What I couldn't find was an explanation of what the "crypto" stuff in that 2D barcode is. I think I've seen something about some other similar systems where there's some mathematical voodoo that goes on that lets you use that data to tally a group of votes without being able to know the details of the individual votes.
Regardless of the details, I think that the data in that barcode has stuff that allows anyone to verify that a) it is an authentic ballot and b) given the full set of ballots, recompute the tally on their own and compare that to the published results.
Wish I could remember where I saw that or how it worked though.
The mechanism shown in the video actually does address this. The voting form layout is randomized, and the "key" portion (the bit matching checkboxes with candidates) is destroyed at the voting location. The receipt shows your marks, so you can verify that the marks counted match the marks you made, but it does not show what those marks mean. Even if someone forces you to give up your receipt, they have no way of knowing how you voted, only that the system recorded the vote correctly.
Yes, storing and providing full text search over a large pile of email consumes resources... duuuh?
Also they're measuring the performance of Thunderbird while converting to the new system, not in its steady state. This is like complaining that Firefox uses a lot more CPU importing settings from IE than IE uses when looking at your home page.
Their claim as to how long it took to do the full text indexing of the mail seems dubious to me. I've got a similar amount of mail, and the time it took to index was more like minutes, not days.
The other thing to do if you want longish term reliability is to add redundancy to whatever you're storing with a tool like par2, http://www.par2.net/ and http://www.quickpar.org.uk/ are your friend.
Raid5 will help you if you lose a whole drive (e.g. siezes up from sitting still for a long time), the par2 data will both allow you to verify that the data hasn't been corrupted, and if it is (e.g. a couple sectors go bad), it will let you recover the data.
Jeez, I remember those. The home PC when I was a wee one used a bunch of their stuff for a Northstar Z80 based system that ran CP/M. Had Wordstar, a copy of Logo by the Lisp Corporation, I think my parents used an ancient version of DBase, and an early version of Microsoft Money. Somewhere in my place or theirs, I think there's even a poster with the elephant:)
My downstairs neighbor had FIOS put in. In addition to disconnecting the neighbor's copper all the way to the pole, the wonderfully helpful FIOS installer:
Cut the copper line where it entered my building
Filled the hole in the wall with silicone goo (preventing rewiring of the copper)
Disconnected the copper all the way up to the pole
Changed/disconnected my copper connection at the CO
Plugged the FIOS unit in the basement into an outlet that is on my electric meter
It took me 3 weeks of fighting with Verizon (who insisted on taking 2 days to make a service appointment window, and insisted that they be 8a-7p) and my DSL provider (who was horribly frustrated by their inability to get Verizon to simply run a clean *bleeping* loop) to get things back up and running.
In your Finland example, if your hypothesis were that eating ice-cream caused drowning, you would take a random sample of people who ate ice-cream, randomise every significant factor (time of year, time of day, weather, location etc.) and then you have to show that with a statistical certainty of >0.95 (or so) that eating ice-cream is a factor that divides your samples into the two groups of those who die and those who don't die. For this particular hypothesis you're very likely to reject your alternative hypothesis at this point. This is impossible to do with an observational study, and unnecessary with a controlled one.
To do what you described with an observational study you would need two things that are unavailable:
Knowledge of every possible contributing factor for the outcomes in question
A subject for every possible combination of every contributing factor
Even in a very simple study, you could not enumerate all the possible causes because you do not know what might be causes, and you cannot find enough subjects because even with a fairly short list of causes, you end up with more combinations than there are people on the planet. Even with clever design so that you don't need a subject for every possible combination, you still won't find enough people on the planet to conduct your study, much less enough people going to the beach in Finland.
And all that is unnecessary anyways. The whole point of the controlled study described by Frequency Domain is that you vary single hypothesized causes, leaving all else the same, and look for changes in outcome. If you are only changing one variable, then the other factors will retain their distribution, unless they are causatively changed by the variable in question. This non-changing aspect is further reinforced by the nature of a double blind randomized study. In such a study, the patient population is randomly distributed among the possible groups (e.g. gets drug, gets placebo), with information escrowed in such a way that neither the patient nor the patient's doctor nor the people conducting the study know who fell into which group until the study is ended and the information retrieved from escrow.
So, with such a study setup, you can indeed rule out unrelated causes with a very high degree of confidence.
However, with an observational study in which you cannot tweak variables in a controlled fashion to measure outcomes, you are left with only ever being able to provide statistical confidences on correlations, never on causes.
If you know the positioning of each of the three dots, you can still do a certain amount of sub-pixel rendering, just not in quite the same way. Doing some more research, I think this was just on the Apple II, which may well have used a trinitron tube and thus used the same style of subpixel rendering as ClearType/CoolType/etc.
Obviously, few (if any) business users need anything more than a Pentium III running at 500 MHz. That processor is perfectly acceptable for business applications like OpenOffice.
Obviously you haven't gone and looked at what gets installed on many business PCs. My employer's standard systems are 2.4ghz p4 on the desktop, 1.7gz p4m on the laptops, 512-1025m ram depending on system usage.
Everyone complains that they're slow. Why? Lets see:
Software distribution systems that check through 5000 packages every 2-4 hours to check for critical updates.
Office addins chiefly our document management system. Ripping all the addins out of Word cuts its startup time by a factor of 5 for me, approximately. However, I'm pretty rare in not needing any of those addins.
For a stripped down system running just the office apps with no addins and some basic virus scanning software, yes, old hardware does fine. But a business desktop does so much more than that, because a business needs to do more things to manage their computers. And the home user, who doesn't have all that, uses the processor power for games and goofing around with their pictures and home videos.
And never underestimate how many business users don't just sit there making office documents. There are large numbers that do development, visualization, number crunching, and other compute intensive tasks on a regular, if not continuous basis.
And whatever makes people think that these fancy web applications need less horsepower? JavaScript + DHTML is going to be less efficient for UI work than native code, thus requiring a faster processor.
I always figured it said that because the whole sequence of movies was meant to be a tale / documentary from the future about how things got to be the way they were. That is, the events of SW are in the future for us, but in the far past from the point of view of the storyteller.
Learing how your system works is great. But I already know. Most of the time when I want to install linux on a new system, I just want to get it installed. Not having excess packages is great, and control over the install is great, but extra typing and waiting is not. In that vein, RedHat's installer is tolerable, but not optimal in that it's hard to get as minimal an install as I'd like. Debian's installer is pretty good, though it requires doing a bit more manual work than I'd like. Anything more manual than Debian's installer is just too much for me.
When it comes to Linux newbies, install time isn't the time to learn about the system. You want to learn about the system when you have all the emotionally-satisfying and high usability eye candy and applications installed. If people have to read lots of documentation in order to install the system, they are likely to get frustrated and possibly give up. Of course, there's more motivation to learn how the system works when you need to do so in order to install it. On the other hand, if the goal is world domination, you don't want to limit it to only the die hard techies.
And as much as I'm a command line junkie, I don't think joe user should be required to to be one too in order to do the basic stuff with his computer. Expecting joe admin to be savvy about the commandline is well and good, but not joe user.
In my experience, on any filesystem over a few gigs in size, ReiserFS's overhead is actually smaller than ext2/3. I converted a 240-ish gig partition from ext3 to reiserfs and regained something on the order of 15-20 gigs as I recall. This wasn't due to tail packing, either, as most of the files on that partition were many many megs in size.
Go get a goddamn calculator, as your brain is obviously incapable of performing arithmetic:
1000/1024 ~= 0.977
So a 10k thing in hard drive k will really only be 9.77k in computer k. That's called overstating (or overestimating to use your poor vocabulary -- it's stating sizes, not estimating them!) not understating. This gross lack of brain-usage on your part seems to lead to your subsequent thrashing about for bullshit justifications for the manufacturers.
Partition tables and boot sectors combined take less than a few k. Hardly noticeable on a freaking floppy disk, let alone an 80G hard drive.
And computer users don't sue about the OS using RAM because the RAM is there, it's just being used! They're suing the HD manufacturers because the space just isn't there. Ditto with your filesystem overhead remark.
As far as bad sectors are concerned, the days of the OS having to deal with bad sectors are almost gone. All modern HDs have space space inside that they use to transparently deal with sectors going bad.
Why is everyone so bound and determined to justify the hard drive manufacturers in this? Regardless of the technical correctness, the intent of the manufacturers was to bullshit consumers pure and simple. That is illegal and punishable. And as far as correctness is concerned, some standards body saying "you should do XYZ" is far from XYZ becoming an accepted and established standard. Standards have to be adopted in order to be meaningful and valid!
Yes, but it's a two-by-four, not a two inch by four inch piece of wood. Even a high-school-non-graduate working at a construction site doesn't tend to have problems with this.
Most of the time standards bodies set down standards that reflect the existing reality, and if they change it at all, it is a minimal change to allow a precise and unambiguous definition (e.g. converting to atomic wavelengths instead of the length of a particular hunk of metal). For them to declare a standard that is blatantly at odds with the state of the industry is, frankly bullshit.
And when it comes to deceptive advertising, if I'm not mistaken, intent plays a role. A company that, through an honest mistake (typo, whatever), mis-advertises the price/size of an item for sale, they do not generally get held liable. When a company deliberately miscontrues information or represents it in a deliberately confusing manner, that's a different case.
The HD manufacturers didn't go and say "We're going to use SI units from now on because we think it's more correct." They just quietly switched to 10^3 units and added a disclaimer. If this is not evidence of intent to deceive, I don't know what is. There are many good reasons why data storage is measured in powers of two, and in fact they have much to do with the reasons that SI units are based on powers of ten. In each context, those are the natural divisors.
And two-by-four, at least originally, had meaning. It was the size of the rough lumber before it was planed down to be (more or less) smooth. I have no idea if the machines and whatnot that are used have improved so that they don't have to shave off as much wood, but that is the original meaning. The same applies to two-by-sixes, two-by-eights, one-by-sixes, etc. ad nauseam.
I've seen this shit before, and (for me) it was a synchronization issue. Basically, mouse movements and clicks are reported as multiple bytes over the ps/2 port. This usually happens when you bump the mouse while switching machines, so half the movement packet goes to one pc, and the other half to the other pc. In X11, I had luck convincing it to re-synchronize by (without moving the mouse AT ALL!) clicking each button in turn once. This usually meant left, right, middle, up, down, though it may be different for you (e.g. left middle right, or down, up).
I think hot swap has more to do with the raid controller than the drives themselves. Of course, the physics of it have some to do with the enclosure as well, but push comes to shove, one can open up a running machine and unscrew a hard drive, it's just not as much fun as just releasing a latch and pulling on a lever.
I have one of the (lower end) 3ware ide-raid cards, and they claim to support hot swap. You have to use their admin tool to tell the controller to deactivate the drive, but supposedly you can do that and then unplug the ide drive, plug in a replacement, and tell the controller to activate the new drive.
I've had some problems with ext2/3 on raid systems and under load. The ext* filesystems seem to like buffering up as much as they can, and then suddenly deciding to flush everything to disk, blocking i/o processes across the whole system while they do. Reiser has been, and continues to be (for me at least), much better behaved about scheduling its writes so that it keeps the i/o load on the disks match to the i/o load from user space processes, isntead of batching it up and then having to wait while it flushes buffers.
Case in point: I was doing some high bitrate multimedia capture a while back. In my system at that time, I had both 10krpm u2w scsi disks, and some 5400rpm ide disks. In order to not get frame drops due to long flushes to disk with ext2/3, I had to jump through all sorts of hoops and basically have a process calling sync() every couple seconds, even when writing to the scsi disk. When I switched over to Reiser, I could take out the sync() calls and write to the slow arse ide disk with zero frame drops.
Also, formatting 2TB with ext3 took 24 hours, but only about 30 seconds with Reiser.
The only problem I've encountered with Reiser is that occaisonally, on a hard crash, some file that was being written to will end up with a whole bunch of garbage at the end of it, usually a block (4kb) or so of null bytes. I know from reading the Reiser mailing lists that I'm not the only one to have seen this. I've never seen it with ext3, but I haven't used ext3 as much, so YMMV.
Lets say Coca-Cola's secret recepie makes it out into some free, comunity maintained cookbook. This cookbook is freely copied and contributed to and modified by millions of people around the world. Numerous people have their own variants of this cookbook that they have added personal recipies to and distrbuted as such variants.
Now, someone finds that there is a bona fide violation of trade secret, that coca cola's recipie is in part of one of the cook book's many soda recipies (bear with the strained analogy here...). This is discovered and conceded to by a large publishing company that distributes nicely bound printed versions of this cook book.
SCO's is asserting that people must cease all use of Linux whatsoever, rather than revealing which pieces of Linux are in violation and allowing those to be fixed. This is analogous to Coca Cola asserting that, rather than revealing which recipies are in violation, people must throw away the whole cook book and never look at a single recipie inside it ever again.
I believe that there are some important legal principles that are relevant here, and one of them is the right to know the crime with which you are being accused. If SCO is going to attack any users of Linux, those users have a right to know what they did that was in violation. The cops aren't allowed to arrest you and throw you in jail saying only "you violated some traffic laws driving to and from work some time in the last few years." It is patently obvious that the vast majority of the linux code was not stolen from proprietary sources, so SCO cannot make any claim that the whole of linux is in violation and must be removed from existence.
It sure would be nice of some one ofthe people who signed that NDA and got to see the code would anonymously post enough information about what bits they were shown so that people could figure out what the code was.
There must be I think, but, again, I'd like to hear about them.
As another poster said, there are many causes for asthma. I myself have suffered from asthma, and I've personally (i.e. not really confirmed by any medical professional) observed at least two different forms of it.
One, which correlates very well with the Buteyko theory, is brought on by heavy exercise in cold weather. I've known and heard of many people who suffer from so called 'exercise induced asthma'. I seem to suffer from this a little, but it is only ever noticeable in cold (more or less below freezing) weather. Since I'm usually not in very good shape, the whole over-breathing thing makes a lot of sense to me, especially considering that the best way for me to have the exercised-induced asthma go away was to sit down and breathe slowly.
However, the other form of asthma from which I have suffered is rather different, and is generally triggered by allergic reactions, and exacerbated by having a cold at the time. When I was younger and the asthma & allergies were both worse, this could cause attacks of such severity that the *most* I could breathe still left me light headed! A theory that the asthma is invoked by overbreathing causing a lack of CO2 just doesn't hold water there.
And, for some more technical refutation: One of the types of inhalers I had was a Chromolyn-Sodium(sp?) one. This was not a theraputic, but rather a prophylactic. It worked by retarding the replication of mast cells, which are related to the immune system and the causes of other alergic reactions. These cells, IIRC, when activated, secrete a chemical that triggers the inflamation of the bronchial passages that is asthma.
So, in short, I believe that the Buteyko theory/therapy could help people with excercise-induced-type asthma, I have a hard time believing that it can do much for allergic-type asthma. For people with chronic asthma of the exercise type, the therapy could be a great help. For those who only get it when exercising, it isn't much good, since they are breathing so heavily because they need the O2 for the exertion! And for those with allergic-type, I can't see how it could do any good.
I was working on a program that did some video-conferencing stuff using OpenH323 this past spring for senior project, and I had one of the most bizzarre bugs I've ever encountered.
The program would mostly run fine, but now & then one of the threads would crash horribly. GDB was (almost) no help. I could see enough from the backtrace that the stack was being smashed, so I guessed that there was some null pointer deref going on (since there were no arrays that could overflow in this bit of code). However, single stepping through that thread almost never produced errors, and when it did, they were non-deterministic1
Finally, after much staring at code, I started to have a clue. I started single stepping through other threads, and finally found my problem.
Thread B was smashing the stack of Thread A (throug a null pointer deref as I had originally guessed). It did this in such a way that Thread B didn't get messed up at all, and Thread A could often continue running for some time. Furthermore, the way in which it smashed the stack let Thread A get several levels deep in bogus function calls!
'scuse me if I'm being dense and not getting the joke somewhere here, but the one time I ate baked alaska, it was delicious. So... what's wrong with actually making people happy?
Yes, I suspect that they are probably lying, or have defined PMPO to be, say, 10x actual peak power output, but I still have a bone to pick with your argument.
A simple application of Ohm's Law reveals that 9V into 8 ohms could yield a maximum current of (I = E/R) 1.125 amps.
If I remember my intro EE stuff correctly, current sources in parallel add current linearly. So, if I have 25 of the capacitors you describe in parallel[1], then I can discharge them in parallel and get 28.125 amps, and 253.125 watts. Of course, you couldn't sustain this for long, and if you did, you'd almost certainly burn up bits of the PCB. But you can get more than 1.125 amps out of some caps charged to 9V.
[1]: Yes, sufficiently ridiculous and space wasting that you'd not find it in cheapshit computer speakers, but I could solder it together in a few minutes and demonstrate it.
It's a *CAT*. If it forgot to come in before I went to bed, it can sleep on the porch. If it's got a rodent, it can sit and eat it on the porch no matter what time of day it is, because I can look outside and see if it's got one before I let it in.
And yes, cats are quite capable of learning to ask to be let in before you go to bed. In the course of my life, my family has had 3 cats, and even the dumbest male[1] of them could figure this out.
If stuff is coming in your pet door that you don't want, the solution is to not use the pet door.
[1]: Of course, he was the most fun to be around too:)
Slashdot Mirror
What I couldn't find was an explanation of what the "crypto" stuff in that 2D barcode is. I think I've seen something about some other similar systems where there's some mathematical voodoo that goes on that lets you use that data to tally a group of votes without being able to know the details of the individual votes.
Regardless of the details, I think that the data in that barcode has stuff that allows anyone to verify that a) it is an authentic ballot and b) given the full set of ballots, recompute the tally on their own and compare that to the published results.
Wish I could remember where I saw that or how it worked though.
The mechanism shown in the video actually does address this. The voting form layout is randomized, and the "key" portion (the bit matching checkboxes with candidates) is destroyed at the voting location. The receipt shows your marks, so you can verify that the marks counted match the marks you made, but it does not show what those marks mean. Even if someone forces you to give up your receipt, they have no way of knowing how you voted, only that the system recorded the vote correctly.
Yes, storing and providing full text search over a large pile of email consumes resources ... duuuh?
Also they're measuring the performance of Thunderbird while converting to the new system, not in its steady state. This is like complaining that Firefox uses a lot more CPU importing settings from IE than IE uses when looking at your home page.
Their claim as to how long it took to do the full text indexing of the mail seems dubious to me. I've got a similar amount of mail, and the time it took to index was more like minutes, not days.
The other thing to do if you want longish term reliability is to add redundancy to whatever you're storing with a tool like par2, http://www.par2.net/ and http://www.quickpar.org.uk/ are your friend.
Raid5 will help you if you lose a whole drive (e.g. siezes up from sitting still for a long time), the par2 data will both allow you to verify that the data hasn't been corrupted, and if it is (e.g. a couple sectors go bad), it will let you recover the data.
Jeez, I remember those. The home PC when I was a wee one used a bunch of their stuff for a Northstar Z80 based system that ran CP/M. Had Wordstar, a copy of Logo by the Lisp Corporation, I think my parents used an ancient version of DBase, and an early version of Microsoft Money. Somewhere in my place or theirs, I think there's even a poster with the elephant :)
+5 Nostalgia indeed!
It took me 3 weeks of fighting with Verizon (who insisted on taking 2 days to make a service appointment window, and insisted that they be 8a-7p) and my DSL provider (who was horribly frustrated by their inability to get Verizon to simply run a clean *bleeping* loop) to get things back up and running.
To do what you described with an observational study you would need two things that are unavailable:
Even in a very simple study, you could not enumerate all the possible causes because you do not know what might be causes, and you cannot find enough subjects because even with a fairly short list of causes, you end up with more combinations than there are people on the planet. Even with clever design so that you don't need a subject for every possible combination, you still won't find enough people on the planet to conduct your study, much less enough people going to the beach in Finland.
And all that is unnecessary anyways. The whole point of the controlled study described by Frequency Domain is that you vary single hypothesized causes, leaving all else the same, and look for changes in outcome. If you are only changing one variable, then the other factors will retain their distribution, unless they are causatively changed by the variable in question. This non-changing aspect is further reinforced by the nature of a double blind randomized study. In such a study, the patient population is randomly distributed among the possible groups (e.g. gets drug, gets placebo), with information escrowed in such a way that neither the patient nor the patient's doctor nor the people conducting the study know who fell into which group until the study is ended and the information retrieved from escrow.
So, with such a study setup, you can indeed rule out unrelated causes with a very high degree of confidence.
However, with an observational study in which you cannot tweak variables in a controlled fashion to measure outcomes, you are left with only ever being able to provide statistical confidences on correlations, never on causes.
If you know the positioning of each of the three dots, you can still do a certain amount of sub-pixel rendering, just not in quite the same way. Doing some more research, I think this was just on the Apple II, which may well have used a trinitron tube and thus used the same style of subpixel rendering as ClearType/CoolType/etc.
c.f. http://www.grc.com/ctwho.htm
OSX, CoolType? FAH! The Apple I/II used sub-pixel rendering to display things nicely on a tv screen.
Obviously, few (if any) business users need anything more than a Pentium III running at 500 MHz. That processor is perfectly acceptable for business applications like OpenOffice.
Obviously you haven't gone and looked at what gets installed on many business PCs. My employer's standard systems are 2.4ghz p4 on the desktop, 1.7gz p4m on the laptops, 512-1025m ram depending on system usage.
Everyone complains that they're slow. Why? Lets see:
Software distribution systems that check through 5000 packages every 2-4 hours to check for critical updates.
Office addins chiefly our document management system. Ripping all the addins out of Word cuts its startup time by a factor of 5 for me, approximately. However, I'm pretty rare in not needing any of those addins.
For a stripped down system running just the office apps with no addins and some basic virus scanning software, yes, old hardware does fine. But a business desktop does so much more than that, because a business needs to do more things to manage their computers. And the home user, who doesn't have all that, uses the processor power for games and goofing around with their pictures and home videos.
And never underestimate how many business users don't just sit there making office documents. There are large numbers that do development, visualization, number crunching, and other compute intensive tasks on a regular, if not continuous basis.
And whatever makes people think that these fancy web applications need less horsepower? JavaScript + DHTML is going to be less efficient for UI work than native code, thus requiring a faster processor.
I always figured it said that because the whole sequence of movies was meant to be a tale / documentary from the future about how things got to be the way they were. That is, the events of SW are in the future for us, but in the far past from the point of view of the storyteller.
Learing how your system works is great. But I already know. Most of the time when I want to install linux on a new system, I just want to get it installed. Not having excess packages is great, and control over the install is great, but extra typing and waiting is not. In that vein, RedHat's installer is tolerable, but not optimal in that it's hard to get as minimal an install as I'd like. Debian's installer is pretty good, though it requires doing a bit more manual work than I'd like. Anything more manual than Debian's installer is just too much for me.
When it comes to Linux newbies, install time isn't the time to learn about the system. You want to learn about the system when you have all the emotionally-satisfying and high usability eye candy and applications installed. If people have to read lots of documentation in order to install the system, they are likely to get frustrated and possibly give up. Of course, there's more motivation to learn how the system works when you need to do so in order to install it. On the other hand, if the goal is world domination, you don't want to limit it to only the die hard techies.
And as much as I'm a command line junkie, I don't think joe user should be required to to be one too in order to do the basic stuff with his computer. Expecting joe admin to be savvy about the commandline is well and good, but not joe user.
In my experience, on any filesystem over a few gigs in size, ReiserFS's overhead is actually smaller than ext2/3. I converted a 240-ish gig partition from ext3 to reiserfs and regained something on the order of 15-20 gigs as I recall. This wasn't due to tail packing, either, as most of the files on that partition were many many megs in size.
Go get a goddamn calculator, as your brain is obviously incapable of performing arithmetic:
1000/1024 ~= 0.977
So a 10k thing in hard drive k will really only be 9.77k in computer k. That's called overstating (or overestimating to use your poor vocabulary -- it's stating sizes, not estimating them!) not understating. This gross lack of brain-usage on your part seems to lead to your subsequent thrashing about for bullshit justifications for the manufacturers.
Partition tables and boot sectors combined take less than a few k. Hardly noticeable on a freaking floppy disk, let alone an 80G hard drive.
And computer users don't sue about the OS using RAM because the RAM is there, it's just being used! They're suing the HD manufacturers because the space just isn't there. Ditto with your filesystem overhead remark.
As far as bad sectors are concerned, the days of the OS having to deal with bad sectors are almost gone. All modern HDs have space space inside that they use to transparently deal with sectors going bad.
Why is everyone so bound and determined to justify the hard drive manufacturers in this? Regardless of the technical correctness, the intent of the manufacturers was to bullshit consumers pure and simple. That is illegal and punishable. And as far as correctness is concerned, some standards body saying "you should do XYZ" is far from XYZ becoming an accepted and established standard. Standards have to be adopted in order to be meaningful and valid!
Yes, but it's a two-by-four, not a two inch by four inch piece of wood. Even a high-school-non-graduate working at a construction site doesn't tend to have problems with this.
Most of the time standards bodies set down standards that reflect the existing reality, and if they change it at all, it is a minimal change to allow a precise and unambiguous definition (e.g. converting to atomic wavelengths instead of the length of a particular hunk of metal). For them to declare a standard that is blatantly at odds with the state of the industry is, frankly bullshit.
And when it comes to deceptive advertising, if I'm not mistaken, intent plays a role. A company that, through an honest mistake (typo, whatever), mis-advertises the price/size of an item for sale, they do not generally get held liable. When a company deliberately miscontrues information or represents it in a deliberately confusing manner, that's a different case.
The HD manufacturers didn't go and say "We're going to use SI units from now on because we think it's more correct." They just quietly switched to 10^3 units and added a disclaimer. If this is not evidence of intent to deceive, I don't know what is. There are many good reasons why data storage is measured in powers of two, and in fact they have much to do with the reasons that SI units are based on powers of ten. In each context, those are the natural divisors.
And two-by-four, at least originally, had meaning. It was the size of the rough lumber before it was planed down to be (more or less) smooth. I have no idea if the machines and whatnot that are used have improved so that they don't have to shave off as much wood, but that is the original meaning. The same applies to two-by-sixes, two-by-eights, one-by-sixes, etc. ad nauseam.
For all the linux geeks out there, try motion. I used to use it for some mundane things, and I can testify that it works very nicely :)
I've seen this shit before, and (for me) it was a synchronization issue. Basically, mouse movements and clicks are reported as multiple bytes over the ps/2 port. This usually happens when you bump the mouse while switching machines, so half the movement packet goes to one pc, and the other half to the other pc. In X11, I had luck convincing it to re-synchronize by (without moving the mouse AT ALL!) clicking each button in turn once. This usually meant left, right, middle, up, down, though it may be different for you (e.g. left middle right, or down, up).
I think hot swap has more to do with the raid controller than the drives themselves. Of course, the physics of it have some to do with the enclosure as well, but push comes to shove, one can open up a running machine and unscrew a hard drive, it's just not as much fun as just releasing a latch and pulling on a lever.
I have one of the (lower end) 3ware ide-raid cards, and they claim to support hot swap. You have to use their admin tool to tell the controller to deactivate the drive, but supposedly you can do that and then unplug the ide drive, plug in a replacement, and tell the controller to activate the new drive.
I've had some problems with ext2/3 on raid systems and under load. The ext* filesystems seem to like buffering up as much as they can, and then suddenly deciding to flush everything to disk, blocking i/o processes across the whole system while they do. Reiser has been, and continues to be (for me at least), much better behaved about scheduling its writes so that it keeps the i/o load on the disks match to the i/o load from user space processes, isntead of batching it up and then having to wait while it flushes buffers.
Case in point: I was doing some high bitrate multimedia capture a while back. In my system at that time, I had both 10krpm u2w scsi disks, and some 5400rpm ide disks. In order to not get frame drops due to long flushes to disk with ext2/3, I had to jump through all sorts of hoops and basically have a process calling sync() every couple seconds, even when writing to the scsi disk. When I switched over to Reiser, I could take out the sync() calls and write to the slow arse ide disk with zero frame drops.
Also, formatting 2TB with ext3 took 24 hours, but only about 30 seconds with Reiser.
The only problem I've encountered with Reiser is that occaisonally, on a hard crash, some file that was being written to will end up with a whole bunch of garbage at the end of it, usually a block (4kb) or so of null bytes. I know from reading the Reiser mailing lists that I'm not the only one to have seen this. I've never seen it with ext3, but I haven't used ext3 as much, so YMMV.
Lets say Coca-Cola's secret recepie makes it out into some free, comunity maintained cookbook. This cookbook is freely copied and contributed to and modified by millions of people around the world. Numerous people have their own variants of this cookbook that they have added personal recipies to and distrbuted as such variants.
...). This is discovered and conceded to by a large publishing company that distributes nicely bound printed versions of this cook book.
Now, someone finds that there is a bona fide violation of trade secret, that coca cola's recipie is in part of one of the cook book's many soda recipies (bear with the strained analogy here
SCO's is asserting that people must cease all use of Linux whatsoever, rather than revealing which pieces of Linux are in violation and allowing those to be fixed. This is analogous to Coca Cola asserting that, rather than revealing which recipies are in violation, people must throw away the whole cook book and never look at a single recipie inside it ever again.
I believe that there are some important legal principles that are relevant here, and one of them is the right to know the crime with which you are being accused. If SCO is going to attack any users of Linux, those users have a right to know what they did that was in violation. The cops aren't allowed to arrest you and throw you in jail saying only "you violated some traffic laws driving to and from work some time in the last few years." It is patently obvious that the vast majority of the linux code was not stolen from proprietary sources, so SCO cannot make any claim that the whole of linux is in violation and must be removed from existence.
It sure would be nice of some one ofthe people who signed that NDA and got to see the code would anonymously post enough information about what bits they were shown so that people could figure out what the code was.
There must be I think, but, again, I'd like to hear about them.
As another poster said, there are many causes for asthma. I myself have suffered from asthma, and I've personally (i.e. not really confirmed by any medical professional) observed at least two different forms of it.
One, which correlates very well with the Buteyko theory, is brought on by heavy exercise in cold weather. I've known and heard of many people who suffer from so called 'exercise induced asthma'. I seem to suffer from this a little, but it is only ever noticeable in cold (more or less below freezing) weather. Since I'm usually not in very good shape, the whole over-breathing thing makes a lot of sense to me, especially considering that the best way for me to have the exercised-induced asthma go away was to sit down and breathe slowly.
However, the other form of asthma from which I have suffered is rather different, and is generally triggered by allergic reactions, and exacerbated by having a cold at the time. When I was younger and the asthma & allergies were both worse, this could cause attacks of such severity that the *most* I could breathe still left me light headed! A theory that the asthma is invoked by overbreathing causing a lack of CO2 just doesn't hold water there.
And, for some more technical refutation: One of the types of inhalers I had was a Chromolyn-Sodium(sp?) one. This was not a theraputic, but rather a prophylactic. It worked by retarding the replication of mast cells, which are related to the immune system and the causes of other alergic reactions. These cells, IIRC, when activated, secrete a chemical that triggers the inflamation of the bronchial passages that is asthma.
So, in short, I believe that the Buteyko theory/therapy could help people with excercise-induced-type asthma, I have a hard time believing that it can do much for allergic-type asthma. For people with chronic asthma of the exercise type, the therapy could be a great help. For those who only get it when exercising, it isn't much good, since they are breathing so heavily because they need the O2 for the exertion! And for those with allergic-type, I can't see how it could do any good.
I was working on a program that did some video-conferencing stuff using OpenH323 this past spring for senior project, and I had one of the most bizzarre bugs I've ever encountered.
The program would mostly run fine, but now & then one of the threads would crash horribly. GDB was (almost) no help. I could see enough from the backtrace that the stack was being smashed, so I guessed that there was some null pointer deref going on (since there were no arrays that could overflow in this bit of code). However, single stepping through that thread almost never produced errors, and when it did, they were non-deterministic1
Finally, after much staring at code, I started to have a clue. I started single stepping through other threads, and finally found my problem.
Thread B was smashing the stack of Thread A (throug a null pointer deref as I had originally guessed). It did this in such a way that Thread B didn't get messed up at all, and Thread A could often continue running for some time. Furthermore, the way in which it smashed the stack let Thread A get several levels deep in bogus function calls!
'scuse me if I'm being dense and not getting the joke somewhere here, but the one time I ate baked alaska, it was delicious. So ... what's wrong with actually making people happy?
Yes, I suspect that they are probably lying, or have defined PMPO to be, say, 10x actual peak power output, but I still have a bone to pick with your argument.
A simple application of Ohm's Law reveals that 9V into 8 ohms could yield a maximum current of (I = E/R) 1.125 amps.
If I remember my intro EE stuff correctly, current sources in parallel add current linearly. So, if I have 25 of the capacitors you describe in parallel[1], then I can discharge them in parallel and get 28.125 amps, and 253.125 watts. Of course, you couldn't sustain this for long, and if you did, you'd almost certainly burn up bits of the PCB. But you can get more than 1.125 amps out of some caps charged to 9V.
[1]: Yes, sufficiently ridiculous and space wasting that you'd not find it in cheapshit computer speakers, but I could solder it together in a few minutes and demonstrate it.
It's a *CAT*. If it forgot to come in before I went to bed, it can sleep on the porch. If it's got a rodent, it can sit and eat it on the porch no matter what time of day it is, because I can look outside and see if it's got one before I let it in.
:)
And yes, cats are quite capable of learning to ask to be let in before you go to bed. In the course of my life, my family has had 3 cats, and even the dumbest male[1] of them could figure this out.
If stuff is coming in your pet door that you don't want, the solution is to not use the pet door.
[1]: Of course, he was the most fun to be around too