Yes, most developers moved to Linux and stopped writing that pesky, unstable software that anyone actually uses.
Keeping a kernel that is 10 years behind the leading edge in file systems or communications, especially by kicking it all out of the kernel and saying "Naah-naah-naah! Not my problem!!!!" is like having a very secure car that doesn't have a reverse gear, seats, or door handles. It certainly helps contribute to stability. But the associated software to handle USB, firewire, packet filtering, or network file systems just isn't up to speed.
No, no, they're transforming our Victorian women into even more ghastly, skelatal, cold-blooded creatures!!! (http://www.guardian.co.uk/books/booksblog/2009/jan/30/jane-austen-zombies)
Resolving the large numbers of packages and their minor variations such as i386, x86_64, etc. across all the systems means a potentially large number of distinct Nagios query templates. Generating and managing those directly is a fascinating problem. Building it from scratch for a single in-house usage is a lot of work, usually better spent elsewhere.
It really makes me wish that RedHat Satellilte Server did _not_ require an Oracle backend. For RedHat themselves, with so many thousands of clients, it makes sense to use Oracle for 'RedHat Network', on which the Satellite Server is based. But for the rest of us, almost any lightweight database would have served better and required far fewer resources to run. But now that Oracle has bought Sun, I think we can rely on Oracle to insist that such corporate level supported services require Oracle, not MySQL.
No, I don't want to suggest Diebold has a monopoly on being too incompetent to be relied on to _prevent_ exploits. They're just surprisingly bad at it, and the processes that selected Diebold as a major vendor in this field are still around. And right now, they seem to be the major player in the electronic voting field.
You are still being unclear. Are you suggesting that OCR should be used instead of manual counting to help cut costs (which it does: manpower is often very, very expensive)? Or are you suggesting that OCR is prone to the same vulnerabilities as basic electronic counting (which it often is, especially if run by Diebold's historically incompetent, Windows based systems)?
Secure storage lockers are not as expensive as electronic kiosks. The _transportation_ and handling of the paper trail, to get them to the lockers, is expensive. And it's a large, recurring cost, in addition to that of the machines themselves. (Mechanical vote counting machines are also expensive, but after the Florida dangling chads fiasco, I think they've demonstrated many of their advantages.)
Mirroring CentOS or Fedora is fairly easy, although if you can afford it, please contribute back to the community by making your mirror available externally. (Rate limit it, but make it available: please don't be a freeloader.)
Mirroring RHEL, on which CentOS is based, is pretty awkward. Since the 'yum-rhn-plugin' knows to talk only to the authorized, secured RedHat repository or a RedHat Satellite Server, you pretty much have to find a way to build a mirror machine for _each RedHat Distribution_, whether i386 or x86_64, whether version 4.x or 5.x, whether 5Client or 5Server. This is a pain in the neck and sucks up licenses, or at least virtual instances on a server.
For RedHat? No. 'yum update' is fairly resource intensive. And changing applications in the middle of system operations is _BAD, BAD, BAD_ practice. I do _not_ want you silently updating MySQL or HTTPD in the midst of production services, because the update process may introduce an incompatibility with the existing configurations, especially if some fool has been doing things like installing CPAN or PHP modules without using RPM packages, or has manually stripped out and replaced Apache without RPM management.
And heaven forbid that you have a kernel with local modifications and special patches for special hardware whose version number is exceeded by the next RedHa kernel, and it replace yours, and the hardware fail to boot at the next reboot. That is very painful indeed to cope with if you haven't set up remote boot management or spent extra effort to lock down your packages.
We oldtimers, low uid or not, have been burned enough to know not to lick the frozen lightpole.
In theory, you can filter out mail to such addresses when you create them, or expire them very quickly to control the sharing of the spammable addresses. In practice, botnets seem quite likely to flood the potential address spaces and find your clever little tweaks. And the mail logs of large domains will also be very valuable to spammers to contact exactly such short-term addresses.
I do not understand your point. Manual vote counting typically takes at least double counting, close observation by trusted observers, and meticulous manual record keeping. That takes a lot of time and a lot of manpower. Storing the physical paper trail safely and securely as well is also expensive. And interpreting poorly filled in "ovals" as you describe them, or accidental marks on the page, can be difficult.
Are you saying to use OCR instead of manual counting? Please don't. That can provide the venue for just the sort of electronic frauds that Diebold has now become infamous for being too politically involved, too lax, and too incompetent to be certain does not occur.
Do you have _any_ idea of the expense of manual vote counting? It may cost more, it may cost less, but the overall cost is not wildly greater than that of manual vote counting.
Switzerland seems to do a good job with manual counting. But Switzerland is much smaller, and in many ways a lot saner than the US: they've also done a much better job of handling confidential banking for centuries, and of sane, respectful negotiations among diverse groups.
No about the sociopathy. One's mental state while under drugs is not the same as one's normal mental condition, and this sounds like the flying-high, stupid behavior of a teenager on cocaine. And her family didn't give her access to the car: she stole it from them. I'm really sorry for them: they were apparently starting the long, hard process of dealing with a cocaine user in the family. She doesn't deserve that insult: she might well seriously regret her behavior, when off the cocaine, that she engaged in. But she won't get the chance.
The article doesn't say how much cocaine was found in her system at the time of the accident. But it doesn't take being rich or privileged for a live-at-home teenager in trouble to steal a family car and get in a dangerous or even fatal accident. I'm remembering some of the crack users of my younger days, when crack became popular. It was devastating.) Mind you, it takes some wealth to afford cocaine rather than crack. And being pretty and young (not blonde, she had brown hair) can contribute to its availability. And I remember some parties I attended when younger: she'd have found it quite available: I don't think that's changed.
Now, all that said: we all need a sense of scale. A pretty white girl wiped out in a drug-related car accident is a family tragedy. But given the scale of other abuses in the world (such as genocide in places where we've created the local genocides, or wholesale rape as a day-to-day occurence in US prisons, or children starving to death worldwide), we need to look outside our privileged enclaves and go do some people some good. So get over this and go contribute some blood, or help with a Big Brother program for kids who really need a lot more help to avoid even worse fates.
I did an early testdrive on an old 486/33 box, to compare it to HURD's absolutely broken installation toolkit. It was awkward compared to modern distros, certainly, but the fact that it installed and gave me a full GNU toolchain was very exciting. It wasn't stable enough yet for business operation, but the fact that the kernel did what it needed to do in a basic install was the missing component of full free software distributions, and this was very exciting for political and business reasons.
Take a look at the twists and bends, especially at the ends. A wiring closet run by cretins will leave cable tips on the floor where they get stepped on, jammed through too-small ducts and the latches pulled off the ends, hand-strung cable that wasn't properly terminated but done in house "to save money", too-short cables stretched to their maximum to reach awkward places, bundles bound so tightly with tie-wraps that they're ripping the cladding off the cable and are likely to pinch them so hard they cause intermittent breaks, power cables woven in massive bundles around critical high frequency connections, over-long cables which were never bundled leaving a rat's nest on the floor, hundreds of unlabeled junctions organized "by convention" that the next technician violated, cables over the maximum length because people don't coount the distance between the desktop and the ceiling duct where the cable runs, cables damaged by doors and ceiling tiles left on them, etc.
This doesn't even mention the old 10Base2 problem of "I'll just stick a 10 foot side-cable from the middle without running a loop, it worked fine in my dorm!".
An annual wiring inspection and cleanup is a good idea in almost any production environment. Even if you don't replace cables, it's a good time to re-organize your existing setup and replace or expand equipment as needed. And while doing it, replace all possible tie-wraps with Velcro to ease pressure on the cables.
Of course, whether we are running _from_ or _to_ something makes a number of differences. I personally appreciate the messianic abilities of a good coward like Rincewind, of Diskworld fame, who in full flight easily skipped across modest bodies of water with barely a ripple to mark his passage.
Unless you've created a fan page for Chuck Berry: he was successfully sued in 1990 for videotaping women in the bathroom of his restaurant, including a camera in the seat (as I remember the lawsuit at the time).
Oh, my goodness. You haven't actually read any NDA contracts lately, haven't you? Between the issues of insider trading and giving away trade secrets and simple security concerns, I and many of my peers are very hamstrung in conversations about how and where we work. Such issues are precisely why I use an alias here: my opinions expressed here are my own, and my employers shouldn't be liable for them. Nor should I have to feel at risk for material I mention for which the NDA expired years ago, but which my claiming the direct experience and showing my resume for might lead to..... confusion about whose policy it is.
And insider trading is a big, big problem for important tech people. We know things about corporate roadmaps and failed technologies, and even about who's complaining about not getting their email reliably from a potential corporate buyer, or fixing the printer and finding pink slips in the output, or fixing the email and finding that the CFO is about to resign. (Both have happened to me.) So no, I cannot publicly discuss anything about my job that I wish to. And this is pretty normal, for most workplaces.
There is a distinction between the powers of an individual and a company, but it's not necessary an imbalance. The ability of a disgruntled employee to publish the security holes of a company, such as the dialup numbers and the passwords, is very powerful.
Check out the hardware at at www.crystalrugged.com. I got to play with some of their smaller plug-in units a few years ago, and while they're slightly underpower for the cost, they are _amazingly_ robust and very Telco suitable. I bet they'd be happy to provide servers ruggedized against salt and humidity.
And SSH and SSL keys. Save those so that people who've saved the keys don't get surprosed, especially for SSH, by their conflicting copies of old public keys.
Well, the simple hierarchical structure of UNIX file ownership isn't very complex. That's true. But do you really find that you need more, and that the Samba share granularity isn't enough for your uses?
Goodness, if possible, forget VM's. Use multiple OLPC systems, which are fiscally sensible, extremely low power, and startingly robust. Salt air and water is a problem: consider machines exposed to that for a year or so to be due for replacement.
If Windows or a UNIX system are necessary for basic software compatibility reasons, life is rather different.
Thunderbuses. Tee-hee. And you do raise interesting points.
"Virtualization" seems a slippery word here. There is always some disconnect between reality and mental models, modulated by the sensorium, yes? So all creatures have an internal sensory model of the universe. The idea that AI will develop from what is currently called "virtual reality" is.... intriguing. I'm not sure I buy it, precisely because I, personally, think that the sensorium is neglected and presents serious integration problems for AI research as I understand things.
While synesthesia for human infants seems... a reasonable idea, though not one I'd heard of, there is no need for synesthesia to explain the differing taste of food without the sense of smell. It's merely mis-classification, I think, much as people say they love someone from the heart, feel heartpangs, etc. I just don't see a need for syneshesia to explain that: do you, or find some reference to that as synesthesia?
Hmm. Fractal compression for memory? Hmm. I think holographic models work well, where the information is distributed across a wide span of the medium and recovered by a similar state of excitation. That doesn't take local fracturing of the data: are you aware of any actual fractal momory structures?
However, the porn site need not be illegal in and of itself: stuffing it full of other people's free porn is merely a fast way to set it up, cheaply, by someone who's a thief anyway. And to recognize a US citizen, wouldn't it work better if they _fail_ to solve a simple polynomial equation? At least for managers?
Wait: you've gone lightyears from your comments about the overall computing capacity of neural tissue versus that of the computational structures. Slow down, please: I can't confirm, refute, or agree with your points when you scatter so many non-standard models in a few paragraphs.
"This is how it works" wasn't my point: One of my points was that all awareness is very strongly bound to the senses. The eye preprocesses massively, and sends an already modeled view of the world to the brain. The brain, in turn, has neurological structures that interpret those shapes, edges, movement, colors, and references it to the rest of the creature's experience. To get genuine AI, or something recognizable as such, we'll need to pay a lot of attention to the perceptual abilities of the mechanism: that doesn't take virtual reality, that takes a sensorium.
Brain damage and damage to thought are a fascinating problem: brains, like other tissue, are highly adaptive. Memory is shown to be rather holographic by some fascinating experiments, but damage to key structures can destroy very specific abilities (such as language). Done properly, with hypothermic probes, some neurological disabling can even be done reversibly in animals (according to experiments I read about in college: fascinating stuff).
With that said, it seems a leap to say that the ability to process that data is entirely secondary. The sensory input, environmental control, and interaction with complex phenomona are necessary to _train_ the organism to what most of us would consider intelligence.
Slashdot Mirror
Yes, most developers moved to Linux and stopped writing that pesky, unstable software that anyone actually uses.
Keeping a kernel that is 10 years behind the leading edge in file systems or communications, especially by kicking it all out of the kernel and saying "Naah-naah-naah! Not my problem!!!!" is like having a very secure car that doesn't have a reverse gear, seats, or door handles. It certainly helps contribute to stability. But the associated software to handle USB, firewire, packet filtering, or network file systems just isn't up to speed.
No, no, they're transforming our Victorian women into even more ghastly, skelatal, cold-blooded creatures!!! (http://www.guardian.co.uk/books/booksblog/2009/jan/30/jane-austen-zombies)
Talk about "lie back and think of England".
Resolving the large numbers of packages and their minor variations such as i386, x86_64, etc. across all the systems means a potentially large number of distinct Nagios query templates. Generating and managing those directly is a fascinating problem. Building it from scratch for a single in-house usage is a lot of work, usually better spent elsewhere.
It really makes me wish that RedHat Satellilte Server did _not_ require an Oracle backend. For RedHat themselves, with so many thousands of clients, it makes sense to use Oracle for 'RedHat Network', on which the Satellite Server is based. But for the rest of us, almost any lightweight database would have served better and required far fewer resources to run. But now that Oracle has bought Sun, I think we can rely on Oracle to insist that such corporate level supported services require Oracle, not MySQL.
No, I don't want to suggest Diebold has a monopoly on being too incompetent to be relied on to _prevent_ exploits. They're just surprisingly bad at it, and the processes that selected Diebold as a major vendor in this field are still around. And right now, they seem to be the major player in the electronic voting field.
You are still being unclear. Are you suggesting that OCR should be used instead of manual counting to help cut costs (which it does: manpower is often very, very expensive)? Or are you suggesting that OCR is prone to the same vulnerabilities as basic electronic counting (which it often is, especially if run by Diebold's historically incompetent, Windows based systems)?
Secure storage lockers are not as expensive as electronic kiosks. The _transportation_ and handling of the paper trail, to get them to the lockers, is expensive. And it's a large, recurring cost, in addition to that of the machines themselves. (Mechanical vote counting machines are also expensive, but after the Florida dangling chads fiasco, I think they've demonstrated many of their advantages.)
Mirroring CentOS or Fedora is fairly easy, although if you can afford it, please contribute back to the community by making your mirror available externally. (Rate limit it, but make it available: please don't be a freeloader.)
Mirroring RHEL, on which CentOS is based, is pretty awkward. Since the 'yum-rhn-plugin' knows to talk only to the authorized, secured RedHat repository or a RedHat Satellite Server, you pretty much have to find a way to build a mirror machine for _each RedHat Distribution_, whether i386 or x86_64, whether version 4.x or 5.x, whether 5Client or 5Server. This is a pain in the neck and sucks up licenses, or at least virtual instances on a server.
For RedHat? No. 'yum update' is fairly resource intensive. And changing applications in the middle of system operations is _BAD, BAD, BAD_ practice. I do _not_ want you silently updating MySQL or HTTPD in the midst of production services, because the update process may introduce an incompatibility with the existing configurations, especially if some fool has been doing things like installing CPAN or PHP modules without using RPM packages, or has manually stripped out and replaced Apache without RPM management.
And heaven forbid that you have a kernel with local modifications and special patches for special hardware whose version number is exceeded by the next RedHa kernel, and it replace yours, and the hardware fail to boot at the next reboot. That is very painful indeed to cope with if you haven't set up remote boot management or spent extra effort to lock down your packages.
We oldtimers, low uid or not, have been burned enough to know not to lick the frozen lightpole.
You missed the RedHat model. 5. Insert Oracle server for no good reason. 6. Profit!
In theory, you can filter out mail to such addresses when you create them, or expire them very quickly to control the sharing of the spammable addresses. In practice, botnets seem quite likely to flood the potential address spaces and find your clever little tweaks. And the mail logs of large domains will also be very valuable to spammers to contact exactly such short-term addresses.
I do not understand your point. Manual vote counting typically takes at least double counting, close observation by trusted observers, and meticulous manual record keeping. That takes a lot of time and a lot of manpower. Storing the physical paper trail safely and securely as well is also expensive. And interpreting poorly filled in "ovals" as you describe them, or accidental marks on the page, can be difficult.
Are you saying to use OCR instead of manual counting? Please don't. That can provide the venue for just the sort of electronic frauds that Diebold has now become infamous for being too politically involved, too lax, and too incompetent to be certain does not occur.
Do you have _any_ idea of the expense of manual vote counting? It may cost more, it may cost less, but the overall cost is not wildly greater than that of manual vote counting.
Switzerland seems to do a good job with manual counting. But Switzerland is much smaller, and in many ways a lot saner than the US: they've also done a much better job of handling confidential banking for centuries, and of sane, respectful negotiations among diverse groups.
No about the sociopathy. One's mental state while under drugs is not the same as one's normal mental condition, and this sounds like the flying-high, stupid behavior of a teenager on cocaine. And her family didn't give her access to the car: she stole it from them. I'm really sorry for them: they were apparently starting the long, hard process of dealing with a cocaine user in the family. She doesn't deserve that insult: she might well seriously regret her behavior, when off the cocaine, that she engaged in. But she won't get the chance.
The article doesn't say how much cocaine was found in her system at the time of the accident. But it doesn't take being rich or privileged for a live-at-home teenager in trouble to steal a family car and get in a dangerous or even fatal accident. I'm remembering some of the crack users of my younger days, when crack became popular. It was devastating.) Mind you, it takes some wealth to afford cocaine rather than crack. And being pretty and young (not blonde, she had brown hair) can contribute to its availability. And I remember some parties I attended when younger: she'd have found it quite available: I don't think that's changed.
Now, all that said: we all need a sense of scale. A pretty white girl wiped out in a drug-related car accident is a family tragedy. But given the scale of other abuses in the world (such as genocide in places where we've created the local genocides, or wholesale rape as a day-to-day occurence in US prisons, or children starving to death worldwide), we need to look outside our privileged enclaves and go do some people some good. So get over this and go contribute some blood, or help with a Big Brother program for kids who really need a lot more help to avoid even worse fates.
And gcc. Don't forget the bootstrap recompilation issues for gcc: you needed a _lot_ of very expensive disk space to recompile gcc.
I did an early testdrive on an old 486/33 box, to compare it to HURD's absolutely broken installation toolkit. It was awkward compared to modern distros, certainly, but the fact that it installed and gave me a full GNU toolchain was very exciting. It wasn't stable enough yet for business operation, but the fact that the kernel did what it needed to do in a basic install was the missing component of full free software distributions, and this was very exciting for political and business reasons.
Take a look at the twists and bends, especially at the ends. A wiring closet run by cretins will leave cable tips on the floor where they get stepped on, jammed through too-small ducts and the latches pulled off the ends, hand-strung cable that wasn't properly terminated but done in house "to save money", too-short cables stretched to their maximum to reach awkward places, bundles bound so tightly with tie-wraps that they're ripping the cladding off the cable and are likely to pinch them so hard they cause intermittent breaks, power cables woven in massive bundles around critical high frequency connections, over-long cables which were never bundled leaving a rat's nest on the floor, hundreds of unlabeled junctions organized "by convention" that the next technician violated, cables over the maximum length because people don't coount the distance between the desktop and the ceiling duct where the cable runs, cables damaged by doors and ceiling tiles left on them, etc.
This doesn't even mention the old 10Base2 problem of "I'll just stick a 10 foot side-cable from the middle without running a loop, it worked fine in my dorm!".
An annual wiring inspection and cleanup is a good idea in almost any production environment. Even if you don't replace cables, it's a good time to re-organize your existing setup and replace or expand equipment as needed. And while doing it, replace all possible tie-wraps with Velcro to ease pressure on the cables.
Of course, whether we are running _from_ or _to_ something makes a number of differences. I personally appreciate the messianic abilities of a good coward like Rincewind, of Diskworld fame, who in full flight easily skipped across modest bodies of water with barely a ripple to mark his passage.
Unless you've created a fan page for Chuck Berry: he was successfully sued in 1990 for videotaping women in the bathroom of his restaurant, including a camera in the seat (as I remember the lawsuit at the time).
Oh, my goodness. You haven't actually read any NDA contracts lately, haven't you? Between the issues of insider trading and giving away trade secrets and simple security concerns, I and many of my peers are very hamstrung in conversations about how and where we work. Such issues are precisely why I use an alias here: my opinions expressed here are my own, and my employers shouldn't be liable for them. Nor should I have to feel at risk for material I mention for which the NDA expired years ago, but which my claiming the direct experience and showing my resume for might lead to..... confusion about whose policy it is.
And insider trading is a big, big problem for important tech people. We know things about corporate roadmaps and failed technologies, and even about who's complaining about not getting their email reliably from a potential corporate buyer, or fixing the printer and finding pink slips in the output, or fixing the email and finding that the CFO is about to resign. (Both have happened to me.) So no, I cannot publicly discuss anything about my job that I wish to. And this is pretty normal, for most workplaces.
There is a distinction between the powers of an individual and a company, but it's not necessary an imbalance. The ability of a disgruntled employee to publish the security holes of a company, such as the dialup numbers and the passwords, is very powerful.
Check out the hardware at at www.crystalrugged.com. I got to play with some of their smaller plug-in units a few years ago, and while they're slightly underpower for the cost, they are _amazingly_ robust and very Telco suitable. I bet they'd be happy to provide servers ruggedized against salt and humidity.
And SSH and SSL keys. Save those so that people who've saved the keys don't get surprosed, especially for SSH, by their conflicting copies of old public keys.
Well, the simple hierarchical structure of UNIX file ownership isn't very complex. That's true. But do you really find that you need more, and that the Samba share granularity isn't enough for your uses?
The 1 TB goes in a separate USB box. Cheap, replaceable, surprisingly robust.
Goodness, if possible, forget VM's. Use multiple OLPC systems, which are fiscally sensible, extremely low power, and startingly robust. Salt air and water is a problem: consider machines exposed to that for a year or so to be due for replacement.
If Windows or a UNIX system are necessary for basic software compatibility reasons, life is rather different.
Thunderbuses. Tee-hee. And you do raise interesting points.
"Virtualization" seems a slippery word here. There is always some disconnect between reality and mental models, modulated by the sensorium, yes? So all creatures have an internal sensory model of the universe. The idea that AI will develop from what is currently called "virtual reality" is.... intriguing. I'm not sure I buy it, precisely because I, personally, think that the sensorium is neglected and presents serious integration problems for AI research as I understand things.
While synesthesia for human infants seems... a reasonable idea, though not one I'd heard of, there is no need for synesthesia to explain the differing taste of food without the sense of smell. It's merely mis-classification, I think, much as people say they love someone from the heart, feel heartpangs, etc. I just don't see a need for syneshesia to explain that: do you, or find some reference to that as synesthesia?
Hmm. Fractal compression for memory? Hmm. I think holographic models work well, where the information is distributed across a wide span of the medium and recovered by a similar state of excitation. That doesn't take local fracturing of the data: are you aware of any actual fractal momory structures?
Oh, my. I like the idea.
However, the porn site need not be illegal in and of itself: stuffing it full of other people's free porn is merely a fast way to set it up, cheaply, by someone who's a thief anyway. And to recognize a US citizen, wouldn't it work better if they _fail_ to solve a simple polynomial equation? At least for managers?
Wait: you've gone lightyears from your comments about the overall computing capacity of neural tissue versus that of the computational structures. Slow down, please: I can't confirm, refute, or agree with your points when you scatter so many non-standard models in a few paragraphs.
"This is how it works" wasn't my point: One of my points was that all awareness is very strongly bound to the senses. The eye preprocesses massively, and sends an already modeled view of the world to the brain. The brain, in turn, has neurological structures that interpret those shapes, edges, movement, colors, and references it to the rest of the creature's experience. To get genuine AI, or something recognizable as such, we'll need to pay a lot of attention to the perceptual abilities of the mechanism: that doesn't take virtual reality, that takes a sensorium.
Brain damage and damage to thought are a fascinating problem: brains, like other tissue, are highly adaptive. Memory is shown to be rather holographic by some fascinating experiments, but damage to key structures can destroy very specific abilities (such as language). Done properly, with hypothermic probes, some neurological disabling can even be done reversibly in animals (according to experiments I read about in college: fascinating stuff).
With that said, it seems a leap to say that the ability to process that data is entirely secondary. The sensory input, environmental control, and interaction with complex phenomona are necessary to _train_ the organism to what most of us would consider intelligence.