If we could all use LinuxBIOS, which can boot Linux or *BSD or Windows directly, we wouldn't need any separate bootloader software.
Using separate bootloeader software is what made running Linux or *BSD on x86 possible in the first place. Hardcoding OS loading code into BIOS, beyond the absolute minimum neccessary, is a really bad idea, since it prevents you from running any OS the BIOS loader doesn't support.
Not if the BIOS loader is designed to be extremely flexible and customizable, or if the BIOS itself is open source and can be easily updated to support new operating system:-)
I wish someone would acquire the GRUB, and close it. That piece of crap has caused me more pain than any other open source software. If lilo broke, well just grab a boot disk and rerun it. When (not if) grub breaks, god help you trying to figure out what to do. Are you kidding me? On most Linux systems, you can just run update-grub to reinstall the first-stage bootloader (or grub-install on some). To adjust its settings, you need only edit the file/boot/grub/menu.lst... you don't even have to re-run GRUB after editing that file.
Unlike Lilo, GRUB offers a full-featured command line so that you can edit your boot settings if it doesn't quite work right. No need for a rescue disk almost ever. GRUB reads it configuration file at boot time, unlike LILO which hides it somewhere after the boot sector of the hard disk.
GRUB is very powerful, it can boot off USB drives in a sane way, it can work around all kinds of BIOS bugs, etc. And in my opinion it's easier-to-use than LILO as well! It sounds like you only have trouble with GRUB because you never bothered to learn the one or two commands needed to reinstall it:-) Since I switched from LILO to GRUB, the *only* times I've ever needed to reinstall GRUB are when I install Windows on top of Linux, and it hoses my boot sector. And I lay the blame for that squarely on Microsoft...
I admit, grub is nice when it automagically works. The problem is when it doesn't. GRUB failures are the only reason I reinstall operating systems anymore. Well, sure... *any* bootloader is great when it "just works." It's when it doesn't quite boot right that you start to care about it. And that's why GRUB's command line and other features are so great.
For many of us, the bootloader is just a solution to the fact that the PC BIOS is horribly retarded, and no self-respecting operating system kernel includes the kind of awful code needed to interface with the BIOS. If we could all use LinuxBIOS, which can boot Linux or *BSD or Windows directly, we wouldn't need any separate bootloader software. Someday...
I love how when you go to Microsoft's Open Source Site, the first thing that greets you is a giant FLASH ANIMATION. Well, that's not exactly open-source friendly, now is it!
Since this is the norm when discussing Moore's "law", I'd rather see one of those mythical non-retarded arguments regarding it. There are none. Well, you have a point. There are no really good arguments for why Moore's law *should* hold. On the other hand, it has been going pretty steadily for 40+ years, so I'll remain skeptical of anyone who says that Moore's law will stop.
Exactly. What a frickin' retarded argument... since when has the low end of computing actually dragged down the high end?
We may have unprecedented demand for low-power 200 MHz ARM processors these days, but we also have unprecedented demand for quad-core 2 GHz beasts in 1U rack-mount servers, so we can stuff more and more of them into vast underground data centers. Moore's law applies equally to the low end and the high end. Today we can put a powerful computer in a $500 iPhone, maybe tomorrow we can put it in a $50 iWatch. There's absolutely no economic reason for Moore's Law not to continue unabated.
This is an issue I have recent and intimate knowledge of. XP will *NOT* install on a standard CF card. Even with a CF/IDE converter, Windows sees the CF card as a "Removable Device" and will not install to it. Windows also will only ever see one partition on a removable device. It's also broken when trying to format an existing partition during install, and it corrupts itself when trying to expand it's C: partition when installing from a sysprep'ed disk image. The only way I was able to get it installed was to create a sysprep image the exact size that the finished install will be and write it directly to the flash drive. It's kind of funny to double click on "My Computer" and see the C: drive show up as a removable device with a little removable type icon. This guys blog details the issues a bit more:
The main difference are the write/read cycles the drives can take, SSDs have built in algorithms to evenly spread the writes out over the disk over time, which greatly increases the life of the disk, granted you could probably do this in software, but its another thing to deal with. Standard CF/SD memory can only take a few hundred thousand cycles, which as a system disk is gone in a very short time.
So a CF/SD SSD would work and be cheaper, but would probably not last very long, and be slower. As I see it, the ONLY high-performance way to do write levelling is in software: that is, flash devices should use different filesystem structures from hard disks. In hard disks, fragmentation is very bad, so data should be kept together on the disk. While in flash devices, fragmentation is not an issue, but wear and write granularity are important, so data should be kept in a sort of cyclic log structure.
Under Linux, I think the two-CF-cards-with-an-IDE-adapter drive would be really awesome. Linux has very high-performance software RAID support, and you can use a lot of excellent high-performance flash filesystems. Windows doesn't have good software RAID support or flash filesystem support though, so I guess Windoze users would prefer one of these SSDs that presents itself as a monolithic device with wear levelling in the firmware.
That's what I was thinking. The $140 CF card is rated 40X or 6 MB/s transfer rate. But most 5400RPM 2.5" hard drives can barely do that in a sustained write, in my experience.
If you put two of those $140 CF cards in a striped RAID-0 with the dual-card Addonics adapter, you'll have a 32gb solid state disk, with a speed approaching 12 MB/s rate, for about $300 ($280 for the two CF cards, and $20 for the dual adapter).
Those look really cool! I love the dual adapter, which allows you to put two compactflash cards into one notebook IDE slot: one master and one slave. Add Linux a little bit of software RAID 0, and you've got twice the bandwidth and twice the storage. Brilliant.
make sure your cf-ide adapter supports dma transfers. The CF-IDE adapter is simply a passive mechanical adapter... nothing more than a connector between the pins of the CF card and the pins of the IDE header.
However, you bring up a good point: if the CF card doesn't support DMA, it will be quite slow. The one I linked to apparently doesn't support DMA:-( Anyone know what the prices are like for 16gb CF cards that do support UDMA mode 4? An 8gb CF card supporting DMA costs $110... and it is made by Transcend. It sounds like they may be the leading maker of CF cards that support DMA.
Hopefully other manufacturers will catch up quick, since DMA capabilities don't depend on the raw NAND flash chips, only on the controller chip... so the cost to manufacture a CF card supporting DMA should barely increase.
The two greatest resources I've found for finding Linux wireless card drivers are:
http://linux-wless.passys.nl/ That site is awesome. Thanks for the link! I've been hoping for a searchable database of linux-friendly wireless cards for a while (even thought about making my own)!
On the other hand, a 16gb CompactFlash card is only $140. And the CompactFlash interface is electrically identical to IDE/PATA, so you can use a $5 mechanical adapter to connect a CompactFlash card to your notebook's hard drive bay.
What am I missing here???
I can make my own 16gb solid-state IDE disk for only $150 (and 32gb CF cards are coming out in a few months).
Actually, it doesn't work like that. What actually happens is that the code which is maintained poorly gets dropped.
That's a pretty unfortunate situation if the unmaintained code is still actually used by someone.
(...)
That said, yeah, if someone notices that filesystem FooFS has been completely broken for ages and nobody has even noticed, then that's a pretty good argument for dropping it. But even then it's not just because it's unmaintained, it's because at that point you're pretty sure nobody really gives a crap about it.
The Linux kernel *almost never* drops support for any devices/filesystems unless (a) it's INCREDIBLY obsolete and NO ONE is using it, or (b) it's been superseded by something clearly better and there's a straightforward upgrade path.
For example, if you read the kernel changelog summaries on LWN.net, you'll see that support for IBM PC/XT hard disks was only dropped in the last couple years... although they have been obsolete since the late 80s and perhaps literally no one has used them for 5-10 years. And support for the original "ext" filesystem was removed a few months ago, despite the fact that it's been completely superseded by ext2--which was introduced in 1993.
As Greg Kroah-Hartman has pointed out, the kernel developers are perfectly willing to maintain a driver for which only a single piece of hardware exists in the whole world!
Every time I ever wanted to upgrade *anything* bar the HD in my computer in the past *fifteen* years, I has to buy a new mobo. Might as well get new everything else. That's of course only if you upgrade every 2-3 years or so. More often than that, you may benefit. You're also blowing money for bragging rights. Hence ricer. Well... I don't agree. For example, with my current desktop, I'm perfectly happy with the HDD, graphics card, monitor, keyboard, mouse, optical drive (it's only about 1.5 years old), case, and power supply. But I could really use more RAM and a faster processor, so I'll probably go with a new motherboard, processor, and RAM... which will set me back about $300 compared to $700 for the whole new computer.
I'm definitely not a ricer. I've never bought a case with lots of LEDs, I don't go to LAN parties, I don't play 3D games... I mostly use my computer for software development, office-related stuff, web browsing, and as a server.
Off the top of my head, here are several categories of power users who really need upgradability and expandability to get their jobs done:
Audio pros: They need to be able to add high-end sound cards which take up a whole drive bay for I/O jacks and controls.
Device driver writers: They need to test different combinations of hardware.
Video pros: They may need high-end encoder/decoder cards, and room for many internal HDDs.
Scientists (me at work!): We need to be able to interface our computers to data acquisition boards and custom interfaces for specific instruments. Many of these use legacy interfaces like ISA, oversized PCI cards, serial ports, etc. which aren't available for laptops... and the instruments themselves cost 10-100X what the computer does.
Just because desktop sales have leveled off doesn't mean they're going away. Sheesh! Okay, so everyone who needs and wants a desktop has one today. That's why sales are leveling off! It doesn't mean we no longer want them or need them!!! (By comparison, I doubt toothbrush sales have experienced phenomenal growth in the last few years... but that doesn't mean people don't need or won't buy toothbrushes anymore.)
There are many cool and exciting new uses for laptops/PDAs/tablets, but desktops have many uses as well. For example, most computer users have a desk at home or work where they get a lot of work done: there's no need to have that computer be mobile, and desktops are CHEAPER and MORE UPGRADABLE and MORE RELIABLE.
Upgradibility in particular is a huge issue for power users and hardware enthusiasts:
Upgrade speed: With the nicely designed OEM case of my Acer minitower, I can have the case open in less than 15 seconds. I can replace an expansion card in about 30 more seconds. I can replace a RAM stick in about 30 seconds. I can add a new SATA hard drive in a couple minutes. I can replace the processor in 3-5 minutes. The power supply in 5-10 minutes maybe. I can do a whole mobo swap in probably 10-20 minutes. And I don't need any tools. By contrast, with my laptop, it takes maybe 5 minutes to replace the hard drive, and I have to mess with a bunch of fiddly little screws. To replace the RAM or optical drive I have to remove several panels and it probably takes 10-15 minutes. Replacing a MiniPCI wifi card is a huge pain and probably takes at least half an hour. And everything else simply can't be upgraded.
Upgrade cost: Desktop computer motherboards, drives, expansion cards, power supplies, cases mostly use standard form factors and connectors. I can mix and match parts to my heart's content. Not so with the laptop! The hard disk is a standard FF, the optical drive is sorta standard, the miniPCI wireless card is sorta standard... and that's it. Good luck replacing the graphics card on your laptop, or the RAM on some models, or the processor, or the motherboard.
Upgrade options: You can upgrade everything on a desktop. You can't easily upgrade anything but the HDD and RAM on most laptops, doing anything more requires tiny screwdrivers, a lot of patience, and the knowledge that you can easily hose your motherboard and have to replace the whole thing. And if you need specialty expansion cards for things like GPS or data acquisition, your only option is an external device--with lower performance, more clutter, and often more limited selection.
As far as I know, these cheap hardware RNGs use random fluctuations of thermal noise... meaning they basically sample the voltage across a little resistor and record the random fluctuations.
This is cheap and effective, but is vulnerable to a particularly nasty attack: cooling down the temperature. Imagine you have a data center cranking out random numbers using a thermal RNG. The FBI/CIA/MI5/Mossad/KGB/Elvis could pump in liquid nitrogen overnight until it's cold enough that the range of fluctuations is reduced to the point that they're no longer random enough... for example, if the RNG is designed to spit out numbers between 0 and 1.0 at room temperature, it might most only produce numbers between 0 and 0.25 at the temperature of liquid nitrogen. This could be enough to completely subvert some cryptographic systems.
You can talk quantum electrodynamics all day long, until you're blue in the face, but the outcome is still 50/50, and it's still determined every bit as much by the surroundings as it is by the nuclei making the emission. Incorrect! Such quantum phenomena are not "determined" by the state of the nuclei even on a theoretical level. It is a common misconception that quantum mechanics gives probabilistic results only because the experimenters are unable to measure the "underlying" phenomena accurately (known as the hidden variables).
In fact, this has been essentially disproved by experimental verification of (building on the earlier Einstein-Podolsky-Rosen paradox).
Bell's theorem states that "No physical theory of local hidden variables can ever reproduce all of the predictions of quantum mechanics." In other words, no underlying mechanism could possibly produce the same results as quantum mechanics. I feel that this is the most elegant and fundamentally significant result of quantum mechanics.
So if random numbers are generated through quantum mechanical processes, their origin is in a very real sense not only unknown but truly unknowable.
(PS- I should mention that Bell's Theorem makes certain assumptions about locality, which is intimately related to general relativity. Both quantum mechanics and general relativity have been extremely well-tested, but it is known that they can't both be right in all circumstances... but to make a long story short, if you're not right next to a black hole then both quantum mechanics and general relativity should make sense.)
It's called Inotify. Everything and its mom uses it on Linux nowadays...
The PowerTop folks have been going around and finding a lot of programs that do polling rather than asynchronous notification, and fixing them to use things like inotify.
Granted, you do get some good jokes, but most of the time people who are not remotely funny spend the day playing unfunny jokes on each other. But you're not allowed to complain, because then you get accused of having no sense of humour. Not to mention the fact that it's freakin' July!
Incorrect. A plurality system pushes politicians towards each other. That may be true... and we often see periods where there's a surprising disconnect between political leaders and popular opinion. For example, something like 70% of the US wants to get out of Iraq soon according to some polls I've seen, and we just elected a Democratic legislative majority based largely on that sentiment, and yet they haven't made any specific plans to withdraw yet.
But when election time comes around, the politicians have to go running back to the people and promise to deliver on the issues that matter to them. Otherwise the other party's candidate will, and will win. Sure, it's messy and halting and pandering, but that's democracy in action:-)
In the USA, we don't end up with makeshift parliamentary coalitions held together by fringe parties that hold the majority hostage to their agendas (such as I read about in Italy, Israel, France, etc.) That's why our system rewards centrism and promotes stability.
If one party shifts right and is popular, the other party will shift right to catch up, which is what happens in the USA. Your Democrat party, in other countries, would be considered a fairly far right wing party. So what? Our political "center" is further to the right than that of, say, the Netherlands or Germany or the UK.
The American people as a whole would be considered "fairly far right wing" by the standards of many other democracies: our population is more religious, less concerned with the environment, broadly supports gun ownership rights, broadly supports Israel over the Palestinians, broadly supports capital punishment, etc.
Or had been smart enough to realize that plurality voting systems (instead of ranked or condorcet methods) would ultimately result in the creation of an entrenched two-party duopoly from which there appears to be no escape.
Everyone loves to hate that entrenched two-party duopoly, and I don't think the Founding Fathers intended it... but I think it's done a lot towards ensuring the stability and prosperity of the USA in the long view.
A plurality system inexorably pushes politicians towards the center, rather than the fringes. In countries with proportional representation, a fringe party can get into the legislature with, say, 3% of the national vote. In a plurality system, that party won't be more than a blip UNLESS it manages to gain a plurality of the vote in some regional district. So the plurality system encourages centrist politics, while at the same time respecting regional political differences.
Honestly... why would I pay $700 for the privilege of having not one but TWO large companies dictate to me how I can use a piece of hardware.
Dell called, your new Vista PC is ready. Actually my new Dell runs Ubuntu 7.10. Actually, I dual boot Vista Home Premium to run EXACTLY ONE program (Altera Quartus II 7.1). And I hate pretty much every minute of it.
Dell doesn't lock its computers down to prevent me from running another OS. Sure, I'm no fan of the ~$30-50 OEM tax for Windows, but it's a far cry from an entire system that is enslaved to Apple and AT&T.
I actually paid just under $600 for this nice dual-core Turion 64 X2 TL-60 rig, with a 6-hour battery upgrade... so I paid LESS than the iPhone. And no subscription fees for this one.
Honestly... why would I pay $700 for the privilege of having not one but TWO large companies dictate to me how I can use a piece of hardware. Till I can buy a phone that lets me do whatever I want with it, they're all crap to me.
The iPhone craze has truly reduced epic proportions... so much so that my GRANDMA emails me and says she wants to buy me one. I mean, I don't want to be ungrateful, I'm probably the only person I know who actively DOES NOT WANT an iPhone, I should take advantage of my most-favored-grandson status, but.... yeeesssh!
Slashdot Mirror
Using separate bootloeader software is what made running Linux or *BSD on x86 possible in the first place. Hardcoding OS loading code into BIOS, beyond the absolute minimum neccessary, is a really bad idea, since it prevents you from running any OS the BIOS loader doesn't support.
Not if the BIOS loader is designed to be extremely flexible and customizable, or if the BIOS itself is open source and can be easily updated to support new operating systemUnlike Lilo, GRUB offers a full-featured command line so that you can edit your boot settings if it doesn't quite work right. No need for a rescue disk almost ever. GRUB reads it configuration file at boot time, unlike LILO which hides it somewhere after the boot sector of the hard disk.
GRUB is very powerful, it can boot off USB drives in a sane way, it can work around all kinds of BIOS bugs, etc. And in my opinion it's easier-to-use than LILO as well! It sounds like you only have trouble with GRUB because you never bothered to learn the one or two commands needed to reinstall it
For many of us, the bootloader is just a solution to the fact that the PC BIOS is horribly retarded, and no self-respecting operating system kernel includes the kind of awful code needed to interface with the BIOS. If we could all use LinuxBIOS, which can boot Linux or *BSD or Windows directly, we wouldn't need any separate bootloader software. Someday...
I love how when you go to Microsoft's Open Source Site, the first thing that greets you is a giant FLASH ANIMATION. Well, that's not exactly open-source friendly, now is it!
Since this is the norm when discussing Moore's "law", I'd rather see one of those mythical non-retarded arguments regarding it. There are none. Well, you have a point. There are no really good arguments for why Moore's law *should* hold. On the other hand, it has been going pretty steadily for 40+ years, so I'll remain skeptical of anyone who says that Moore's law will stop.
Exactly. What a frickin' retarded argument... since when has the low end of computing actually dragged down the high end?
We may have unprecedented demand for low-power 200 MHz ARM processors these days, but we also have unprecedented demand for quad-core 2 GHz beasts in 1U rack-mount servers, so we can stuff more and more of them into vast underground data centers. Moore's law applies equally to the low end and the high end. Today we can put a powerful computer in a $500 iPhone, maybe tomorrow we can put it in a $50 iWatch. There's absolutely no economic reason for Moore's Law not to continue unabated.
XP will *NOT* install on a standard CF card. Even with a CF/IDE converter, Windows sees the CF card as a "Removable Device" and will not install to it. Windows also will only ever see one partition on a removable device. It's also broken when trying to format an existing partition during install, and it corrupts itself when trying to expand it's C: partition when installing from a sysprep'ed disk image. The only way I was able to get it installed was to create a sysprep image the exact size that the finished install will be and write it directly to the flash drive. It's kind of funny to double click on "My Computer" and see the C: drive show up as a removable device with a little removable type icon. This guys blog details the issues a bit more:
http://thebs413.blogspot.com/2005/12/windows-xp-e
So a CF/SD SSD would work and be cheaper, but would probably not last very long, and be slower. As I see it, the ONLY high-performance way to do write levelling is in software: that is, flash devices should use different filesystem structures from hard disks. In hard disks, fragmentation is very bad, so data should be kept together on the disk. While in flash devices, fragmentation is not an issue, but wear and write granularity are important, so data should be kept in a sort of cyclic log structure.
The author of LogFS, a log-structured flash filesystem, has written a very convincing paper on this point: http://lazybastard.org/~joern/logfs1.pdf
Under Linux, I think the two-CF-cards-with-an-IDE-adapter drive would be really awesome. Linux has very high-performance software RAID support, and you can use a lot of excellent high-performance flash filesystems. Windows doesn't have good software RAID support or flash filesystem support though, so I guess Windoze users would prefer one of these SSDs that presents itself as a monolithic device with wear levelling in the firmware.
That's what I was thinking. The $140 CF card is rated 40X or 6 MB/s transfer rate. But most 5400RPM 2.5" hard drives can barely do that in a sustained write, in my experience.
If you put two of those $140 CF cards in a striped RAID-0 with the dual-card Addonics adapter, you'll have a 32gb solid state disk, with a speed approaching 12 MB/s rate, for about $300 ($280 for the two CF cards, and $20 for the dual adapter).
Certainly cheaper than the 32gb solid-state disk for $430 from Transcend. And you can upgrade it piecewise as larger CF cards come out.
Those look really cool! I love the dual adapter, which allows you to put two compactflash cards into one notebook IDE slot: one master and one slave. Add Linux a little bit of software RAID 0, and you've got twice the bandwidth and twice the storage. Brilliant.
However, you bring up a good point: if the CF card doesn't support DMA, it will be quite slow. The one I linked to apparently doesn't support DMA
Hopefully other manufacturers will catch up quick, since DMA capabilities don't depend on the raw NAND flash chips, only on the controller chip... so the cost to manufacture a CF card supporting DMA should barely increase.
http://linux-wless.passys.nl/ That site is awesome. Thanks for the link! I've been hoping for a searchable database of linux-friendly wireless cards for a while (even thought about making my own)!
On the other hand, a 16gb CompactFlash card is only $140 . And the CompactFlash interface is electrically identical to IDE/PATA, so you can use a $5 mechanical adapter to connect a CompactFlash card to your notebook's hard drive bay.
What am I missing here???
Inquiring minds want to know. Maybe I can start selling cheapo 16gb solid state drives on eBay for $180 and make a killing
That's a pretty unfortunate situation if the unmaintained code is still actually used by someone.
(...)
That said, yeah, if someone notices that filesystem FooFS has been completely broken for ages and nobody has even noticed, then that's a pretty good argument for dropping it. But even then it's not just because it's unmaintained, it's because at that point you're pretty sure nobody really gives a crap about it.
The Linux kernel *almost never* drops support for any devices/filesystems unless (a) it's INCREDIBLY obsolete and NO ONE is using it, or (b) it's been superseded by something clearly better and there's a straightforward upgrade path.For example, if you read the kernel changelog summaries on LWN.net, you'll see that support for IBM PC/XT hard disks was only dropped in the last couple years... although they have been obsolete since the late 80s and perhaps literally no one has used them for 5-10 years. And support for the original "ext" filesystem was removed a few months ago, despite the fact that it's been completely superseded by ext2--which was introduced in 1993.
As Greg Kroah-Hartman has pointed out, the kernel developers are perfectly willing to maintain a driver for which only a single piece of hardware exists in the whole world!
Every time I ever wanted to upgrade *anything* bar the HD in my computer in the past *fifteen* years, I has to buy a new mobo. Might as well get new everything else. That's of course only if you upgrade every 2-3 years or so. More often than that, you may benefit. You're also blowing money for bragging rights. Hence ricer. Well... I don't agree. For example, with my current desktop, I'm perfectly happy with the HDD, graphics card, monitor, keyboard, mouse, optical drive (it's only about 1.5 years old), case, and power supply. But I could really use more RAM and a faster processor, so I'll probably go with a new motherboard, processor, and RAM... which will set me back about $300 compared to $700 for the whole new computer.
I'm definitely not a ricer. I've never bought a case with lots of LEDs, I don't go to LAN parties, I don't play 3D games... I mostly use my computer for software development, office-related stuff, web browsing, and as a server.
Off the top of my head, here are several categories of power users who really need upgradability and expandability to get their jobs done:
There are many cool and exciting new uses for laptops/PDAs/tablets, but desktops have many uses as well. For example, most computer users have a desk at home or work where they get a lot of work done: there's no need to have that computer be mobile, and desktops are CHEAPER and MORE UPGRADABLE and MORE RELIABLE.
Upgradibility in particular is a huge issue for power users and hardware enthusiasts:
By contrast, with my laptop, it takes maybe 5 minutes to replace the hard drive, and I have to mess with a bunch of fiddly little screws. To replace the RAM or optical drive I have to remove several panels and it probably takes 10-15 minutes. Replacing a MiniPCI wifi card is a huge pain and probably takes at least half an hour. And everything else simply can't be upgraded.
As far as I know, these cheap hardware RNGs use random fluctuations of thermal noise... meaning they basically sample the voltage across a little resistor and record the random fluctuations.
This is cheap and effective, but is vulnerable to a particularly nasty attack: cooling down the temperature. Imagine you have a data center cranking out random numbers using a thermal RNG. The FBI/CIA/MI5/Mossad/KGB/Elvis could pump in liquid nitrogen overnight until it's cold enough that the range of fluctuations is reduced to the point that they're no longer random enough... for example, if the RNG is designed to spit out numbers between 0 and 1.0 at room temperature, it might most only produce numbers between 0 and 0.25 at the temperature of liquid nitrogen. This could be enough to completely subvert some cryptographic systems.
Correction: that was supposed to read "In fact, this has been essentially disproved by experimental verification of Bell's theorem ..."
In fact, this has been essentially disproved by experimental verification of (building on the earlier Einstein-Podolsky-Rosen paradox).
Bell's theorem states that "No physical theory of local hidden variables can ever reproduce all of the predictions of quantum mechanics." In other words, no underlying mechanism could possibly produce the same results as quantum mechanics. I feel that this is the most elegant and fundamentally significant result of quantum mechanics.
So if random numbers are generated through quantum mechanical processes, their origin is in a very real sense not only unknown but truly unknowable.
(PS- I should mention that Bell's Theorem makes certain assumptions about locality, which is intimately related to general relativity. Both quantum mechanics and general relativity have been extremely well-tested, but it is known that they can't both be right in all circumstances... but to make a long story short, if you're not right next to a black hole then both quantum mechanics and general relativity should make sense.)
It's called Inotify. Everything and its mom uses it on Linux nowadays...
The PowerTop folks have been going around and finding a lot of programs that do polling rather than asynchronous notification, and fixing them to use things like inotify.
But when election time comes around, the politicians have to go running back to the people and promise to deliver on the issues that matter to them. Otherwise the other party's candidate will, and will win. Sure, it's messy and halting and pandering, but that's democracy in action
In the USA, we don't end up with makeshift parliamentary coalitions held together by fringe parties that hold the majority hostage to their agendas (such as I read about in Italy, Israel, France, etc.) That's why our system rewards centrism and promotes stability. If one party shifts right and is popular, the other party will shift right to catch up, which is what happens in the USA. Your Democrat party, in other countries, would be considered a fairly far right wing party. So what? Our political "center" is further to the right than that of, say, the Netherlands or Germany or the UK.
The American people as a whole would be considered "fairly far right wing" by the standards of many other democracies: our population is more religious, less concerned with the environment, broadly supports gun ownership rights, broadly supports Israel over the Palestinians, broadly supports capital punishment, etc.
This is pretty damn funny. Mod parent UP :-)
Everyone loves to hate that entrenched two-party duopoly, and I don't think the Founding Fathers intended it... but I think it's done a lot towards ensuring the stability and prosperity of the USA in the long view.
A plurality system inexorably pushes politicians towards the center, rather than the fringes. In countries with proportional representation, a fringe party can get into the legislature with, say, 3% of the national vote. In a plurality system, that party won't be more than a blip UNLESS it manages to gain a plurality of the vote in some regional district. So the plurality system encourages centrist politics, while at the same time respecting regional political differences.
Dell called, your new Vista PC is ready. Actually my new Dell runs Ubuntu 7.10. Actually, I dual boot Vista Home Premium to run EXACTLY ONE program (Altera Quartus II 7.1). And I hate pretty much every minute of it.
Dell doesn't lock its computers down to prevent me from running another OS. Sure, I'm no fan of the ~$30-50 OEM tax for Windows, but it's a far cry from an entire system that is enslaved to Apple and AT&T.
I actually paid just under $600 for this nice dual-core Turion 64 X2 TL-60 rig, with a 6-hour battery upgrade... so I paid LESS than the iPhone. And no subscription fees for this one.
Honestly... why would I pay $700 for the privilege of having not one but TWO large companies dictate to me how I can use a piece of hardware. Till I can buy a phone that lets me do whatever I want with it, they're all crap to me.
The iPhone craze has truly reduced epic proportions... so much so that my GRANDMA emails me and says she wants to buy me one. I mean, I don't want to be ungrateful, I'm probably the only person I know who actively DOES NOT WANT an iPhone, I should take advantage of my most-favored-grandson status, but.... yeeesssh!