Hooray! I have all kinds of big plans for such a device. I'm assuming that this is the kind that I can turn around: use it as a client with a host PC, and yet as a host with a client peripheral?
See the old Saturday Night Live sketch about the "All-Drug Olympics" for a take on why a seperate "enhanced" Olympics might be a bad idea. The punch line involves a weightlifter ripping his arms off trying to do a snatch lift. Commentator: "Oh, that's really going to hurt when the drugs wear off."
I would pay a substantial sum for a car "TiVo" radio with the following properties:
TiVo-like ability to play one radio stream while recording another.
TiVo-like ability to wake up even with the car turned off and record my favorite programs. (Car radios don't consume much power when they aren't playing audio, and the unit could detect a low-battery condition and refuse.)
Mechanism for getting recordings out of the car: some sort of flash card and/or USB port would do nicely.
Builtin CD player with MP3 capabilities.
I like listening to non-music radio (mostly NPR). Recently, I constantly find myself reaching to back over something my radio just said and hear it again---only to realize my radio's not a TiVo. When I tell my friends about something cool I just heard, it would be nice to be able to just hand them the MP3 as well. It would be nice to have a menu of automatically recorded stuff for when nothing's on (often) or when I'm out of radio range. It would be nice to record a program while listening to a CD or a previous program. Etc.
Why doesn't this device seem to exist? Have I missed a product?
Who gets sued when damages come as a result of security flaws in open source software (and yes, such flaws do exist)?
The author of the software, of course. For most open source software, this would effectively prevent a lawsuit, since most open source developers don't have deep enough pockets to make it worth the trouble. For the stuff written by e.g. IBM, though, you bet they'd have to be careful. (Note that the proposal is really to allow consumers to sue over gross negligence, not just the existence of a security flaw.)
In reality of course, most software comes with disclaimers saying the company is not liable for such damages, so this whole issue is moot.
In reality, the legal status of such disclaimers tends to be highly questionable. I can put a disclaimer on my Happy Fun Ball (TM) saying I'm not responsible if it explodes and kills your kid: trust me that this isn't lawsuit-proofing. There are many kinds of liability that cannot be disclaimed. The principal reason for such disclaimers is to discourage licensees from suing, not to protect the licensor from judgement.
The potential advantage of this CFC thing over the gumstix (which is cool) is that the stupid client USB port on the gumstix means that it's going to be a struggle to attach USB peripherals. With the CF bus, I should be able to attach CF peripherals to the CFC easily. Presumably the next gumstix will based on a part with the new USB 2.0 controllers which can be switched between host and client modes. This would be good.
OTOH, the gumstix ARM should be substantially faster than the CFC Moto ColdFire part. Neither has an FPU, so CPU speed will matter in some applications. Not sure why the CFC didn't go Xscale like everyone else these days. Jamming 8 400MHz Xscale parts into a CF bus starts to look like a little low-power NUMA supercomputer node:-).
His Myth 1: Linux is just as easy to install as Windows. -- My mom can install Windows (without any help from me -- I just tell her "If you don't know what to do, just click Next" -- and when she's done, she has a fully functional OS. The linux installation experience is dramatically more complicated, and it's unlikely the end-product will work right if it was done by a novice (he pretty much admits this).
I installed Windows XP for the first time recently. It didn't talk
to about 1/2 my HW until extensive banging around was performed. It still randomly locks up fairly regularly, even though it's running behind a firewall with virus and spyware checkers installed and up-to-date and all Windows patches.
Meanwhile, ever booted a Knoppix CD? I suppose it gets easier than that, but I don't know how. It generally autoconfigures all your hardware, and brings you up running Linux with zero questions. I do my Linux HD installs these days with its built-in HD installer---retains all the configuration it has found, so things just work.
Try this with your mom. For real. Go down to the store and buy a Fry's box that comes loaded with ThizLinux. Hand your mom that box and a Windows XP CD, and stay out of her way for a while. See what happens. I have a theory...
I don't get it. If everybody has a direct neural Feed, why talk at all, except through it? Seems like it would revolutionize society in ways too hard to write about...oh, now I get it.
See Alfred Bester's famous novel The Demolished Man for an interesting take on a future in which just a subset of folks are telepaths. Highly recommended.
Puzzle games can work for intermittent intense play if the puzzles are kept interesting and doable in a sitting. The famous Mac title The Fool's Errand comes immediately to mind. Sliding blocks puzzles get boring pretty fast.
One of my colleagues likes to tease our students by referring to this volunteer grid stuff as "H-bomb@home". "Sure, your SW says it's doing gravity-wave calculations. I claim that USDoD is using it to do H-bomb (or bioweapon, or whatever) design simulations for free on your computer. Go ahead, prove me wrong."
I think Gelernter's Mirror Worlds enunciates Kay's vision nicely. Basically, if you can simulate something accurately, you can run "virtual experiments" and monitor system differences from simulation, which can be convenient indeed.
You left out the punch line. Larson walks away with the most money ever won in a game show, and promptly loses it all (and then some) in a bad real estate investment scheme. A year later, he's begging the producers to let him back on the show, believing that he can beat their new super-tough wheel. Producers: "Uh, no."
Even going 10m on a single pump stage is quite difficult. The normal procedure used for electric jet pumps in wells (which routinely pump 75m), is to "stage" the pumps: provide a series of intermediate pump mechanisms every few meters along the way that repressurize the water for the next leg.
I would think that a staged Archimedes' screw type design would work well in this application.
If I tell you I'm running Linux with a gui, and I want to reboot, can you tell me, without looking over my shoulder, where, spacially, on my interface, I need to go? Even if I tell you "I'm using Gnome" or "I'm using KDE", can you then tell me where the 'Log-out' or 'Reboot' button is? No- because it's very customizable, any button could be anywhere.
<CTRL-ALT-F1><CTRL-ALT-DEL> will almost always get it. But I agree that that's somewhat beside the point.:-)
Ammonia is being used as an alternative to hydrogen because it is safer? I don't understand: I had thought that gaseous ammonia is extremely poisonous. For an unmanned vehicle, surely the risk of poisoning someone trumps the risk of burning up the vehicle?
According to Nickle, 100/2.54 is the repeating decimal 39.{370078740157480314960629921259842519685039}. The approximation 39.3700787 ought to be close enough for most work:-). Heck, I'd think 39.37 would pretty much do it. As a point of comparison, the old UNIX "units" program gives 39.370079
Amen. I learned UNIX back in 1984 under BSD 4.1. (Actually, did some work earlier under 2.9.) The system calls are still pretty similar (although I miss MPXIO:-). Device drivers look almost identical in their basic structure. The password file is the same except for the obvious improvement of supporting longer hashes. libc has some more functions in it, but all the old ones work. Etc, etc.
Biggest change is we have X now, but even that is pretty darn similar to X of 1990, except that we have (a) much better HW to run it on (no more 8 bit pseudocolor---yay!) and (b) much better toolkits to run on top of it (no more Xt---yay!).
Probably 80% of the C code I write today would run on the first boxes I worked on with only trivial modifications.
On the administration front, similar. Probably the biggest change has been the introduction of NIS and NFS, but to anyone who knew the old stuff, this isn't a big learning curve.
One of the cool things about UNIX is that improvements tend to be incremental in their learning curve, and unimpactful on existing knowledge. This is evidence that the design was right in the first place...
Heck, probably couldn't even take a digital picture of the magic t-shirt. Nobody clue in the primitive societies, or I'll have to go back to film cameras to steal their souls...
I'm sure mandatory banknote recognition in Europe will be as effective as mandatory weak crypto was in the US. As long as there's someplace in the internet-connected world that doesn't ban some kind of software, it will be available everywhere---this is one of the strengths of libre software.
The thing you may have missed is that a major reason current OSS and vendor supplied OpenGL drivers cause so much grief is that they have to sit inside DRI. This means that they are intimately entangled with the details of how the X server works---they need to know where they're supposed to be painting, how to get out of the way of X cursors, etc.
In the new world, the OpenGL driver can simply take over the screen. This is what they desperately want to do. This means that driver versions won't need to track X server versions, that graphic chip init can move into the driver where it belongs, that differences in HW capabilities are obscured to some degree: in short, it massively simplifies the driver architecture.
(BTW, not sure when you tried Mesa Indirect last. It's certainly fast enough to do fairly elaborate 2D rendering in SW on any reasonable modern processor I've tried.)
Don't take my word for it. Check out the refereed papers in the upcoming Usenix Annual Technical Conference Freenix Track on this topic. They should help you see where things are headed, and why it is a good idea.
Slashdot Mirror
Hooray! I have all kinds of big plans for such a device. I'm assuming that this is the kind that I can turn around: use it as a client with a host PC, and yet as a host with a client peripheral?
Looking forward to it---thanks for the note!
See the old Saturday Night Live sketch about the "All-Drug Olympics" for a take on why a seperate "enhanced" Olympics might be a bad idea. The punch line involves a weightlifter ripping his arms off trying to do a snatch lift. Commentator: "Oh, that's really going to hurt when the drugs wear off."
I would pay a substantial sum for a car "TiVo" radio with the following properties:
I like listening to non-music radio (mostly NPR). Recently, I constantly find myself reaching to back over something my radio just said and hear it again---only to realize my radio's not a TiVo. When I tell my friends about something cool I just heard, it would be nice to be able to just hand them the MP3 as well. It would be nice to have a menu of automatically recorded stuff for when nothing's on (often) or when I'm out of radio range. It would be nice to record a program while listening to a CD or a previous program. Etc.
Why doesn't this device seem to exist? Have I missed a product?
Here's the sizes for my antialiased KDE desktop with an xterm and firefox open on it.
Original commenter looks right to me. Interesting.(IANAL.)
Who gets sued when damages come as a result of security flaws in open source software (and yes, such flaws do exist)?
The author of the software, of course. For most open source software, this would effectively prevent a lawsuit, since most open source developers don't have deep enough pockets to make it worth the trouble. For the stuff written by e.g. IBM, though, you bet they'd have to be careful. (Note that the proposal is really to allow consumers to sue over gross negligence, not just the existence of a security flaw.)
In reality of course, most software comes with disclaimers saying the company is not liable for such damages, so this whole issue is moot.
In reality, the legal status of such disclaimers tends to be highly questionable. I can put a disclaimer on my Happy Fun Ball (TM) saying I'm not responsible if it explodes and kills your kid: trust me that this isn't lawsuit-proofing. There are many kinds of liability that cannot be disclaimed. The principal reason for such disclaimers is to discourage licensees from suing, not to protect the licensor from judgement.
The potential advantage of this CFC thing over the gumstix (which is cool) is that the stupid client USB port on the gumstix means that it's going to be a struggle to attach USB peripherals. With the CF bus, I should be able to attach CF peripherals to the CFC easily. Presumably the next gumstix will based on a part with the new USB 2.0 controllers which can be switched between host and client modes. This would be good.
OTOH, the gumstix ARM should be substantially faster than the CFC Moto ColdFire part. Neither has an FPU, so CPU speed will matter in some applications. Not sure why the CFC didn't go Xscale like everyone else these days. Jamming 8 400MHz Xscale parts into a CF bus starts to look like a little low-power NUMA supercomputer node :-).
His Myth 1: Linux is just as easy to install as Windows. -- My mom can install Windows (without any help from me -- I just tell her "If you don't know what to do, just click Next" -- and when she's done, she has a fully functional OS. The linux installation experience is dramatically more complicated, and it's unlikely the end-product will work right if it was done by a novice (he pretty much admits this).
I installed Windows XP for the first time recently. It didn't talk to about 1/2 my HW until extensive banging around was performed. It still randomly locks up fairly regularly, even though it's running behind a firewall with virus and spyware checkers installed and up-to-date and all Windows patches.
Meanwhile, ever booted a Knoppix CD? I suppose it gets easier than that, but I don't know how. It generally autoconfigures all your hardware, and brings you up running Linux with zero questions. I do my Linux HD installs these days with its built-in HD installer---retains all the configuration it has found, so things just work.
Try this with your mom. For real. Go down to the store and buy a Fry's box that comes loaded with ThizLinux. Hand your mom that box and a Windows XP CD, and stay out of her way for a while. See what happens. I have a theory...
I don't get it. If everybody has a direct neural Feed, why talk at all, except through it? Seems like it would revolutionize society in ways too hard to write about...oh, now I get it.
See Alfred Bester's famous novel The Demolished Man for an interesting take on a future in which just a subset of folks are telepaths. Highly recommended.
A computer is a machine for getting wrong answers quickly.
Perl can help with that.
Puzzle games can work for intermittent intense play if the puzzles are kept interesting and doable in a sitting. The famous Mac title The Fool's Errand comes immediately to mind. Sliding blocks puzzles get boring pretty fast.
Lets look at this from a physics standpoint. It is a glove with wires coming out. There are ...no rockets to fire...
Hey, credit where credit is due! That's my idea!
BTW, note the similarity of the actual scuplting system to this proposal.
That's cool. Thanks for the tip!
One of my colleagues likes to tease our students by referring to this volunteer grid stuff as "H-bomb@home". "Sure, your SW says it's doing gravity-wave calculations. I claim that USDoD is using it to do H-bomb (or bioweapon, or whatever) design simulations for free on your computer. Go ahead, prove me wrong."
IMHO it's an interesting point.
I think Gelernter's Mirror Worlds enunciates Kay's vision nicely. Basically, if you can simulate something accurately, you can run "virtual experiments" and monitor system differences from simulation, which can be convenient indeed.
[Michael Larson story ...]
You left out the punch line. Larson walks away with the most money ever won in a game show, and promptly loses it all (and then some) in a bad real estate investment scheme. A year later, he's begging the producers to let him back on the show, believing that he can beat their new super-tough wheel. Producers: "Uh, no."
Hope KenJen does better with his winnings.
Even going 10m on a single pump stage is quite difficult. The normal procedure used for electric jet pumps in wells (which routinely pump 75m), is to "stage" the pumps: provide a series of intermediate pump mechanisms every few meters along the way that repressurize the water for the next leg.
I would think that a staged Archimedes' screw type design would work well in this application.
If I tell you I'm running Linux with a gui, and I want to reboot, can you tell me, without looking over my shoulder, where, spacially, on my interface, I need to go? Even if I tell you "I'm using Gnome" or "I'm using KDE", can you then tell me where the 'Log-out' or 'Reboot' button is? No- because it's very customizable, any button could be anywhere.
<CTRL-ALT-F1><CTRL-ALT-DEL> will almost always get it. But I agree that that's somewhat beside the point. :-)
Ammonia is being used as an alternative to hydrogen because it is safer? I don't understand: I had thought that gaseous ammonia is extremely poisonous. For an unmanned vehicle, surely the risk of poisoning someone trumps the risk of burning up the vehicle?
In fact remember to wonder why the next person you see says something, in general.
Hell yeah. In fact, I'll bet you are paid behind the scenes by, uh, somebody to support, uh...rational skepticism.
Nevermind.
According to Nickle, 100/2.54 is the repeating decimal 39.{370078740157480314960629921259842519685039}. The approximation 39.3700787 ought to be close enough for most work :-). Heck, I'd think 39.37 would pretty much do it. As a point of comparison, the old UNIX "units" program gives 39.370079
Amen. I learned UNIX back in 1984 under BSD 4.1. (Actually, did some work earlier under 2.9.) The system calls are still pretty similar (although I miss MPXIO :-). Device drivers look almost identical in their basic structure. The password file is the same except for the obvious improvement of supporting longer hashes. libc has some more functions in it, but all the old ones work. Etc, etc.
Biggest change is we have X now, but even that is pretty darn similar to X of 1990, except that we have (a) much better HW to run it on (no more 8 bit pseudocolor---yay!) and (b) much better toolkits to run on top of it (no more Xt---yay!).
Probably 80% of the C code I write today would run on the first boxes I worked on with only trivial modifications.
On the administration front, similar. Probably the biggest change has been the introduction of NIS and NFS, but to anyone who knew the old stuff, this isn't a big learning curve.
One of the cool things about UNIX is that improvements tend to be incremental in their learning curve, and unimpactful on existing knowledge. This is evidence that the design was right in the first place...
Making a decent color printer could be very hard. Replacing its firmware and/or its compute hardware should be pretty easy.
Heck, probably couldn't even take a digital picture of the magic t-shirt. Nobody clue in the primitive societies, or I'll have to go back to film cameras to steal their souls...
I'm sure mandatory banknote recognition in Europe will be as effective as mandatory weak crypto was in the US. As long as there's someplace in the internet-connected world that doesn't ban some kind of software, it will be available everywhere---this is one of the strengths of libre software.
The thing you may have missed is that a major reason current OSS and vendor supplied OpenGL drivers cause so much grief is that they have to sit inside DRI. This means that they are intimately entangled with the details of how the X server works---they need to know where they're supposed to be painting, how to get out of the way of X cursors, etc.
In the new world, the OpenGL driver can simply take over the screen. This is what they desperately want to do. This means that driver versions won't need to track X server versions, that graphic chip init can move into the driver where it belongs, that differences in HW capabilities are obscured to some degree: in short, it massively simplifies the driver architecture.
(BTW, not sure when you tried Mesa Indirect last. It's certainly fast enough to do fairly elaborate 2D rendering in SW on any reasonable modern processor I've tried.)
Don't take my word for it. Check out the refereed papers in the upcoming Usenix Annual Technical Conference Freenix Track on this topic. They should help you see where things are headed, and why it is a good idea.