There are no (NO!) commands so complicated that are too complex not to be scripted.
Nonsense. I challenge anyone in the world to write a program- not just a script, but a program runnable on their choice of any computer in the world- that will do the following:
Download pictures from a digital camera.
Identify the subject matter of each picture (i.e. recognize the people if it's a portrait, the name of a building if it's an architectural picture, etc.)
Decide on appropriate categories for different groups of pictures, and
Sort the pictures into different directories and subdirectories according to their categories.
The first and last step of that process is certainly scriptable- I use scripts to creat a browsable catalog of my photographs after I've sorted them- but the middle two steps are not. The problem is AI complete. Since AI good enough to look at a photograph and identify the subject is not available, the task requires human interaction to complete it. Scripts are very valuable but they aren't useful for non-standardized tasks requiring judgment.
Pointy flashy clicky things just distract from getting real work done.
Familiarity breeds contempt. I'm guessing that you used a "pointy flashy clicky" web browser to post that message, but given that you're reading Slashdot I can accept the point that it's distracting you from getting real work done.
More seriously, though, there are plenty of tasks that are complicated enough that they can't be scripted. My guess is that you didn't rely on a script to post your comment to/., and you probably don't rely on one when you compose email, draw pictures, play music, or organize your pr0n collection. Programs that let you do those kinds of things inherently require a more or less complicated UI, and that means that somebody needs to design that UI, hopefully so that it's easy to use. You can't ignore the effort that went into designing those programs' UIs just because you're so familiar with them that you've stoped noticing. To the contrary- the ability to use those programs so easily that you forget about their UI is evidence that it's well designed.
Hmm. Perhaps you need to go back and look at this article for some pointers. The most popular suggestion seems to be waiting until you get so sick that you can't drink coffee and don't notice how miserable the withdrawl symptoms are.
The article gives a LD50 (lethal dose in 50% of cases) of 192 mg/kg in rats. About.com lists a range of 60-120 mg/cup for brewed coffee, so that would give a range of 6-12 g in 100 cups. That would be over the listed LD50 for a light person drinking strong coffee. Of course, that doesn't account for the possibility of tolerance. Somebody who regularly drinks a pot or two of strong coffee per day will have some physiological tolerance for caffeine and will likely have a higher lethal dose than somebody (or some rat) that wasn't a regular user.
Re:The Long Answer
on
Death by Coffee?
·
· Score: 5, Informative
No, coffee isn't dehydrating because it contains too many electrolytes. It's dehydrating because caffeine is a diuretic, i.e. a drug that induces urination. In any case, though, the comment about serving with a glass of water points out something important- that you can theoretically overcome issues with overhydration/dehydration/mineral depletion/etc. by drinking something else or taking electrolytes at over the same time period that you're taking the coffee, negating its negative effects (other than the potential caffeine toxicity).
No. Fedora is trying very hard to avoid IP issues, so they've deliberately refrained from including things like mp3 decoders and DVD decoders that might get them into legal trouble. Fortunately, Fedora does have apt and yum available, so it's easy to add external repositories, like FreshRPMS or Livna, both of which do include mp3 players and DVD decoders. It's very convenient, and avoids a lot of legal headaches for RedHat.
And, as somebody else has pointed out, there's nothing to prevent you from running both a hardware firewall and a software firewall. In the unlikely event that there's an unpatched vulnerability in the hardware firewall, you'd have the backup protection of the software firewall to protect you. The chance of both of them being vulnerable simultaneously and there being a worm written specifically to take advantage of the situation seems to be vanishingly small.
A driver is responsible for the upkeep of his vehicle if his negligence causes an accident
The analogy breaks down, though, because the problem isn't with user failing to maintain his product, but with the product containing a manufacturing defect. Patching buggy software is the computer equivalent of taking a car in for a recall. Punishing computer users when their computers get infected is like punishing drivers when they get into accidents caused by failure of recalled parts. There has to be some kind of grace period during which the creator is considered at fault for making a defective product, rather than the user for not having it fixed.
The highest packet rate they saw was more than 23,000 per hour, sustained for at least one hour.
Perhaps equally scary is that the worm seems to have saturated its host population in under an hour. Since infection rate is slower in a small population like this one, a worm infecting via an exploit in a popular program could propagate even faster. If a worm writer were to discover and exploit a previously unknown vulnerability in a very widely deployed program, the consequences could be ghastly.
The people handing out the domains in your whack-a-mole ( good imagery, btw ), example will hand them out to spammers nilly-willy, but for.mail, they will not.
It's a different group of people. The ordinary.com/.net etc. registrars are purely for profit organizations. Anyone in the world can go to them with the name of an available domain and some money and get the domain they're asking for. The people running.mail are not the same group. It's being proposed by a group of anti-spam activists who have been devoting a huge amount of time and energy to wiping out spam. There's no particular reason to believe that they will give domains to any known spammer, especially because these are the same people who are doing the best job of tracking down the spammers and all their known aliases.
But this proposal is quite different from SPF. Under SPF, anyone with a domain is allowed to define which computers are valid mail senders for that domain, but there's no further restriction. That would prevent spammers (and email worms) from falsifying their sender address, but it doesn't directly confront the issue of spam. A spammer with his own domain, presumably hosted by a spam-friendly service provider, can still define his own computers as being permitted senders for that domain and send out spam. He'll presumably be stopped once people recognize the domain and start blocking mail from it, but that just makes it a matter of playing whack-a-mole; the spammer just buys new domains in bulk from a cheap registrar and switches every time people start blocking the old one.
What.mail does is different. It defines a known, and defended, whitelist domain. Mail from a.mail address should be safe, because the registrar actually takes steps to make sure that spammers aren't allowed to register there. One part of the proposal that I haven't seen mentioned here is that all mail sent to abuse@somedomain.mail is directed to the.mail registrar, rather than the domain owner. That means that spam complaints will be sent to a third party with the power to revoke the domain if the complaint is valid. Obviously what would be really good would be to combine the two proposals, so that somebody couldn't forge mail from a.mail server, but they do address different points.
I think that you're misreading what I wrote. The point is that there are two ways of obeying the CAN-SPAM act:
Putting a legitimate address in the mail, having and opt-out, etc.
Refusing to spam.
My point is that the original article seems to say that neither group 1 (spammers who follow the rules) nor group 2 (non-spammers) would be allowed to register under.mail. This would obviously be stupid, and isn't what SpamHaus is saying.
.' The interesting twist is that companies that comply with the US CAN-SPAM act - which SpamHaus opposed due to the legalization of bulk unsolicited commercial e-mail - would not be eligibile to register a.mail address.
That's not quite correct. The SpamHaus rules wouldn't ban anyone who obeyed the CAN-SPAM act. Presumably most ordinary companies obey CAN-SPAM by refusing to do anything that vaguely resembles spamming, and they'd be just fine under the SpamHaus rules. What SpamHaus wants to do is to use a stricter definition of what constitutes spam, so that some senders who meet the terms of CAN-SPAM still wouldn't qualify.
That's the wrong attitude to take. If a Linux-based server is compromised because of software flaws, that's a perfectly legitimate point in an argument about security, just as the compromise of a Windows-based server because of a software flaw would be. If there's a real vulnerability that let somebody crack the system (as opposed to a misconfiguration or incorrect belief that the system was broken into) it needs to be fixed pronto, rather than written off as a PR event.
The flipside is that you can have problems when the same donor wants to give multiple buildings. My alma mater has no fewer than 4 buildings donated by and named after Arnold Beckman scattered across campus. It gets annoying when you have to call them "The Beckman Institute", "Beckman Auditorium", "Beckman Behavioral Biology", or whatever rather than just "Beckman". Not to mention how confusing it gets when you have multiple donors with similar names, like Spaulding and Spalding.
There's more than one legitimate meaning of "take". From dictionary.com:
1. To get into one's possession by force, skill, or artifice...
10. To accept and place under one's care or keeping.
They're not taking the money in sense 1, but they certainly are taking it in sense 10. I hope that you will admit that there is irony in accepting money from a rich donor and then using it in a way that may serve to undermine the source of that donor's wealth. Whether you view that as pleasant or unpleasant irony obviously depends on your view of Mr. Gates.
You haven't been paying very careful attention to University naming practices, have you? Most universities will name a building whatever the donor who gives it to them says to name it. If Bill Gates wants a building name after himself, his mother, or his favorite pet goldfish from when he was six, any school in the country will oblige him as long as he's writing the check. Besides, you could easily argue that there's a certain pleasant irony in taking a big chunk of money from Mr. Gates and using it to build a facility where the researchers will be doing work that will benefit Free Software.
Canon.RAW images fit this description as the image has had no post-processing on it. It's exactly as it comes off the CCD.
Extremely minor nit: Canon digital SLRs use CMOS sensors, not CCDs. Until quite recently everyone thought that CMOS was too noisy to generate really first rate pictures, but the Canons are very competitive with CCD-based SLRs. CMOS also has the advantage of costing only about 1/3 as much as a comprably sized CCD. My impression is that this is why all of the existing digital SLRs that have full frame (i.e. 24mm x 36mm) sensors use CMOS, rather than CCD; it's not economically feasible to use a CCD that big.
The cameras are quite expensive. The cheapest camera mentioned, the EOS-10D, runs about $1500 retail. The EOS-1D, which they mention as the main camera that SI uses, is about $4500. (That's actually for the updated Mark II, but AFAIK the original 1D was about the same price.) The EOS-1Ds is $8000. The Nikon D-1X and D-2H are about $4000 and $3000, respectively. And the lenses can make the cameras seem cheap in comparison. The article mentioned 600mm F/4 ($7200), 400mm F/2.8 ($6500), 300mm F/2.8 ($3900), and 70-200mm F/2.8 ($1600) lenses. A professional sports photographer may very well have $50,000 in cameras and lenses.
As another poster points out, most manufacturers' RAW formats do use lossless compression, so the data is being squeezed at least a bit. More importantly, though, PNG won't do the job. Good quality digital cameras- basically anything in the digital SLR class- have more color depth than the common image formats can handle. PNG and JPEG use 24 bit color(plus 8 bits of alpha for PNG), while the cameras can produce 36 bit color. You can't display all that color depth on a monitor, which is why PNG and JPEG don't bother keeping it, but it is useful when manipulating the pictures in Photoshop/Cinepaint and when printing to paper. IIRC, most camera companies export their RAW formats to 48 bit TIFFs for processing in common image manipulation programs.
Only because Amazon is far too literal. If you search for The Adolescence of P1 on Amazon, you get all of that drek at the top of your search, but if you search for The Adolescence of P-1 (which is the correct spelling of the title) the right book is the top match. Google also give the correct page on Amazon when given the correct spelling, and it manages to get it in the top 10 when given the incorrectly spelled version. Given that Google is searching the whole web and not just Amazon, I'd say that Google wins that one handily.
Actually, one side benefit of insurance is that it tends to force people to adopt good standards. Insurance companies understand the concepts of "adverse selection" (people who are bad risks are more likely to want insurance) and "moral hazard" (people who are insured are more likely to be sloppy and/or dishonest) and take steps to minimize these problems. One big thing that they do is to force their customers- or at least provide them with strong incentives- to adopt good practices that reduce the rate at which payouts are necessary.
As an example, before selling a business theft insurance, the insurance company is likely to perform an audit of their security. The company might reduce the rate if the business wanting insurance installs a high quality alarm system or refuse to insure them at all unless they hire guards from a reputable security company. For fire insurance, they might require good alarms, sprinkler systems, and employee training. The same thing will probably be true of things like programming malpractice insurance. Insurance companies will give much better rates to people who have certification proving that they've passed professional training in writing secure programs, have adequate pre-release testing, etc.
Presumably not. The OED is looking for the first instances of words in English; it is, after all, the Oxford English Dictionary. Verne wrote in French, so his books wouldn't be a good source of the first appearances of words in English.
They're not able to scope out the course in advance, since they were only told what it was two hours before the race started. This is a reasonable requirement for a military test, since soldiers generally don't get a chance to preview their driving path before taking it.
Your assumption that they're crossing a barren plain is incorrect. The Mojave desert is not an easy place to drive. Quite the contrary; it's an area that dedicated off-roaders love because of the challenge of driving there. DARPA chose a test that they expected none of the entrants to be able to beat; my impression is that even making it 7 miles is an enormous accomplishment.
Slashdot Mirror
Nonsense. I challenge anyone in the world to write a program- not just a script, but a program runnable on their choice of any computer in the world- that will do the following:
The first and last step of that process is certainly scriptable- I use scripts to creat a browsable catalog of my photographs after I've sorted them- but the middle two steps are not. The problem is AI complete. Since AI good enough to look at a photograph and identify the subject is not available, the task requires human interaction to complete it. Scripts are very valuable but they aren't useful for non-standardized tasks requiring judgment.
Familiarity breeds contempt. I'm guessing that you used a "pointy flashy clicky" web browser to post that message, but given that you're reading Slashdot I can accept the point that it's distracting you from getting real work done.
More seriously, though, there are plenty of tasks that are complicated enough that they can't be scripted. My guess is that you didn't rely on a script to post your comment to /., and you probably don't rely on one when you compose email, draw pictures, play music, or organize your pr0n collection. Programs that let you do those kinds of things inherently require a more or less complicated UI, and that means that somebody needs to design that UI, hopefully so that it's easy to use. You can't ignore the effort that went into designing those programs' UIs just because you're so familiar with them that you've stoped noticing. To the contrary- the ability to use those programs so easily that you forget about their UI is evidence that it's well designed.
Hmm. Perhaps you need to go back and look at this article for some pointers. The most popular suggestion seems to be waiting until you get so sick that you can't drink coffee and don't notice how miserable the withdrawl symptoms are.
The article gives a LD50 (lethal dose in 50% of cases) of 192 mg/kg in rats. About.com lists a range of 60-120 mg/cup for brewed coffee, so that would give a range of 6-12 g in 100 cups. That would be over the listed LD50 for a light person drinking strong coffee. Of course, that doesn't account for the possibility of tolerance. Somebody who regularly drinks a pot or two of strong coffee per day will have some physiological tolerance for caffeine and will likely have a higher lethal dose than somebody (or some rat) that wasn't a regular user.
No, coffee isn't dehydrating because it contains too many electrolytes. It's dehydrating because caffeine is a diuretic, i.e. a drug that induces urination. In any case, though, the comment about serving with a glass of water points out something important- that you can theoretically overcome issues with overhydration/dehydration/mineral depletion/etc. by drinking something else or taking electrolytes at over the same time period that you're taking the coffee, negating its negative effects (other than the potential caffeine toxicity).
No. Fedora is trying very hard to avoid IP issues, so they've deliberately refrained from including things like mp3 decoders and DVD decoders that might get them into legal trouble. Fortunately, Fedora does have apt and yum available, so it's easy to add external repositories, like FreshRPMS or Livna, both of which do include mp3 players and DVD decoders. It's very convenient, and avoids a lot of legal headaches for RedHat.
And, as somebody else has pointed out, there's nothing to prevent you from running both a hardware firewall and a software firewall. In the unlikely event that there's an unpatched vulnerability in the hardware firewall, you'd have the backup protection of the software firewall to protect you. The chance of both of them being vulnerable simultaneously and there being a worm written specifically to take advantage of the situation seems to be vanishingly small.
The analogy breaks down, though, because the problem isn't with user failing to maintain his product, but with the product containing a manufacturing defect. Patching buggy software is the computer equivalent of taking a car in for a recall. Punishing computer users when their computers get infected is like punishing drivers when they get into accidents caused by failure of recalled parts. There has to be some kind of grace period during which the creator is considered at fault for making a defective product, rather than the user for not having it fixed.
Perhaps equally scary is that the worm seems to have saturated its host population in under an hour. Since infection rate is slower in a small population like this one, a worm infecting via an exploit in a popular program could propagate even faster. If a worm writer were to discover and exploit a previously unknown vulnerability in a very widely deployed program, the consequences could be ghastly.
It's a different group of people. The ordinary .com/.net etc. registrars are purely for profit organizations. Anyone in the world can go to them with the name of an available domain and some money and get the domain they're asking for. The people running .mail are not the same group. It's being proposed by a group of anti-spam activists who have been devoting a huge amount of time and energy to wiping out spam. There's no particular reason to believe that they will give domains to any known spammer, especially because these are the same people who are doing the best job of tracking down the spammers and all their known aliases.
But this proposal is quite different from SPF. Under SPF, anyone with a domain is allowed to define which computers are valid mail senders for that domain, but there's no further restriction. That would prevent spammers (and email worms) from falsifying their sender address, but it doesn't directly confront the issue of spam. A spammer with his own domain, presumably hosted by a spam-friendly service provider, can still define his own computers as being permitted senders for that domain and send out spam. He'll presumably be stopped once people recognize the domain and start blocking mail from it, but that just makes it a matter of playing whack-a-mole; the spammer just buys new domains in bulk from a cheap registrar and switches every time people start blocking the old one.
What .mail does is different. It defines a known, and defended, whitelist domain. Mail from a .mail address should be safe, because the registrar actually takes steps to make sure that spammers aren't allowed to register there. One part of the proposal that I haven't seen mentioned here is that all mail sent to abuse@somedomain.mail is directed to the .mail registrar, rather than the domain owner. That means that spam complaints will be sent to a third party with the power to revoke the domain if the complaint is valid. Obviously what would be really good would be to combine the two proposals, so that somebody couldn't forge mail from a .mail server, but they do address different points.
I think that you're misreading what I wrote. The point is that there are two ways of obeying the CAN-SPAM act:
My point is that the original article seems to say that neither group 1 (spammers who follow the rules) nor group 2 (non-spammers) would be allowed to register under .mail. This would obviously be stupid, and isn't what SpamHaus is saying.
That's not quite correct. The SpamHaus rules wouldn't ban anyone who obeyed the CAN-SPAM act. Presumably most ordinary companies obey CAN-SPAM by refusing to do anything that vaguely resembles spamming, and they'd be just fine under the SpamHaus rules. What SpamHaus wants to do is to use a stricter definition of what constitutes spam, so that some senders who meet the terms of CAN-SPAM still wouldn't qualify.
That's the wrong attitude to take. If a Linux-based server is compromised because of software flaws, that's a perfectly legitimate point in an argument about security, just as the compromise of a Windows-based server because of a software flaw would be. If there's a real vulnerability that let somebody crack the system (as opposed to a misconfiguration or incorrect belief that the system was broken into) it needs to be fixed pronto, rather than written off as a PR event.
The flipside is that you can have problems when the same donor wants to give multiple buildings. My alma mater has no fewer than 4 buildings donated by and named after Arnold Beckman scattered across campus. It gets annoying when you have to call them "The Beckman Institute", "Beckman Auditorium", "Beckman Behavioral Biology", or whatever rather than just "Beckman". Not to mention how confusing it gets when you have multiple donors with similar names, like Spaulding and Spalding.
There's more than one legitimate meaning of "take". From dictionary.com:
They're not taking the money in sense 1, but they certainly are taking it in sense 10. I hope that you will admit that there is irony in accepting money from a rich donor and then using it in a way that may serve to undermine the source of that donor's wealth. Whether you view that as pleasant or unpleasant irony obviously depends on your view of Mr. Gates.
You haven't been paying very careful attention to University naming practices, have you? Most universities will name a building whatever the donor who gives it to them says to name it. If Bill Gates wants a building name after himself, his mother, or his favorite pet goldfish from when he was six, any school in the country will oblige him as long as he's writing the check. Besides, you could easily argue that there's a certain pleasant irony in taking a big chunk of money from Mr. Gates and using it to build a facility where the researchers will be doing work that will benefit Free Software.
Extremely minor nit: Canon digital SLRs use CMOS sensors, not CCDs. Until quite recently everyone thought that CMOS was too noisy to generate really first rate pictures, but the Canons are very competitive with CCD-based SLRs. CMOS also has the advantage of costing only about 1/3 as much as a comprably sized CCD. My impression is that this is why all of the existing digital SLRs that have full frame (i.e. 24mm x 36mm) sensors use CMOS, rather than CCD; it's not economically feasible to use a CCD that big.
The cameras are quite expensive. The cheapest camera mentioned, the EOS-10D, runs about $1500 retail. The EOS-1D, which they mention as the main camera that SI uses, is about $4500. (That's actually for the updated Mark II, but AFAIK the original 1D was about the same price.) The EOS-1Ds is $8000. The Nikon D-1X and D-2H are about $4000 and $3000, respectively. And the lenses can make the cameras seem cheap in comparison. The article mentioned 600mm F/4 ($7200), 400mm F/2.8 ($6500), 300mm F/2.8 ($3900), and 70-200mm F/2.8 ($1600) lenses. A professional sports photographer may very well have $50,000 in cameras and lenses.
As another poster points out, most manufacturers' RAW formats do use lossless compression, so the data is being squeezed at least a bit. More importantly, though, PNG won't do the job. Good quality digital cameras- basically anything in the digital SLR class- have more color depth than the common image formats can handle. PNG and JPEG use 24 bit color(plus 8 bits of alpha for PNG), while the cameras can produce 36 bit color. You can't display all that color depth on a monitor, which is why PNG and JPEG don't bother keeping it, but it is useful when manipulating the pictures in Photoshop/Cinepaint and when printing to paper. IIRC, most camera companies export their RAW formats to 48 bit TIFFs for processing in common image manipulation programs.
Only because Amazon is far too literal. If you search for The Adolescence of P1 on Amazon, you get all of that drek at the top of your search, but if you search for The Adolescence of P-1 (which is the correct spelling of the title) the right book is the top match. Google also give the correct page on Amazon when given the correct spelling, and it manages to get it in the top 10 when given the incorrectly spelled version. Given that Google is searching the whole web and not just Amazon, I'd say that Google wins that one handily.
Actually, one side benefit of insurance is that it tends to force people to adopt good standards. Insurance companies understand the concepts of "adverse selection" (people who are bad risks are more likely to want insurance) and "moral hazard" (people who are insured are more likely to be sloppy and/or dishonest) and take steps to minimize these problems. One big thing that they do is to force their customers- or at least provide them with strong incentives- to adopt good practices that reduce the rate at which payouts are necessary.
As an example, before selling a business theft insurance, the insurance company is likely to perform an audit of their security. The company might reduce the rate if the business wanting insurance installs a high quality alarm system or refuse to insure them at all unless they hire guards from a reputable security company. For fire insurance, they might require good alarms, sprinkler systems, and employee training. The same thing will probably be true of things like programming malpractice insurance. Insurance companies will give much better rates to people who have certification proving that they've passed professional training in writing secure programs, have adequate pre-release testing, etc.
Presumably not. The OED is looking for the first instances of words in English; it is, after all, the Oxford English Dictionary. Verne wrote in French, so his books wouldn't be a good source of the first appearances of words in English.
They're not able to scope out the course in advance, since they were only told what it was two hours before the race started. This is a reasonable requirement for a military test, since soldiers generally don't get a chance to preview their driving path before taking it.
Your assumption that they're crossing a barren plain is incorrect. The Mojave desert is not an easy place to drive. Quite the contrary; it's an area that dedicated off-roaders love because of the challenge of driving there. DARPA chose a test that they expected none of the entrants to be able to beat; my impression is that even making it 7 miles is an enormous accomplishment.