Yep. There's a reason people are willing to spend orders of magnitude more money on "server grade" hardware (hot swap RAIDed drives, redundant hot swap power supplies, etc) than on the equivalent horsepower in desktop systems, and even to have redundant servers and systems on hot standby.
Suppose your airline or nationwide hotel chain reservation system goes out. Or the control system for your semiconductor fab dies. I work with a guy who did some work for a stock exchange -- systems on carts on hot standby, if so much as a mouse acted up they'd swap out the whole system, it was faster (thus cheaper) than trying to diagnose and fix the problem on the spot.
Okay, I'll admit there might be some bias because I work for HP, but I'm on the managed hosting for services customers side so I'm a customer of HP's hardware support, I call the same 800 number any other customer does. And yes, the HP (nee Compaq) Proliant server line is pretty good when it comes to redundant hardware that fails operational (hot swappable, etc), but the few times where something serious has gone wrong (eg system board failure in an 8-way Xeon in the middle of the night) they've had the tech and the part on site in less than four hours.
Usually though the part gives us warning that failure is impending (eg disks or memory) and hardware support just delivers the part to me the next day for a hot swap.
Ironically, UFS is not as portable as it should seem to be (lack of Linux support the last time I checked).
Linux supports UFS, and there's even some support for UFS2 (ro support since 2.6.5). The portability problem is because various Unix vendors added their own extensions to UFS, which aren't necessarily compatible with each other. This also limits Linux's ability to support particular variations of UFS.
I don't know if such gizmos exist for modern tape cartridges, but back in the day when we still used open reel 9-track tapes, it was standard practice to run a new tape through a tape cleaner before first used. That scraped off any loose oxide particles, as well as dirt, etc. We'd periodically do that after using the tape a certain number of times, too.
I've heard of doing that as a step in recovering data from a bad tape -- even a thinned out oxide layer retains some of the original magnetic field, and cleaning it keeps the heads from being confused by oxide particles from elsewhere on the tape -- but never had to resort to that.
... every single Linux distribution I've seen for the past years has used/var/www for storing web pages?
Guess you haven't seen SUSE then, or indeed any LSB-compliant (or FHS 2.3 compliant) distro then.
The Filesystem Hierarchy Standard says to put that stuff under/srv (typically in/srv/www) and the Linux Standard Base says to follow the FHS. FHS 2.3 has been the standard for nearly two years, since January 2004.
Just because RedHat screws it up... (At least AS4 has/srv)
The only parts which are repeat manufactured are SRBs and those are extremely simple.
No, the SRB's are re-used (although "crashed and salvaged" might be more accurate). The ET's are disposed of and manufactured new for each flight.
The dies and CNC code is what is costed [sic] most in manufacturing.
Right, so it doesn't cost much more to build 30 than it does to build 5. Even if you're largely hand-building the things (vs using CNC), you still need dies and templates. Besides, with any halfway modern CADD system, the CNC codes are just another output.
And technicians are much faster at servicing one part instead of 4.
They're not talking about one part. One combustion chamber and nozzle, perhaps, but three electric motors and some hairy mechanical linkages and flexible plumbing, plus valves. That's a lot more to maintain than four simple chambers and nozzles and no moving parts except the valves.
We are talking about replacing multiple pieces with one piece
No, you're talking about replacing simple, rigid plumbing fixtures with something combining flexible plumbing and electric motors, adding at least five points of failure (3 motors plus flex fuel and oxidizer lines). More than that when you consider the linkages and failure points within each motor.
(I'm not counting valves -- yes the gimballed only has one set vs a quad's four sets, but (a) valves are pretty simple and (b) design of 4-way series/parallel valves to allow fail-operational in case of a valve stuck open or a valve stuck closed has about 50 years of aerospace engineering experience behind it.)
Several grams can cost on the order of thousands of dollars to reach high earth orbit.
Well, we're talking manned ops, so low Earth orbit, not high Earth orbit. But yes, if you design to weight rather than designing to cost, stuff costs a lot more. There hasn't been a real operational need (vs engineering desire or habit) to design to weight since the Apollo program. (NB, obviously weight is a factor that can't be totally ignored, the mistake is making it the primary criterion rather than cost and safety. The Soviets/Russians had better heavy-lift capability from the start, so they never fell into that trap. That may have cost them the Moon, -- doing Moon direct or even Lunar Orbit Rendezvous is weight critical -- but they dominate LEO.) (Weight is less of an issue for the Moon if you're doing an Earth Orbit Rendezvous rather than trying to lift everything in one shot.)
The nice thing about repetitive structures is that they lead to cost reductions through volume manufacturing and assembly. They're also easier to train technicians to service.
The nice thing about fixed quads for an RCS system is that their only moving parts are the valves -- pretty simple parts. Moving parts are failure points.
The nice thing about multiple quads vs just a few gimballed thrusters is that they provide redundancy and backup for each other.
But then NASA has always been in love with the complex solution that might save a few grams of weight, and require a few more engineers and technicians to manage it -- which return requires a few more managers to manage them, and so on...
If Microsoft made some error in implementing their own Win32 API, i.e. not to the correct specification,
According to Microsoft, if such a situation were to arise, the specification would be in error. Where differences exist between Microsoft's documentation and Microsoft's implementation, the implementation is correct. (At least in released code.)
Macro evolution has yet to be tested experimentally,
Nonsense. There's no difference between macro and micro evolution -- that distinction was invented by creationists to explain away the observed evolution of e.g. antibiotic resistance in bacteria as mere "micro" evolution.
If macro evolution applies to very clear and obvious morphological differences between species, then we can predict that the degree of difference between species at the "macro" (morphological) level will be mirrored by the degree of difference at the "micro" (DNA) level. We know DNA changes happen and can be selected for over time (so called micro evolution).
And indeed, gene sequencing of more and more species reveals that to be the case. So, we have a testable prediction -- accumulated changes at the genetic level will result in changes at the morphological level -- which seems to be the case. If we ever turn up a pair of very dissimilar species that have nearly identical gene sequences, we'll have to rethink that. Ditto if we turn up two very similar species that have massively different genomes.
the drake equation calculates our odds of contacting ANOTHER civilization
No, it doesn't. It calculates the number of civilizations capable of contact, based on assumptions about the odds of the various factors.
Even if one knows the number of contact-capable civilizations, computing the odds of actual contact requires assumptions about a bunch of other factors not in the Drake equation.
palettes were a horrible hack, and I'm very glad to have seen the last of them
Oh, they have their uses, particulary in an app where you want to visually manipulate all pixels with value = N in some way, without wasting time. I.e. instead of doing something like:
for (x=0; x < image.width; x++) {
for (y=0; y < image.height; y++) {
if (pixel_at(image, x, y) == N) {
set_pixel(image, x, y, some_new_value);
}
}
}
I'd rather just do something like
set_palette(N, some_new_value);
This is especially true where I want to highlight something (a feature on a map, say) by blinking it (toggling between two different colors). Although it's true enough that simple one-time changes (eg using false color to emphasize a subtle difference in a grayscale image) are pretty fast on modern hardware.
X lets you (in theory, and if the hardware supports it) do different visuals for different windows, so you could have 24-bit direct in most of the display with selected 8-bit pseudocolor windows for apps that need it. In theory. If the hardware supports it.
That may be one reason why Disney doesn't use the official DVD logo anywhere on their discs or packaging. Instead, they're "Disney DVD"s, with Disney's own logo for them.
How much of that $215 billion would have been spent anyway on routine costs (including housing, training, etc) if the deployed forces had just stayed home? Sure, it costs a lot to deploy, but even a standing army (navy, air force...) ain't cheap.
In other words, what's the incremental cost of the war in Iraq?
It's okay if the guy had previously signed a security agreement saying that it was okay, or there were prominently placed notices that bags brought onto the premises were subject to search. He's always free not to sign such (and not be allowed in) or not to bring bags (briefcases, etc) onto the premises.
Both fairly common practices in places that worry about that sort of thing.
Now, maybe that ranger was clueless or maybe by "the region" he meant central Florida -- the Florida panther's range is more southwestern Florida -- but it's not like the critters couldn't walk from A to B if they had a mind to.
If that is what you saw, consider yourself lucky to have seen it. (There probably wasn't much risk to you -- if the behaviour is anything like the mountain lions around these parts (Colorado), it'd leave two adults on bikes alone. A kid or a dog on its own, on the other hand...)
So what if it does? We have no shortage of altruistic knowledge-seekers, especially if the chilling effect of potential lawsuits by greedy knowledge seekers is reduced.
But whether it does or doesn't, that's irrelevant to the topic at hand, which is discussing trying to prevent the lock-up or proprietization of knowledge which is already free and has been for hundreds of years.
From the US PTO's track record of granting patents to almost anyone who pays the fee, and ignoring any "prior art" that isn't in a previous patent (and sometimes not even then), this may be futile.
Oh, it is certainly worth doing, and I applaud the effort. Not every country's patent system is as messed up as the US's is.
In 500m years, the continents will have drifted enough for North America to have smashed into the east coast of Asia. (It won't happen exactly like that, but that gives the scale). The only traces of humankind left will be on the Moon, or buried thousands of feet underground.
Heck, the dinosaur stomping grounds of only 80m years ago are hundreds of feet below ground here, except for a few miles to the west where the rock has been bent through about 45 degrees as the Rocky Mountains uplifted. (I'm a few miles from Morrison, CO). The dinosaurs could have strip-mined half the planet and we wouldn't see much evidence of it.
Slashdot Mirror
Some of us -- said author included -- spell it the civilized way, not that heathen sassenach variation. ;-)
-- Alastair (not Reynolds)
Yep. There's a reason people are willing to spend orders of magnitude more money on "server grade" hardware (hot swap RAIDed drives, redundant hot swap power supplies, etc) than on the equivalent horsepower in desktop systems, and even to have redundant servers and systems on hot standby.
Suppose your airline or nationwide hotel chain reservation system goes out. Or the control system for your semiconductor fab dies. I work with a guy who did some work for a stock exchange -- systems on carts on hot standby, if so much as a mouse acted up they'd swap out the whole system, it was faster (thus cheaper) than trying to diagnose and fix the problem on the spot.
Okay, I'll admit there might be some bias because I work for HP, but I'm on the managed hosting for services customers side so I'm a customer of HP's hardware support, I call the same 800 number any other customer does. And yes, the HP (nee Compaq) Proliant server line is pretty good when it comes to redundant hardware that fails operational (hot swappable, etc), but the few times where something serious has gone wrong (eg system board failure in an 8-way Xeon in the middle of the night) they've had the tech and the part on site in less than four hours.
Usually though the part gives us warning that failure is impending (eg disks or memory) and hardware support just delivers the part to me the next day for a hot swap.
It would be neat if they did have extraterrestrial craft but if we had such technology, why would we still be dependent on oil?
A severe shortage of dilithium crystals.
Also, the naquada was all mined out several thousand years ago...
Ironically, UFS is not as portable as it should seem to be (lack of Linux support the last time I checked).
Linux supports UFS, and there's even some support for UFS2 (ro support since 2.6.5). The portability problem is because various Unix vendors added their own extensions to UFS, which aren't necessarily compatible with each other. This also limits Linux's ability to support particular variations of UFS.
I don't know if such gizmos exist for modern tape cartridges, but back in the day when we still used open reel 9-track tapes, it was standard practice to run a new tape through a tape cleaner before first used. That scraped off any loose oxide particles, as well as dirt, etc. We'd periodically do that after using the tape a certain number of times, too.
I've heard of doing that as a step in recovering data from a bad tape -- even a thinned out oxide layer retains some of the original magnetic field, and cleaning it keeps the heads from being confused by oxide particles from elsewhere on the tape -- but never had to resort to that.
... every single Linux distribution I've seen for the past years has used /var/www for storing web pages?
/srv (typically in /srv/www) and the Linux Standard Base says to follow the FHS. FHS 2.3 has been the standard for nearly two years, since January 2004.
/srv)
Guess you haven't seen SUSE then, or indeed any LSB-compliant (or FHS 2.3 compliant) distro then.
The Filesystem Hierarchy Standard says to put that stuff under
Just because RedHat screws it up... (At least AS4 has
The only parts which are repeat manufactured are SRBs and those are extremely simple.
No, the SRB's are re-used (although "crashed and salvaged" might be more accurate). The ET's are disposed of and manufactured new for each flight.
The dies and CNC code is what is costed [sic] most in manufacturing.
Right, so it doesn't cost much more to build 30 than it does to build 5. Even if you're largely hand-building the things (vs using CNC), you still need dies and templates. Besides, with any halfway modern CADD system, the CNC codes are just another output.
And technicians are much faster at servicing one part instead of 4.
They're not talking about one part. One combustion chamber and nozzle, perhaps, but three electric motors and some hairy mechanical linkages and flexible plumbing, plus valves. That's a lot more to maintain than four simple chambers and nozzles and no moving parts except the valves.
We are talking about replacing multiple pieces with one piece
No, you're talking about replacing simple, rigid plumbing fixtures with something combining flexible plumbing and electric motors, adding at least five points of failure (3 motors plus flex fuel and oxidizer lines). More than that when you consider the linkages and failure points within each motor.
(I'm not counting valves -- yes the gimballed only has one set vs a quad's four sets, but (a) valves are pretty simple and (b) design of 4-way series/parallel valves to allow fail-operational in case of a valve stuck open or a valve stuck closed has about 50 years of aerospace engineering experience behind it.)
Several grams can cost on the order of thousands of dollars to reach high earth orbit.
Well, we're talking manned ops, so low Earth orbit, not high Earth orbit. But yes, if you design to weight rather than designing to cost, stuff costs a lot more. There hasn't been a real operational need (vs engineering desire or habit) to design to weight since the Apollo program. (NB, obviously weight is a factor that can't be totally ignored, the mistake is making it the primary criterion rather than cost and safety. The Soviets/Russians had better heavy-lift capability from the start, so they never fell into that trap. That may have cost them the Moon, -- doing Moon direct or even Lunar Orbit Rendezvous is weight critical -- but they dominate LEO.) (Weight is less of an issue for the Moon if you're doing an Earth Orbit Rendezvous rather than trying to lift everything in one shot.)
The nice thing about repetitive structures is that they lead to cost reductions through volume manufacturing and assembly. They're also easier to train technicians to service.
The nice thing about fixed quads for an RCS system is that their only moving parts are the valves -- pretty simple parts. Moving parts are failure points.
The nice thing about multiple quads vs just a few gimballed thrusters is that they provide redundancy and backup for each other.
But then NASA has always been in love with the complex solution that might save a few grams of weight, and require a few more engineers and technicians to manage it -- which return requires a few more managers to manage them, and so on...
If Microsoft made some error in implementing their own Win32 API, i.e. not to the correct specification,
According to Microsoft, if such a situation were to arise, the specification would be in error. Where differences exist between Microsoft's documentation and Microsoft's implementation, the implementation is correct. (At least in released code.)
Jet fuel is kerosene. Hell of a lot less flammable than gasoline or even alcohol.
Macro evolution has yet to be tested experimentally,
Nonsense. There's no difference between macro and micro evolution -- that distinction was invented by creationists to explain away the observed evolution of e.g. antibiotic resistance in bacteria as mere "micro" evolution.
If macro evolution applies to very clear and obvious morphological differences between species, then we can predict that the degree of difference between species at the "macro" (morphological) level will be mirrored by the degree of difference at the "micro" (DNA) level. We know DNA changes happen and can be selected for over time (so called micro evolution).
And indeed, gene sequencing of more and more species reveals that to be the case. So, we have a testable prediction -- accumulated changes at the genetic level will result in changes at the morphological level -- which seems to be the case. If we ever turn up a pair of very dissimilar species that have nearly identical gene sequences, we'll have to rethink that. Ditto if we turn up two very similar species that have massively different genomes.
I'm not holding my breath.
the drake equation calculates our odds of contacting ANOTHER civilization
No, it doesn't. It calculates the number of civilizations capable of contact, based on assumptions about the odds of the various factors.
Even if one knows the number of contact-capable civilizations, computing the odds of actual contact requires assumptions about a bunch of other factors not in the Drake equation.
palettes were a horrible hack, and I'm very glad to have seen the last of them
Oh, they have their uses, particulary in an app where you want to visually manipulate all pixels with value = N in some way, without wasting time. I.e. instead of doing something like:
for (x=0; x < image.width; x++) {
for (y=0; y < image.height; y++) {
if (pixel_at(image, x, y) == N) {
set_pixel(image, x, y, some_new_value);
}
}
}
I'd rather just do something like
set_palette(N, some_new_value);
This is especially true where I want to highlight something (a feature on a map, say) by blinking it (toggling between two different colors). Although it's true enough that simple one-time changes (eg using false color to emphasize a subtle difference in a grayscale image) are pretty fast on modern hardware.
X lets you (in theory, and if the hardware supports it) do different visuals for different windows, so you could have 24-bit direct in most of the display with selected 8-bit pseudocolor windows for apps that need it. In theory. If the hardware supports it.
That may be one reason why Disney doesn't use the official DVD logo anywhere on their discs or packaging. Instead, they're "Disney DVD"s, with Disney's own logo for them.
How much of that $215 billion would have been spent anyway on routine costs (including housing, training, etc) if the deployed forces had just stayed home? Sure, it costs a lot to deploy, but even a standing army (navy, air force...) ain't cheap.
In other words, what's the incremental cost of the war in Iraq?
To use AMD chips.
(Although personally I'd prefer to avoid the whole x86 thing altogether. I compromise with x86-64, and some older PPC and Sparc boxes.)
Hey, some of us 5-digit UID "kiddies" read that when it first appeared in CACM.
Those 6-digit whippersnappers, now...
It's okay if the guy had previously signed a security agreement saying that it was okay, or there were prominently placed notices that bags brought onto the premises were subject to search. He's always free not to sign such (and not be allowed in) or not to bring bags (briefcases, etc) onto the premises.
Both fairly common practices in places that worry about that sort of thing.
Um, google for "florida panther".
Now, maybe that ranger was clueless or maybe by "the region" he meant central Florida -- the Florida panther's range is more southwestern Florida -- but it's not like the critters couldn't walk from A to B if they had a mind to.
If that is what you saw, consider yourself lucky to have seen it. (There probably wasn't much risk to you -- if the behaviour is anything like the mountain lions around these parts (Colorado), it'd leave two adults on bikes alone. A kid or a dog on its own, on the other hand...)
So what if it does? We have no shortage of altruistic knowledge-seekers, especially if the chilling effect of potential lawsuits by greedy knowledge seekers is reduced.
But whether it does or doesn't, that's irrelevant to the topic at hand, which is discussing trying to prevent the lock-up or proprietization of knowledge which is already free and has been for hundreds of years.
From the US PTO's track record of granting patents to almost anyone who pays the fee, and ignoring any "prior art" that isn't in a previous patent (and sometimes not even then), this may be futile.
Oh, it is certainly worth doing, and I applaud the effort. Not every country's patent system is as messed up as the US's is.
Ajax is (or was) also a laundry detergent. Remember the "stronger than dirt" ad slogan?
So, while it's great France may have had success during the age of the longbow,
Um, no. See the Battle of Agincourt.
In 500m years, the continents will have drifted enough for North America to have smashed into the east coast of Asia. (It won't happen exactly like that, but that gives the scale). The only traces of humankind left will be on the Moon, or buried thousands of feet underground.
Heck, the dinosaur stomping grounds of only 80m years ago are hundreds of feet below ground here, except for a few miles to the west where the rock has been bent through about 45 degrees as the Rocky Mountains uplifted. (I'm a few miles from Morrison, CO). The dinosaurs could have strip-mined half the planet and we wouldn't see much evidence of it.