It most definitely isn't irrelevant (at least once you extend it to Rice's theorem) in the real world. It may not be particularly relevant to NaCl, but that's not the only thing out there.
You can argue that some other things are more relevant in program analysis (e.g. the fact that language inclusion even between nondeterministic finite automata is PSPACE-complete, or that language inclusion between context-free languages is undecidable) than that you can't talk about arbitrary programs, but it's still important.
Proving general code safe using static analysis is undecidable (that whole halting problem/Rice's theorem thing).
However, that doesn't mean you can't do something like what the JVM does: impose extra requirements that go beyond safety. That is, reject huge swaths of safe programs because they are too hard to prove safe, and just make it so that your SDK doesn't generate code like that.
A more useful system, IMHO, would be one that automatically logged off every PC in a room after a motion detector noted a period of inactivity. We do have issues where people leave for the day, go into another area or just close the door and leave systems up. That's a much bigger attack surface than leaving a PC logged in with 8 other employees wandering around.
And that depends on your domain. In many places, e.g. a software development house, sure. However, in something like a doctor's office, where even the other people in an office shouldn't have access to all the systems, this is much less true.
Right, because everyone who knows the dangers is perfect and is never distracted. Way better to force the user to conform to the computer than make the computer conform to the user.
I agree. I read the graphic novel a couple months before the move came out and thought the squid ending was dumb, at a couple different levels. I thought the movie's version made it way less contrived and more believable.
To be specific, Watson got correct answers at the end of each question. He didn't get incorrect answers (that he could potentially have learned from) as demonstrated by the 1920s thing.
Frankly, I think the whole winning at Jeopardy thing is more of a demonstration of how relatively simple a task Jeopardy actually is.... There's a reason why answering trivia questions is a game show, and not a profession.
To be fair to the Watson folks, Jeopardy is much more than just trivia from a "write a computer program" point of view. Many of the questions are worded so as to be deliberately obtuse, involve puns, etc.; the thing that makes Jeopardy questions different from other trivia games (besides the question/answer inversion) pretty much lines up with many of the reasons that NLP is as difficult as it is.
This is a very different situation than, say, Deep Blue. It's pretty well-known how you can brute force complete knowledge games like chess. The impressiveness of Deep Blue came from the fact that they were able to do it extremely quickly. (They also probably came up with some nice pruning techniques and such.)
However, what isn't clear is how to brute force language. I'm not an AI researcher, but I don't even know what that means. You could give me the computer from the USS Enterprise but with the software deleted, show me how to take advantage of it's computing power, and I'd have almost no idea how to write a program that behaves like Watson.
You say a 3-year-old has better language skills than Watson, and that's true. But how many computer systems do? Any? It's hard to say because Watson is so niche, but it's definitely up there.
So don't minimize what the IBM crew did; coming up with that is most decidedly not a "trivial task."
Ah, no worries. I see what you meant... I read the "and 100% of the security" as a bit of sarcasm, and referring to "100% of the sarcasm of the old XP way (not having the dialog box)."
If my credit card company wants to talk to me and ask me security questions, they can ask the merchant to put me on the phone.
It's not likely for the CC's company's protection, it's for the merchant's. It's the merchant that bears most of the risk for fraudulent transactions, and it's their decision to add an extra validation step.
Instead Microsoft buried the controls and made it next to impossible to turn off for a particular disk... I think you could disable it by holding shift, or alt, or control, or something. Nobody can remember that and there's no indication that it's working.
Sure there is. You get a dialog that pops up about "filter keys".
I'm not sure exactly what that has to do with autorun though.
Disclaimer: the content of this post is sarcastic.
Because autorun doesn't happen by default in Vista and Win 7. That "task window" you mention appears for media that would have autorun in previous editions of Windows, and basically removes the security problems.
And if you found it naggy then turn it off. Personally, I really like it. And I think the moving of the autorun functionality to the task window is a pretty good compromise between the convenience aspect of autorun and the security enhancement of not autorunning.
About the only thing that immediately comes to mind in terms of what I would improve is to add the ability to say "always take this action for this particular volume". Then I could tell it to never do anything when I plug in my camera for instance, but to open that dialog when I put in my USB stick. I don't know how to do that.
Um, that's kind of my point. What one person considers an "everyday experience" is far from that for many others.
It's a self-reinforcing loop here: classes don't feel the need to teach manuals, so people don't learn manuals, so people don't buy manuals (and rental car companies don't stock them), so classes don't feel the need to teach them.
And so even people who are kind of interested in learning (like me) don't learn, because... where are we going to? Am I really going to go out and pay for a course so that I can learn a skill so that when I buy a new car I can put a lot of extra effort into getting a car that will let me use it so that I can work more while driving? There's a bit of hyperbole in that sentence, but I think a lot of truth too.
I'll agree that there are a number of ways that driver's ed and licensing programs here could be improved, and that places in Europe do both rather better -- but I think "they teach manuals" is rather far down on that list.
All of the sites you mentioned, with the exception of Google when live searching is enabled, are traditional CGI-based web sites in which your browser loads a new page to fetch new content.
That's a little too simplistic a view in the case of YouTube: there are portions of that site that use AJAX to update the page on the fly. (Try loading new comments, or going to the page for a user and loading videos from their 'uploads' list.) And in the case of/., I think with the previous redesign it was also possible to load comments via AJAX. (I don't know for sure, as I had it turned off.)
Hey, I'd love to be able to drive a manual, but I'll have my current car for some time more. When I do finally get out of grad school and get around to getting a new car, I might get some old beat-up stick shift for a while, learn to drive with it, then actually buy the new car.
(Just to clarify: in case it isn't obvious, I'm talking about the USA when it comes to automatics vs sticks. I realize that, for whatever reason, sticks are a lot more popular in most other places.)
The problem is that some of those things aren't really "basic everyday activities." Most dramatically, it certainly isn't a basic everyday activity for most people to drive a stick shift -- because most people don't have stick shifts. When are you going to drive one? There have only been a couple times ever where I've even really been in a situation in which I might have driven a stick if I knew how. (I'm not speaking from the authority of age here, but I have been driving for over a decade.)
From what I picked up in an earlier story, what was going on was Tunsia was injecting Javascript into the initial page. Presuming that's correct, sending the password alone wouldn't be enough; the initial page would need to be HTTPS too.
Bah, don't even bother with trying to down sample it's a waste of effort and kind of contradicts the meaning of the RAW format.
True, it would technically not be RAW; as the other reply said, I guess the thing to do would be to use 16-bit TIFF or something.
The only reason is to try and save space, in which case it's better to use a drive compression solution, or just manually.zip or.rar them up and keep a copy out when you're working.
Have you ever tried to zip a RAW file? I have... it more-or-less doesn't work. At least from my experience with my CR2s, you'd be reasonably lucky to drop the size 5%.
I disagree with the prioritizing of hard drive space, but if you do, the only ways to really save space are downsample or use a lossy format. If you've going to do one of the two, I think downsampling is the way to go; even if YOU have use for the full resolution, I agree with the guy I originally replied to who basically said that most people don't.
You generally make (what are popularly known as) HDR images and panoramas from multiple shots. If you've got an HDR panorama, as your parent suggests, there could easily be 30 pictures that went into it, and counting it as 30 pictures wouldn't be unreasonable.
That said... I would say I do a fair bit of photography and I have only a little over 10,000 shots from 5 years, and lots of those are crap (and marked as such).
RAW does not give you so much that it is worth storing stuff in it, unless you're a professional photographer and do a lot of image correction yourself, but many amateurs also like to use it as it sounds technical, and in theory gives you more control, even though it balloons the size of an image collection and does not in fact offer a huge amount more control.
I disagree, and basically unquestionably for actual shooting. Between the slight resolution boost you get (really, JPG compression artifact elimination) which allows for tighter crops, the extra dynamic range you get which allows for both correction of underexposure and adjustment if the scene is actually contrasty, and the ability to do lossless white balance correction, RAW can be the difference between a great shot and an unremarkable one. (I've had this happen.)
That said... you don't necessarily have to keep around the RAWs. Once you develop them and are happy, a reasonably low-res (though 1024? that's crazy) JPG wouldn't be an awful idea.
I'm quite happy with the fact that I'm archiving RAWs -- storage is dirt cheap -- but it isn't a necessity. But I recommend shooting in RAW for anyone who cares about their picture quality.
Count me in the group who loves Flickr. I don't use it for backup really, just pictures I want to share around. (A dozen or two a month or so on average I'd say.) I whine and complain about 98% of the software I use, but I'm actually quite happy with the software that's in my photo workflow. (That's Adobe Lightroom locally and Flickr remotely.)
On one hand, the poster is right: high resolution isn't really useful unless you want huge size or a very large crop.
But on the other hand, there are a couple reasons you might not want to do that. First, I recommend doing that resizing in post (instead of in-camera) if at all. This gives you the freedom to look at them and go "oh, actually I do want to crop that tiny section" before you lose the ability. Second, I still recommend shooting RAW if your camera can. The resolution doesn't matter, but the likely extra dynamic range and the lossless white balancing adjusting does. Then you have a decision as to whether you keep the RAWs around, or post-process to JPGs and save those. You can definitely do the latter and reclaim space, but I'm a fan of the former -- and my workflow doesn't provide any opportunity to downsample. I don't even know of any tools that will let you downsample a RAW and still get a RAW, though I suppose perhaps some DNG conversion tool may let you do it.
(Downsampling makes a lot more sense for someone like this submitter than it does for me for instance. I shoot a fair number of photos, but even at nearly 30 MB a shot (18 MP or so in RAW) the main reason I whine about the size is the flash card itself -- and if you take my advice to downsample on the computer, it doesn't get around that problem. But I have a desktop with a ton of space and a 500 GB USB drive. For me, storage is very cheap not just in monetary cost but in terms of what I need to do to use it. Someone like the submitter may have a bigger problem with the latter.)
Slashdot Mirror
It most definitely isn't irrelevant (at least once you extend it to Rice's theorem) in the real world. It may not be particularly relevant to NaCl, but that's not the only thing out there.
You can argue that some other things are more relevant in program analysis (e.g. the fact that language inclusion even between nondeterministic finite automata is PSPACE-complete, or that language inclusion between context-free languages is undecidable) than that you can't talk about arbitrary programs, but it's still important.
Proving general code safe using static analysis is undecidable (that whole halting problem/Rice's theorem thing).
However, that doesn't mean you can't do something like what the JVM does: impose extra requirements that go beyond safety. That is, reject huge swaths of safe programs because they are too hard to prove safe, and just make it so that your SDK doesn't generate code like that.
A more useful system, IMHO, would be one that automatically logged off every PC in a room after a motion detector noted a period of inactivity. We do have issues where people leave for the day, go into another area or just close the door and leave systems up. That's a much bigger attack surface than leaving a PC logged in with 8 other employees wandering around.
And that depends on your domain. In many places, e.g. a software development house, sure. However, in something like a doctor's office, where even the other people in an office shouldn't have access to all the systems, this is much less true.
Right, because everyone who knows the dangers is perfect and is never distracted. Way better to force the user to conform to the computer than make the computer conform to the user.
I agree. I read the graphic novel a couple months before the move came out and thought the squid ending was dumb, at a couple different levels. I thought the movie's version made it way less contrived and more believable.
To be specific, Watson got correct answers at the end of each question. He didn't get incorrect answers (that he could potentially have learned from) as demonstrated by the 1920s thing.
Frankly, I think the whole winning at Jeopardy thing is more of a demonstration of how relatively simple a task Jeopardy actually is. ... There's a reason why answering trivia questions is a game show, and not a profession.
To be fair to the Watson folks, Jeopardy is much more than just trivia from a "write a computer program" point of view. Many of the questions are worded so as to be deliberately obtuse, involve puns, etc.; the thing that makes Jeopardy questions different from other trivia games (besides the question/answer inversion) pretty much lines up with many of the reasons that NLP is as difficult as it is.
This is a very different situation than, say, Deep Blue. It's pretty well-known how you can brute force complete knowledge games like chess. The impressiveness of Deep Blue came from the fact that they were able to do it extremely quickly. (They also probably came up with some nice pruning techniques and such.)
However, what isn't clear is how to brute force language. I'm not an AI researcher, but I don't even know what that means. You could give me the computer from the USS Enterprise but with the software deleted, show me how to take advantage of it's computing power, and I'd have almost no idea how to write a program that behaves like Watson.
You say a 3-year-old has better language skills than Watson, and that's true. But how many computer systems do? Any? It's hard to say because Watson is so niche, but it's definitely up there.
So don't minimize what the IBM crew did; coming up with that is most decidedly not a "trivial task."
Ah, no worries. I see what you meant... I read the "and 100% of the security" as a bit of sarcasm, and referring to "100% of the sarcasm of the old XP way (not having the dialog box)."
Why, do your pressed CDs regularly change their contents?
If my credit card company wants to talk to me and ask me security questions, they can ask the merchant to put me on the phone.
It's not likely for the CC's company's protection, it's for the merchant's. It's the merchant that bears most of the risk for fraudulent transactions, and it's their decision to add an extra validation step.
Instead Microsoft buried the controls and made it next to impossible to turn off for a particular disk... I think you could disable it by holding shift, or alt, or control, or something. Nobody can remember that and there's no indication that it's working.
Sure there is. You get a dialog that pops up about "filter keys".
I'm not sure exactly what that has to do with autorun though.
Disclaimer: the content of this post is sarcastic.
Because autorun doesn't happen by default in Vista and Win 7. That "task window" you mention appears for media that would have autorun in previous editions of Windows, and basically removes the security problems.
And if you found it naggy then turn it off. Personally, I really like it. And I think the moving of the autorun functionality to the task window is a pretty good compromise between the convenience aspect of autorun and the security enhancement of not autorunning.
About the only thing that immediately comes to mind in terms of what I would improve is to add the ability to say "always take this action for this particular volume". Then I could tell it to never do anything when I plug in my camera for instance, but to open that dialog when I put in my USB stick. I don't know how to do that.
Um, that's kind of my point. What one person considers an "everyday experience" is far from that for many others.
It's a self-reinforcing loop here: classes don't feel the need to teach manuals, so people don't learn manuals, so people don't buy manuals (and rental car companies don't stock them), so classes don't feel the need to teach them.
And so even people who are kind of interested in learning (like me) don't learn, because... where are we going to? Am I really going to go out and pay for a course so that I can learn a skill so that when I buy a new car I can put a lot of extra effort into getting a car that will let me use it so that I can work more while driving? There's a bit of hyperbole in that sentence, but I think a lot of truth too.
I'll agree that there are a number of ways that driver's ed and licensing programs here could be improved, and that places in Europe do both rather better -- but I think "they teach manuals" is rather far down on that list.
All of the sites you mentioned, with the exception of Google when live searching is enabled, are traditional CGI-based web sites in which your browser loads a new page to fetch new content.
That's a little too simplistic a view in the case of YouTube: there are portions of that site that use AJAX to update the page on the fly. (Try loading new comments, or going to the page for a user and loading videos from their 'uploads' list.) And in the case of /., I think with the previous redesign it was also possible to load comments via AJAX. (I don't know for sure, as I had it turned off.)
Hey, I'd love to be able to drive a manual, but I'll have my current car for some time more. When I do finally get out of grad school and get around to getting a new car, I might get some old beat-up stick shift for a while, learn to drive with it, then actually buy the new car.
(Just to clarify: in case it isn't obvious, I'm talking about the USA when it comes to automatics vs sticks. I realize that, for whatever reason, sticks are a lot more popular in most other places.)
Basic everyday activities.
The problem is that some of those things aren't really "basic everyday activities." Most dramatically, it certainly isn't a basic everyday activity for most people to drive a stick shift -- because most people don't have stick shifts. When are you going to drive one? There have only been a couple times ever where I've even really been in a situation in which I might have driven a stick if I knew how. (I'm not speaking from the authority of age here, but I have been driving for over a decade.)
It's also too hard to tell the indentation level of comments, and the text box on the "edit comment" page is too narrow.
Read "with a tank of explosive vapors" if you prefer.
From what I picked up in an earlier story, what was going on was Tunsia was injecting Javascript into the initial page. Presuming that's correct, sending the password alone wouldn't be enough; the initial page would need to be HTTPS too.
Bah, don't even bother with trying to down sample it's a waste of effort and kind of contradicts the meaning of the RAW format.
True, it would technically not be RAW; as the other reply said, I guess the thing to do would be to use 16-bit TIFF or something.
The only reason is to try and save space, in which case it's better to use a drive compression solution, or just manually .zip or .rar them up and keep a copy out when you're working.
Have you ever tried to zip a RAW file? I have... it more-or-less doesn't work. At least from my experience with my CR2s, you'd be reasonably lucky to drop the size 5%.
I disagree with the prioritizing of hard drive space, but if you do, the only ways to really save space are downsample or use a lossy format. If you've going to do one of the two, I think downsampling is the way to go; even if YOU have use for the full resolution, I agree with the guy I originally replied to who basically said that most people don't.
You generally make (what are popularly known as) HDR images and panoramas from multiple shots. If you've got an HDR panorama, as your parent suggests, there could easily be 30 pictures that went into it, and counting it as 30 pictures wouldn't be unreasonable.
That said... I would say I do a fair bit of photography and I have only a little over 10,000 shots from 5 years, and lots of those are crap (and marked as such).
RAW does not give you so much that it is worth storing stuff in it, unless you're a professional photographer and do a lot of image correction yourself, but many amateurs also like to use it as it sounds technical, and in theory gives you more control, even though it balloons the size of an image collection and does not in fact offer a huge amount more control.
I disagree, and basically unquestionably for actual shooting. Between the slight resolution boost you get (really, JPG compression artifact elimination) which allows for tighter crops, the extra dynamic range you get which allows for both correction of underexposure and adjustment if the scene is actually contrasty, and the ability to do lossless white balance correction, RAW can be the difference between a great shot and an unremarkable one. (I've had this happen.)
That said... you don't necessarily have to keep around the RAWs. Once you develop them and are happy, a reasonably low-res (though 1024? that's crazy) JPG wouldn't be an awful idea.
I'm quite happy with the fact that I'm archiving RAWs -- storage is dirt cheap -- but it isn't a necessity. But I recommend shooting in RAW for anyone who cares about their picture quality.
Count me in the group who loves Flickr. I don't use it for backup really, just pictures I want to share around. (A dozen or two a month or so on average I'd say.) I whine and complain about 98% of the software I use, but I'm actually quite happy with the software that's in my photo workflow. (That's Adobe Lightroom locally and Flickr remotely.)
I sort of agree with this and sort of disagree.
On one hand, the poster is right: high resolution isn't really useful unless you want huge size or a very large crop.
But on the other hand, there are a couple reasons you might not want to do that. First, I recommend doing that resizing in post (instead of in-camera) if at all. This gives you the freedom to look at them and go "oh, actually I do want to crop that tiny section" before you lose the ability. Second, I still recommend shooting RAW if your camera can. The resolution doesn't matter, but the likely extra dynamic range and the lossless white balancing adjusting does. Then you have a decision as to whether you keep the RAWs around, or post-process to JPGs and save those. You can definitely do the latter and reclaim space, but I'm a fan of the former -- and my workflow doesn't provide any opportunity to downsample. I don't even know of any tools that will let you downsample a RAW and still get a RAW, though I suppose perhaps some DNG conversion tool may let you do it.
(Downsampling makes a lot more sense for someone like this submitter than it does for me for instance. I shoot a fair number of photos, but even at nearly 30 MB a shot (18 MP or so in RAW) the main reason I whine about the size is the flash card itself -- and if you take my advice to downsample on the computer, it doesn't get around that problem. But I have a desktop with a ton of space and a 500 GB USB drive. For me, storage is very cheap not just in monetary cost but in terms of what I need to do to use it. Someone like the submitter may have a bigger problem with the latter.)