interesting were "writing on dirty paper" and "capacity of memory with errors". These were all about similar problems in VERY different areas.
The great thing about theory is that it finds connections you'd never imagine.
If you want to talk about this, my email is dbentley at stanford (it's a university, guess what the TLD is)
FYI: flac (flac.sourceforge.net) is a free lossless audio compression. What this means is that you might only halve the size, but it's a PERFECT reproduction. I'm not sure if any players support it, but a shell script or somesuch could probably decrompress it right before you wanted to play it.
The code quality is horrible, but that might not matter.
I'm doing research on digital watermarking (don't hate me, it's an academic thing, specifically trying to break one scheme). Does anyone know where I can get general purpose code to interact with waves? Like, it will handle all the headers and I can just get the samples?
And this is only sorta OT, because the answer might be Ogg.
This could really help me. If you know of anything, reply.
Too bad this was done in 1993, back when NeXT cubes were plentiful. There's only a finite amount of anything in this world: including space. If we don't recycle stuff, including doing cool stuff and having funn with it, that's no good either. So sit back, and enjoy the pyrotechnics.
Perhaps I'm missing it, but why is programming in Microsoft's new operating system so important? Does Visual Studio.NET offer a better debugger or somesuch?
This page is supposed to be an FUD antidote? What?
This describes what Microsoft USED to do. Microsoft no longer restricts it to their biggest companies (or universities, which for a long time have had access, which noone here seems to realize), but instead allows anyone to look at WinCE code. You can even mess around with it, modify, recompile, as long as it's not for commercial use. This is pretty cool. You can hack with it, play around with it, etc., as long as you don't try to steal Microsoft's code.
Granted, it's not open source by a long shot, but it is a way for Microsoft code to become friendlier.
Oh, and if you check it out, they even allow you to use code in your own. So it's NOT the "Oh-my-god-if-your-seventeenth-cousand-thrice-remo ved-looks-at-this-you-can't-make-anything-more-tha n-shell-scripts-or-Linux-will-be-fucked." Indeed, they're willing to give you ideas.
Shared source isn't what we're into, granted, but it is a lot nicer than we give it credit for. If we're going to be opinionated, let's at least be right.
And what is the success rate for the Athlon's branch prediction?
If you don't know this kind of stuff, then don't criticize Intel. Branch Prediction is hard stuff. If I asked you to do it with 90% accuracy, I bet you couldn't.
Slashdot Mirror
Note: This may look like a flame. But it's not. I'm right.
You're wrong. Very wrong. Why?
From Computer Architecture by Hennessey and Patterson, page 13:
Cost of die = Die Area ^ 4.
So, having 136 times more area per chip would give you a processor that costs about 136 ^ 4 = 300 million times more per chip. Umm, yeah. that's a lot of money.
Why does this happen? Because there are errors in your process, always. So you can't just not use those processors. If you want to know more, email me as slashdot at danbentley daught com.
-Dan
O(10^3)?!?! Being big o to a constant means any number. That's meaningless expression. Not to mention, I think it's understand that factoring 15 is still far away from useful application. Basically, what this poster says is, there are problems, because there are problems.
Insightful?
Are you the Tom Duff that used to work at Bell Labs? In the Unix Room?
-Dan Bentley (son of Jon Bentley)
(@ dbentley (dot stanford edu))
Their hardware is optimized to make the x86 the one that is emulated. They have special registers that makes the optimizations work best their. Comparing two pointers to know if they can be the same or not while dynamically recompiling, etc.
Wheels gave way to tires, and other things. How come we don't have pipes that are multi-way? How come I can't build distributed multi-way large programs from small chunks? Pipes aren't going anywhere.
Don't forget that NIL has significance in LISP, hence giving cool hackerness too it. Maybe even better, have it stand for NIL's a Consistent License. Hehe.
Dude, if you really want, make your own.
The first one should be easy. One chapter in "The Unix Programming Environment" makes an interpreter/compiler for a language almost as powerful as basic, just different.
It would be a great exercise to make your own. You'd learn a lot about compilers and languages. Also, because you maintain it, you have a lot of flexibility in what it does, and how it acts. You'd also, I'm sure, get mad props on slashdot and maybe after you posted about it later, win converts for people who wanted a quick language to do what you want to do.
Oh? Really? You think you could do that?
Learn about CRISP by Ditzel et al. from Bell Labs. Damn near 20 years ago, amazingly efficient stack based RISC processor.
Linux may not be hard to use, but it sure is DIFFERENT. In an emergency, there's no time to learn something different. I wasn't saying Linux was bad. Just not appropriate. Often, people paint Linux as a panacea on this site. It isn't. This is one of those places where it would be detrimental.
Well... This *may* be true of GNU/Linux. To paraphrase the Bard "were it so, 'twas a grievous fault, and grievously hath GNU/Linux answered it."
So PLEASE, STOP using GNU/Linux. Wait until GNU/Hurd comes out! RSN! It will solve problems.
You want a micro-kernel production level OS, you got it.
You want a better GUI, with a built-in Office program that's both monolithic and component-ized? You got it!
You want a girlfriend? Ummm, sure, we can build that in! Version 1.5!
Do you want someone killed? Well, yeah, that's what Free Software is all about! RMS the hit-man, version 2!
Supporting GNU/Linux hurts GNU/Hurd. Supporting any other OS is evil.
GNU/Hurd (in development) forever!
Note: If you do help...
Leave advocacy for later!
Do not push Linux. If people can't use it, it won't help right now.
There probably won't be 802.11b links. This will probably networking Pentium class machines so people can do data entry. They will almost certainly run Windows.
This is good, because this will make sense to the people who can directly help. If you go there, be prepared to deal with these systems and don't bitch about it.
Not to mention, you'll give Linux a bad name if you still push it in this time of crisis.
-Dan
I remember when Bill Cheswick started making them. At the time, both he and my dad were at Bell Labs. He even printed two different views out for me to hang on my dorm room hall. But these have interesting research aspects in small parts. The first is mapping the internet. Damn. He has daily logs going way back, and on his website has videos of the IP's of Bosnia blinking on and off during the days of bombing them. (Google search for him). It's incredible.
But the visualization has interesting problems. My dad did some interesting work on the computational geometry structure that allowed for these things to be visualized. They have various springiness between all the connections that eventually reach the state that's displayed. The colors can be assigned in various ways (the one I remember is that each different part of the IP address is a component of RGB). It's an amazing effort that's a lot less hype and a lot more science than we might think.
For more info, the book Programming Pearls by Jon Bentley has mention of it (my copy is packed away, otherwise I would cite it), if obscure. But if you want to do cool stuff like this, it's important to remember that it's not just scripts, cs theory can help everywhere! A useful lesson to take to heart.
-Dan
BUT, that's what Slashdot, THE COMMUNITY, has decided to be. Those AREN'T journalists. It's not CmdrTaco who's coming down and flaming people. There even exists many legitimate criticisms of Slashdot and Slashdot's journalism. But this guy, in confusing the whole issue, just comes off as stupid.
If you're going to say Slashdot is harsh, say it in an article about the environment of weblog.
If you're going to say journalism is bad, get on them for the all the times they've been had by hoaxes and post press releases for companies submitted by people with the same username as the company.
But if you're going to criticize /., at least do it fairly and in the right forum. Otherwise, you come off seeming like an idiot who doesn't understand what, exactly, he's writing about or what his subject is.
C++ - 15 years
Mac - closing in on 20
x86 - too damn long, more than 20
This is an accomplishment, don't get me wrong. But 10 years is not, no matter what the dot-commers would have you think, a huge milestone in computing. This is just the continued maturation of a product getting better all the time.
Congrats to everybody who's changing the way software is created!
Interesting idea, but it doesn't quite work. True, either sentence could be false, but as long as one and only one is true, the whole statement is self consistent. I believe the idea you're looking for is:
the following sentence is a lie. the preceding sentence is true.
Yeah, I'm sure the code to do nuclear simulations on the top (public) supercomputers in the world is laughably bad.
The point is, this article and others have been doning some amazing work on provably good steganography and making some strides in really making stego fit to the information theory model in good ways.
A lot of the papers cited are less "practical" experiments in steganography but rather information theory which has similar issues. The two most interesting were "writing on dirty paper" and "capacity of memory with errors". These were all about similar problems in VERY different areas.
The great thing about theory is that it finds connections you'd never imagine.
If you want to talk about this, my email is dbentley at stanford (it's a university, guess what the TLD is)
they haven't.
The code quality is horrible, but that might not matter.
-Dan
And this is only sorta OT, because the answer might be Ogg.
This could really help me. If you know of anything, reply.
Emai: dbentley@stanford dot it's-a-university
Too bad this was done in 1993, back when NeXT cubes were plentiful. There's only a finite amount of anything in this world: including space. If we don't recycle stuff, including doing cool stuff and having funn with it, that's no good either. So sit back, and enjoy the pyrotechnics.
I'm lost...
This describes what Microsoft USED to do. Microsoft no longer restricts it to their biggest companies (or universities, which for a long time have had access, which noone here seems to realize), but instead allows anyone to look at WinCE code. You can even mess around with it, modify, recompile, as long as it's not for commercial use. This is pretty cool. You can hack with it, play around with it, etc., as long as you don't try to steal Microsoft's code.
Granted, it's not open source by a long shot, but it is a way for Microsoft code to become friendlier.
Oh, and if you check it out, they even allow you to use code in your own. So it's NOT the "Oh-my-god-if-your-seventeenth-cousand-thrice-remo ved-looks-at-this-you-can't-make-anything-more-tha n-shell-scripts-or-Linux-will-be-fucked." Indeed, they're willing to give you ideas.
Shared source isn't what we're into, granted, but it is a lot nicer than we give it credit for. If we're going to be opinionated, let's at least be right.
If you don't know this kind of stuff, then don't criticize Intel. Branch Prediction is hard stuff. If I asked you to do it with 90% accuracy, I bet you couldn't.
Oy Vey!