Yup, and it works out great until they realize that the same stack leak causing the crash on the 17th save also causes data to get silently lost in some other scenario.
When writing large programs it is critical to validate your inputs and carefully check stuff like this. Just because the way the error was caught was unlikely to come up in real life, doesn't mean that the same bug won't blow up in some other way. The only exception is if you actually ANALYZE the problem and determine that it isn't a big risk. But if you've done that then you've done half the work needed to fix it (unless you're in a shop that writes 10 pounds of documentation for one line of code change).
I knew somebody with a Palm-based smartphone on verizon quite a few years back, and its bluetooth wouldn't work with his 3-series BMW. Apparently verizon crippled the address book feature. That is typical verizon. It will be nice if they turn over a new leaf - I doubt they'd be doing it if they weren't getting killed in the smartphone arena.
Note that no US cell network has a 3G capability that works outside of the US, at least as far as I understand it. Granted, with the alphabet soup mess who can blame me if I am missing something?
That said, just about everybody but Verizon and Sprint have switched to GSM, which is of course the standard everywhere else (except Japan, apparently).
What a mess... You'll never see it fixed since the providers have too much at stake to allow their pet standard to lose. GSM is the obvious choice, but to somehow mandate a switch would probably put verizon at a substantial disadvantage, and that reduction in competition won't help consumers on the other networks either.
I still don't understand why data isn't data. If I pay for data transfer on their net why does it matter if that comes from a laptop connected to the phone (an extra charge) or from an app running on the phone itself? Are the bits fatter?
The problem is the word "unlimited". This word should probably be banned, except when it literally means use without any limits at all.
When verizon sells you unlimited service, they can do so because they know that a typical phone-based use won't use more than a certain amount of bandwidth. On the other hand, if you tether the thing to a laptop you can fire up bittorrent or download a high-def movie. You wouldn't do that on a phone since the phone probably wouldn't have the storage for something like that, apps capable of doing it, and why download a 10GB blu-ray image to watch it on a VGA-sized display?
If everybody just charged reasonable rates per byte then all that nonsense could go away. Just make people pay for what they actually use.
I suspect that if we really want to be secure than the dongle is the only way to go.
For website login something similar to RSA SecurID would work. Better still, give the dongle some way to communicate bidirectionally (USB for web, speaker/mic/modem for phone), and then it can do an RSA challenge/response. The private key never leaves the dongle, EVER.
For encryption you can again use a dongle. In this case you probably need to have the session key leave the dongle since otherwise you need a high-speed link and the dongle has to stay attached to the computer.
The dongle has a keypad/display, and requires the input of a PIN to work. The dongle of course is built to resist hardware-level attacks (SEMs/etc). Upon input of an owner-designated fake PIN, the dongle wipes its memory.
The limiting factors with this approach are:
1. Hardware resistance of the dongle. It has to be VERY resistant to being taken apart without destruction of the keys. 2. For encryption, the implementation on the host PC is important since it will see the session key. It is critical that it never be written to disk or otherwise leak. 3. If a PC is captured in a powered on state it might be possible to break in with the encryption key still in RAM, or to use hardware-level attacks on the PC to retrieve the contents of RAM. 4. Other hardware-level attacks against a host PC are possible, such as tapping various busses/etc.
Weaknesses #2-4 only apply to use for encryption by the host. Encryption performed by the dongle, and authentication have only weakness #1, since the key doesn't leave the device. The application being provided with the authentication could of course have other weaknesses (replay attacks, etc), but those aren't inherent to this design.
I didn't suggest that Obama should go walking into offices in TSA personally investigating things. However, he could appoint somebody to try to promote government openness, and they could go around doing these sorts of things.
I don't think that all civil servants are lazy. The problem is that quite a few are, and for the most part they are more concerned with not messing up than with achieving results. It is like that in private companies too - I'm not under any illusions that this is purely a government phenomenon.
Obama needs to take hold of the reigns and let those under him know that he's serious about reform - if he is actually serious about reform.
I don't buy it. The problem is that the guys on the top (including Obama) are more interested in patronage than results. The reason Obama doesn't clean house in the TSA is because when he is out of office he doesn't want his friend's kids fired for not doing their jobs by the next president. Even when public servants do something way out of line it usually leads to a long administrative process, and most of the time the government employee is paid to not work the whole time.
To be honest, a lot of industry isn't a whole lot better, although financial strain can at least cause some natural weeding out. The problem with the federal government is that it can just print money, so it isn't forced to go through layoffs/etc and cut the dead wood.
There is no legal reason why Obama couldn't take note of inconsistency with his policy and have somebody make a few polite phone calls to straighten things out. If that doesn't fix the issue then they could start firing people until the rest fall into line. It doesn't take too much shake-up to get things moving once people are less complacent about their jobs.
You could probably get something "good enough" by projecting the show on a nice big TV and pointing a video camera at it, and then capturing the audio direct from the stereo (of if you must, from the wires in the speakers). Until TV is piped directly into your visual cortex, there will ALWAYS be an analog hole. Are they going to start watermarking every broadcast and require the recording of ID to buy a DVD or a cable box?
There was a choice to use some precompiled base binaries that would have sped things up some but that sort of defeats the point.
If you're just going to install using the defaults then you have no benefit from not using the canned binaries. Even if you are, just install the canned binaries and do a rebuild in the background while the system is otherwise productive.
Gentoo does have some performance benefits, but most of the benefits are the ability to run almost any combination of libraries and packages desired.
For example, suppose you have some program that doesn't like the latest glibc - then you don't install the latest glibc. If you tried that on ubuntu then none of your other packages would work, as they are all linked against a specific version of the library.
Sure, it can be annoying sometime to have to wait to use a new program while it builds, but for the most part compiling takes place when you aren't otherwise using the computer, but the benefits come when you are using it.
It isn't the best option for everybody, but Gentoo does have a few unique benefits that you won't find elsewhere. Like everything, it has its place...
Yup, cell providers could comply with the law by eliminating all but their prepaid plans. Of course, if any of them had enough of a brain to just quit gouging people then they could quickly take over the whole market since most people wouldn't put up with the typical prepaid plan either.
See my other reply about prepaid plans. They're fine, but as you say they tend to be non-competitive unless you have credit problems.
Makes you wonder where all that tuition money is going. The average student graduates something like $100k in debt it seems - probably a lot more these days. That is an incredible amount of money to spend on an eduction, and for the life of me I can't figure out where it is all going.
Consider on top of all of this the cost of a modern college education. There really is no excuse for somebody to pay $30k in tuition and be taught by somebody who would rather be someplace else. With each student paying several thousand dollars per class there should be PLENTY of money to hire somebody who knows what they're doing. If the administrators can't find somebody, then they should offer a huge discount.
Sometimes I really question whether a four year degree really confers that much value compared to its cost. Sure, an education is a nice thing to have, but so is a BMW 5-series. I can do without the later, perhaps we should make due with less of the former as well.
Aside from what others have pointed out (your join is not equivalent, and virtually all modern rdbms's will do an inner join given the syntax you supplied), I wanted to point out one other thing: ANSI SQL.
I'm not an expert on SQL, but if I'm not mistaken, the syntax you suggested shouldn't be used is the ONLY syntax for an inner join that actually works on every ANSI-compliant database server out there. I don't think that JOIN is a valid keyword in ANSI SQL, even though it is obviously a universal concept.
I think one of the biggest issues out there is that ANSI SQL is getting fairly rusty, and as a result almost everybody uses some vendor-specific SQL variant. An SQL statement written for mysql will most likely not work in Oracle, and vice-versa.
ANSI SQL needs a major update.
(If I'm wrong on the JOIN bit by all means speak up - but the last time I went looking I couldn't find any syntax for JOINs other than the typical a=b or a*=b or a=*b ANSI syntax. I don't think there is any standard way of doing an OUTER join at all.)
Prepaid plans typically are more expensive or have other downsides associated with them.
I'm not asking for free credit for people who might not be able to pay their bills. I'm just saying that consumers should be able to control how much they end up spending on a utility.
It should not be possible to "accidentally" run up a $10k phone bill by carrying an iphone onto a cruise ship without being familiar with the nature of data roaming plans. $10k is an enormous amount of money, and consumers should not have to negotiate down the charges from a position of weakness having already incurred the charges. If somebody signed up for a $100 per month plan, chances are they wanted to spend $100 per month, and not $500 per month. Adding in a mountain of fine print does not make the agreement fair. One should not need to have degrees in law, accounting, and IT to operate a cell phone responsibly.
Phone companies should be free to offer any services they'd like, and consumers should be free to buy them. However, when money changes hands both sides of the trade should fully understand what they're getting into.
Oh, the phone providers will cap your phone if they don't think you have the means to pay the bill. They won't even give you anything but a prepaid plan if they don't trust your credit.
However, they won't cap it at the request of a consumer who doesn't want to get stuck with a bill that they don't want to pay.
This isn't a cap to protect the phone companies - they already have those. This is a cap to protect the consumer from predatory pricing practices.
That's a great idea. Except when lines fall or circuits get shut down etc. you might loose communications with your control system.
And that is why for critical systems you need humans in the loop.
I always scratch my head a little when I see articles about Naval destroyers that can operate with only 12 people on board. I have to wonder what they plan on doing when they get hit by a shell and you end up with a 1" seam with water pouring in. If you had 100 guys to run damage control it would probably be a non-issue. With only 12 people on board even a small leak is going to be a big problem. Sure, you could easily evacuate those 12 people and not risk too many lives, and that is a viable option if a destroyer costs $1M, but if it costs $1B you're going to want to try to save it.
Automation is great for saving costs, but it never will be able to deal with emergencies as well as having a staff of trained people on-call. When we're talking about the national electric grid, I think we can afford to have a few engineers around.
Indeed - almost all of the cell providers should be fined for deceptive trade practices the way they sucker people into huge bills - all agreed to in fine print.
All of this would be fixed by a very simple law - anybody ought to be able to set a limit on their monthly bill. If I call up evilphoneco and tell them that my cell phone bill is capped at $90 per month, then I don't care WHAT services I consume - they can't charge me more than $90 per month. It is their job to keep me from using services I haven't paid for - not my job to avoid accidentally incurring them. And no incurring of debts either - if they deliver me $10k worth of services they can bill me $90 once and then we're even.
Oh, the second part of that law would be that everybody's cap starts out at whatever their basic monthly rate is. Unless somebody specifically requests a higher limit they couldn't be charged for any "optional" services.
And no giving people limited choices like $10/month or unlimited only. People should be able to name their own limits as arbitrarily as they'd like.
It seems like phone companies depend on people making $500 mistakes with their cell phones, and they almost count on people doing it. They get zero mercy when it happens. At best they might be offered an option for $10 per month to cap their bill. That should NOT be something that costs money.
The character set conversion issue described on that page is due to the default MySQL settings as shipped with Gentoo.
If I'm not mistaken, the Gentoo defaults were those recommended by the mysql project. Essentially, Gentoo just passed on the mysql defaults without fiddling with them.
IMHO, no application should rely on database default settings, or if they do they should rely on them being set to the defaults recommended by the database vendor / ANSI / etc. The reason is simple - if you don't do it this way then suddenly two applications using the same database server could have differing configuration requirements, with no way to satisfy them.
Yup, and the geniuses in the courts will probably go ahead and choose not to enforce that provision, but they will enforce the requirement to cover pre-existing conditions. Then we can all watch the insurance industry go bankrupt.
I'm not saying that I'm a big fan of mandatory coverage, but it is the only way you can cover pre-existing conditions. Otherwise everybody will just drop all coverage at all times except when they're sick.
However, it really is the only option if you're going to force coverage of pre-existing conditions. Otherwise people will just wait until they're sick to sign up for insurance. That would bankrupt any insurance company, since they'd have only sick customers and couldn't share the costs with the healthy. Why have insurance if you can just sign up from the hospital lobby and emergency care is otherwise guaranteed?
If you're going to require coverage for pre-existing conditions then coverage has to be mandatory. That's how every other government on the planet effectively does it...
Actually there is a proposal to fund drug research without patents, see An alternative to pharmaceutical patents. There are others but that's the only link I have.
I've proposed something similar, but with a bit of a twist:
Leave patents exactly how they are right now. If a company develops a drug it can charge whatever the market will bear.
In parallel, go ahead and beef up government R&D just as the site you linked proposed. The government programs would fund the whole development process from start to end (not just stopping at the early basic research point). Any drugs that come out of the effort could be manufactured by anybody royalty-free, so they'll be dirt cheap. The government could even hire industry to do parts of the work if it made sense, but this would be a strict work-for-hire effort with the government paying the full bill but keeping the patent rights.
So, in the end we have competition between government and private efforts. Some fear that the government won't do a good job, and if that happens we'll at least have the status quo to fall back on. Some argue that the government can do the whole thing on its own, and if that turns out to be true chances are the private industry will naturally wither away since they can't compete with the resulting cheap drugs. Either way consumers win with the competition, and the risk of messing up and killing the goose that laid the golden egg is a lot lower.
The NCI could have licensed it for say a hundred dollars a dose to any company that wanted to manufacture it.
Absolutely - of course the NCI would actually have to fund development of the drug all the way to completion for anybody to be interested. I'm actually a big fan of this model for drugs that the government does wish to fund - do the development soup-to-nuts and don't sell the patent rights at all.
And those costs are tax deductible.
All corporate costs are tax deductible - corporations pay taxes on profits so if you reduce your profits you reduce your taxes. That doesn't mean that corporations are happy when their profits are reduced.
Also one blockbuster drug can pay for the research, development, and testing for a bunch of other drugs.
Absolutely. But, only if you can sell the blockbuster drug for what the market will bear. If you are forced to sell at near-marginal-cost (the whole point of this thread) then there are no blockbuster drugs.
I'm not saying that the status quo doesn't make any money for pharma companies, just like nobody is saying that the status quo didn't make money for record labels or movie producers (the whole point of this thread). The argument was made that you can still make records and movies when you're only selling product for a hair over the marginal cost of distribution, and my argument was that this doesn't work for bigger-ticket items like movies and drugs.
Ah but companies enjoy patent monopolies and can set their own prices for a number of years before they have to compeat with others for the market share of a drug. Combined with blockbuster drugs there should be enough to cover the cost of drug research. But if that isn't enough then the companies can cut their marketing costs, which are more than research.
Well, sure, that's how it works right now. However, the whole point of this discussion was whether you can make initial-cost-heavy-marginal-cost-light items without IP protection. If you get rid of p2p copying of music it is trivially easy to make money off of movies. If you envision a world where everybody has a super-quality home theater and easy access to free downloads, then it isn't so easy to make money off of movies. Likewise, if you have a world where there are no drug patents then it isn't all that easy to make money off of drugs either.
I'm not trying to suggest that drug companies didn't make money in the 90s. I am suggesting that it is hard for drug and movie makers to make money in the 2010s. This isn't an easy problem to fix unless you just have the government do ALL the drug R&D and pay for movie production as well.
The pharmaceutical industry spend much more on marketing than on research.
Agreed.
Don't tell me drug companies spend a lot to develop drugs.
Drug companies spend a lot to develop drugs. Sorry to say it, but it is still true. If they didn't spend a dime on marketing and administration they'd still be spending many billions of dollars a year on development.
You cited taxol as an example, but it is an extremely atypical example. It is rare for government agencies to do any development of a drug candidate at all. Typically somebody comes up with a proof of concept molecule and a couple of tests that suggests that it might do something. It is extremely rare for a government agency to fund full clinical trials, and those are the most expensive part of a drug development effort by far. Additionally, I couldn't find any clear indication online that the government had funded complete clinical trials - perhaps just phase I trials which don't cost much, and also which don't detect many issues which tend to kill drugs.
I'm not sure what kind of a bidding process was used to sell Taxol, but if it was competitive, then there must have been a reason that nobody offered big money for it. If it wasn't competitive then it was just an example of government corruption.
In any case, I'm sure BMS still invested more money into producing Taxol than is spent on the typical movie or record album, which was the whole point of the comparison. The marginal cost of production doesn't cover those costs. If a new drug like Taxol were allowed to be sold completely royalty-free it would be difficult to even recoup the costs that were paid just to license it from the government in the first place.
Drugs and movies also have other things in common. The cost of failures is amortized into the profits of successes. With drugs it is almost impossible to predict whether a given molecule is going to turn out to be safe, and it isn't uncommon for drug programs to get canceled even after hundreds of millions of dollars are spent on them. For movies you'd think that success would be a lot easier to predict, but there are still lots of flops out there where investors make no money, or little money. So, when doing the accounting for a successful production you do need to consider the risk of failure. A $100M investment that makes $2B seems like a huge success, but if the risk of failure was 99% then it was a foolish investment. It is like saying that we need to do something about all those rich lottery winners getting away with murder - they only have costs of $1 and yet they make $50M. Sure, if you only look at the winners it seems that way, but lottery winners are just lucky fools. The comparison breaks down since no rational investor would ever buy lottery tickets, but drugs and movies are areas where profits can currently be made (but not if the product is priced at marginal cost plus a few percent).
For music this shouldn't be that hard of a problem to solve, as the cost of making music is fairly low. If you want big salaries for the artists there are still some issues, but at least the problem is in the realms of possibility.
For movies the problems are much larger, because the costs are a whole lot higher. You need hundreds of people to make a summer blockbuster, even if you trim out ALL of the fat. You can't get by on a couple of cents here or there in revenue.
The situation isn't unlike that faced by drug companies - the pills cost cents to make, but determining of they are safe costs hundreds of millions of dollars, plus lots of waste on stuff that doesn't pan out.
In a society where the marginal cost of reproduction approaches zero, you need to find new models of paying for the creation of creative works in the first place.
Why do ANY two processes share the same address space? Shouldn't it all be virtualized? So, two different processes could be assigned a block of memory with the same address, but in reality neither corresponds to the same region in physical memory?
Ok, maybe as some kind of an optimization you do give processes non-colliding address spaces. Well, even then there should be some kind of privilege assignments so that processes can't access each other's memories.
Otherwise, stuff like this can happen.
I'll be the first to admit that I'm not an expert in x86 protected mode, but it just seems like a multi-user OS should completely isolate the memory of its processes from each other. If a process wants to mmap something into address page zero, no big deal - 100 processes can all do it and they'll all be happy and none of them will tread over each other. If you spawn a new process it should get a clean address space - ESPECIALLY if it is suid.
Slashdot Mirror
Yup, and it works out great until they realize that the same stack leak causing the crash on the 17th save also causes data to get silently lost in some other scenario.
When writing large programs it is critical to validate your inputs and carefully check stuff like this. Just because the way the error was caught was unlikely to come up in real life, doesn't mean that the same bug won't blow up in some other way. The only exception is if you actually ANALYZE the problem and determine that it isn't a big risk. But if you've done that then you've done half the work needed to fix it (unless you're in a shop that writes 10 pounds of documentation for one line of code change).
Nice to see - wonder if it will last...
I knew somebody with a Palm-based smartphone on verizon quite a few years back, and its bluetooth wouldn't work with his 3-series BMW. Apparently verizon crippled the address book feature. That is typical verizon. It will be nice if they turn over a new leaf - I doubt they'd be doing it if they weren't getting killed in the smartphone arena.
Note that no US cell network has a 3G capability that works outside of the US, at least as far as I understand it. Granted, with the alphabet soup mess who can blame me if I am missing something?
That said, just about everybody but Verizon and Sprint have switched to GSM, which is of course the standard everywhere else (except Japan, apparently).
What a mess... You'll never see it fixed since the providers have too much at stake to allow their pet standard to lose. GSM is the obvious choice, but to somehow mandate a switch would probably put verizon at a substantial disadvantage, and that reduction in competition won't help consumers on the other networks either.
I still don't understand why data isn't data. If I pay for data transfer on their net why does it matter if that comes from a laptop connected to the phone (an extra charge) or from an app running on the phone itself? Are the bits fatter?
The problem is the word "unlimited". This word should probably be banned, except when it literally means use without any limits at all.
When verizon sells you unlimited service, they can do so because they know that a typical phone-based use won't use more than a certain amount of bandwidth. On the other hand, if you tether the thing to a laptop you can fire up bittorrent or download a high-def movie. You wouldn't do that on a phone since the phone probably wouldn't have the storage for something like that, apps capable of doing it, and why download a 10GB blu-ray image to watch it on a VGA-sized display?
If everybody just charged reasonable rates per byte then all that nonsense could go away. Just make people pay for what they actually use.
I suspect that if we really want to be secure than the dongle is the only way to go.
For website login something similar to RSA SecurID would work. Better still, give the dongle some way to communicate bidirectionally (USB for web, speaker/mic/modem for phone), and then it can do an RSA challenge/response. The private key never leaves the dongle, EVER.
For encryption you can again use a dongle. In this case you probably need to have the session key leave the dongle since otherwise you need a high-speed link and the dongle has to stay attached to the computer.
The dongle has a keypad/display, and requires the input of a PIN to work. The dongle of course is built to resist hardware-level attacks (SEMs/etc). Upon input of an owner-designated fake PIN, the dongle wipes its memory.
The limiting factors with this approach are:
1. Hardware resistance of the dongle. It has to be VERY resistant to being taken apart without destruction of the keys.
2. For encryption, the implementation on the host PC is important since it will see the session key. It is critical that it never be written to disk or otherwise leak.
3. If a PC is captured in a powered on state it might be possible to break in with the encryption key still in RAM, or to use hardware-level attacks on the PC to retrieve the contents of RAM.
4. Other hardware-level attacks against a host PC are possible, such as tapping various busses/etc.
Weaknesses #2-4 only apply to use for encryption by the host. Encryption performed by the dongle, and authentication have only weakness #1, since the key doesn't leave the device. The application being provided with the authentication could of course have other weaknesses (replay attacks, etc), but those aren't inherent to this design.
I didn't suggest that Obama should go walking into offices in TSA personally investigating things. However, he could appoint somebody to try to promote government openness, and they could go around doing these sorts of things.
I don't think that all civil servants are lazy. The problem is that quite a few are, and for the most part they are more concerned with not messing up than with achieving results. It is like that in private companies too - I'm not under any illusions that this is purely a government phenomenon.
Obama needs to take hold of the reigns and let those under him know that he's serious about reform - if he is actually serious about reform.
I don't buy it. The problem is that the guys on the top (including Obama) are more interested in patronage than results. The reason Obama doesn't clean house in the TSA is because when he is out of office he doesn't want his friend's kids fired for not doing their jobs by the next president. Even when public servants do something way out of line it usually leads to a long administrative process, and most of the time the government employee is paid to not work the whole time.
To be honest, a lot of industry isn't a whole lot better, although financial strain can at least cause some natural weeding out. The problem with the federal government is that it can just print money, so it isn't forced to go through layoffs/etc and cut the dead wood.
There is no legal reason why Obama couldn't take note of inconsistency with his policy and have somebody make a few polite phone calls to straighten things out. If that doesn't fix the issue then they could start firing people until the rest fall into line. It doesn't take too much shake-up to get things moving once people are less complacent about their jobs.
You could probably get something "good enough" by projecting the show on a nice big TV and pointing a video camera at it, and then capturing the audio direct from the stereo (of if you must, from the wires in the speakers). Until TV is piped directly into your visual cortex, there will ALWAYS be an analog hole. Are they going to start watermarking every broadcast and require the recording of ID to buy a DVD or a cable box?
There was a choice to use some precompiled base binaries that would have sped things up some but that sort of defeats the point.
If you're just going to install using the defaults then you have no benefit from not using the canned binaries. Even if you are, just install the canned binaries and do a rebuild in the background while the system is otherwise productive.
Gentoo does have some performance benefits, but most of the benefits are the ability to run almost any combination of libraries and packages desired.
For example, suppose you have some program that doesn't like the latest glibc - then you don't install the latest glibc. If you tried that on ubuntu then none of your other packages would work, as they are all linked against a specific version of the library.
Sure, it can be annoying sometime to have to wait to use a new program while it builds, but for the most part compiling takes place when you aren't otherwise using the computer, but the benefits come when you are using it.
It isn't the best option for everybody, but Gentoo does have a few unique benefits that you won't find elsewhere. Like everything, it has its place...
Yup, cell providers could comply with the law by eliminating all but their prepaid plans. Of course, if any of them had enough of a brain to just quit gouging people then they could quickly take over the whole market since most people wouldn't put up with the typical prepaid plan either.
See my other reply about prepaid plans. They're fine, but as you say they tend to be non-competitive unless you have credit problems.
Makes you wonder where all that tuition money is going. The average student graduates something like $100k in debt it seems - probably a lot more these days. That is an incredible amount of money to spend on an eduction, and for the life of me I can't figure out where it is all going.
Consider on top of all of this the cost of a modern college education. There really is no excuse for somebody to pay $30k in tuition and be taught by somebody who would rather be someplace else. With each student paying several thousand dollars per class there should be PLENTY of money to hire somebody who knows what they're doing. If the administrators can't find somebody, then they should offer a huge discount.
Sometimes I really question whether a four year degree really confers that much value compared to its cost. Sure, an education is a nice thing to have, but so is a BMW 5-series. I can do without the later, perhaps we should make due with less of the former as well.
Aside from what others have pointed out (your join is not equivalent, and virtually all modern rdbms's will do an inner join given the syntax you supplied), I wanted to point out one other thing: ANSI SQL.
I'm not an expert on SQL, but if I'm not mistaken, the syntax you suggested shouldn't be used is the ONLY syntax for an inner join that actually works on every ANSI-compliant database server out there. I don't think that JOIN is a valid keyword in ANSI SQL, even though it is obviously a universal concept.
I think one of the biggest issues out there is that ANSI SQL is getting fairly rusty, and as a result almost everybody uses some vendor-specific SQL variant. An SQL statement written for mysql will most likely not work in Oracle, and vice-versa.
ANSI SQL needs a major update.
(If I'm wrong on the JOIN bit by all means speak up - but the last time I went looking I couldn't find any syntax for JOINs other than the typical a=b or a*=b or a=*b ANSI syntax. I don't think there is any standard way of doing an OUTER join at all.)
Prepaid plans typically are more expensive or have other downsides associated with them.
I'm not asking for free credit for people who might not be able to pay their bills. I'm just saying that consumers should be able to control how much they end up spending on a utility.
It should not be possible to "accidentally" run up a $10k phone bill by carrying an iphone onto a cruise ship without being familiar with the nature of data roaming plans. $10k is an enormous amount of money, and consumers should not have to negotiate down the charges from a position of weakness having already incurred the charges. If somebody signed up for a $100 per month plan, chances are they wanted to spend $100 per month, and not $500 per month. Adding in a mountain of fine print does not make the agreement fair. One should not need to have degrees in law, accounting, and IT to operate a cell phone responsibly.
Phone companies should be free to offer any services they'd like, and consumers should be free to buy them. However, when money changes hands both sides of the trade should fully understand what they're getting into.
Oh, the phone providers will cap your phone if they don't think you have the means to pay the bill. They won't even give you anything but a prepaid plan if they don't trust your credit.
However, they won't cap it at the request of a consumer who doesn't want to get stuck with a bill that they don't want to pay.
This isn't a cap to protect the phone companies - they already have those. This is a cap to protect the consumer from predatory pricing practices.
That's a great idea. Except when lines fall or circuits get shut down etc. you might loose communications with your control system.
And that is why for critical systems you need humans in the loop.
I always scratch my head a little when I see articles about Naval destroyers that can operate with only 12 people on board. I have to wonder what they plan on doing when they get hit by a shell and you end up with a 1" seam with water pouring in. If you had 100 guys to run damage control it would probably be a non-issue. With only 12 people on board even a small leak is going to be a big problem. Sure, you could easily evacuate those 12 people and not risk too many lives, and that is a viable option if a destroyer costs $1M, but if it costs $1B you're going to want to try to save it.
Automation is great for saving costs, but it never will be able to deal with emergencies as well as having a staff of trained people on-call. When we're talking about the national electric grid, I think we can afford to have a few engineers around.
Indeed - almost all of the cell providers should be fined for deceptive trade practices the way they sucker people into huge bills - all agreed to in fine print.
All of this would be fixed by a very simple law - anybody ought to be able to set a limit on their monthly bill. If I call up evilphoneco and tell them that my cell phone bill is capped at $90 per month, then I don't care WHAT services I consume - they can't charge me more than $90 per month. It is their job to keep me from using services I haven't paid for - not my job to avoid accidentally incurring them. And no incurring of debts either - if they deliver me $10k worth of services they can bill me $90 once and then we're even.
Oh, the second part of that law would be that everybody's cap starts out at whatever their basic monthly rate is. Unless somebody specifically requests a higher limit they couldn't be charged for any "optional" services.
And no giving people limited choices like $10/month or unlimited only. People should be able to name their own limits as arbitrarily as they'd like.
It seems like phone companies depend on people making $500 mistakes with their cell phones, and they almost count on people doing it. They get zero mercy when it happens. At best they might be offered an option for $10 per month to cap their bill. That should NOT be something that costs money.
The character set conversion issue described on that page is due to the default MySQL settings as shipped with Gentoo.
If I'm not mistaken, the Gentoo defaults were those recommended by the mysql project. Essentially, Gentoo just passed on the mysql defaults without fiddling with them.
IMHO, no application should rely on database default settings, or if they do they should rely on them being set to the defaults recommended by the database vendor / ANSI / etc. The reason is simple - if you don't do it this way then suddenly two applications using the same database server could have differing configuration requirements, with no way to satisfy them.
Yup, and the geniuses in the courts will probably go ahead and choose not to enforce that provision, but they will enforce the requirement to cover pre-existing conditions. Then we can all watch the insurance industry go bankrupt.
I'm not saying that I'm a big fan of mandatory coverage, but it is the only way you can cover pre-existing conditions. Otherwise everybody will just drop all coverage at all times except when they're sick.
Yup - I'm not a big fan of this.
However, it really is the only option if you're going to force coverage of pre-existing conditions. Otherwise people will just wait until they're sick to sign up for insurance. That would bankrupt any insurance company, since they'd have only sick customers and couldn't share the costs with the healthy. Why have insurance if you can just sign up from the hospital lobby and emergency care is otherwise guaranteed?
If you're going to require coverage for pre-existing conditions then coverage has to be mandatory. That's how every other government on the planet effectively does it...
Actually there is a proposal to fund drug research without patents, see An alternative to pharmaceutical patents. There are others but that's the only link I have.
I've proposed something similar, but with a bit of a twist:
Leave patents exactly how they are right now. If a company develops a drug it can charge whatever the market will bear.
In parallel, go ahead and beef up government R&D just as the site you linked proposed. The government programs would fund the whole development process from start to end (not just stopping at the early basic research point). Any drugs that come out of the effort could be manufactured by anybody royalty-free, so they'll be dirt cheap. The government could even hire industry to do parts of the work if it made sense, but this would be a strict work-for-hire effort with the government paying the full bill but keeping the patent rights.
So, in the end we have competition between government and private efforts. Some fear that the government won't do a good job, and if that happens we'll at least have the status quo to fall back on. Some argue that the government can do the whole thing on its own, and if that turns out to be true chances are the private industry will naturally wither away since they can't compete with the resulting cheap drugs. Either way consumers win with the competition, and the risk of messing up and killing the goose that laid the golden egg is a lot lower.
The NCI could have licensed it for say a hundred dollars a dose to any company that wanted to manufacture it.
Absolutely - of course the NCI would actually have to fund development of the drug all the way to completion for anybody to be interested. I'm actually a big fan of this model for drugs that the government does wish to fund - do the development soup-to-nuts and don't sell the patent rights at all.
And those costs are tax deductible.
All corporate costs are tax deductible - corporations pay taxes on profits so if you reduce your profits you reduce your taxes. That doesn't mean that corporations are happy when their profits are reduced.
Also one blockbuster drug can pay for the research, development, and testing for a bunch of other drugs.
Absolutely. But, only if you can sell the blockbuster drug for what the market will bear. If you are forced to sell at near-marginal-cost (the whole point of this thread) then there are no blockbuster drugs.
I'm not saying that the status quo doesn't make any money for pharma companies, just like nobody is saying that the status quo didn't make money for record labels or movie producers (the whole point of this thread). The argument was made that you can still make records and movies when you're only selling product for a hair over the marginal cost of distribution, and my argument was that this doesn't work for bigger-ticket items like movies and drugs.
Ah but companies enjoy patent monopolies and can set their own prices for a number of years before they have to compeat with others for the market share of a drug. Combined with blockbuster drugs there should be enough to cover the cost of drug research. But if that isn't enough then the companies can cut their marketing costs, which are more than research.
Well, sure, that's how it works right now. However, the whole point of this discussion was whether you can make initial-cost-heavy-marginal-cost-light items without IP protection. If you get rid of p2p copying of music it is trivially easy to make money off of movies. If you envision a world where everybody has a super-quality home theater and easy access to free downloads, then it isn't so easy to make money off of movies. Likewise, if you have a world where there are no drug patents then it isn't all that easy to make money off of drugs either.
I'm not trying to suggest that drug companies didn't make money in the 90s. I am suggesting that it is hard for drug and movie makers to make money in the 2010s. This isn't an easy problem to fix unless you just have the government do ALL the drug R&D and pay for movie production as well.
The pharmaceutical industry spend much more on marketing than on research.
Agreed.
Don't tell me drug companies spend a lot to develop drugs.
Drug companies spend a lot to develop drugs. Sorry to say it, but it is still true. If they didn't spend a dime on marketing and administration they'd still be spending many billions of dollars a year on development.
You cited taxol as an example, but it is an extremely atypical example. It is rare for government agencies to do any development of a drug candidate at all. Typically somebody comes up with a proof of concept molecule and a couple of tests that suggests that it might do something. It is extremely rare for a government agency to fund full clinical trials, and those are the most expensive part of a drug development effort by far. Additionally, I couldn't find any clear indication online that the government had funded complete clinical trials - perhaps just phase I trials which don't cost much, and also which don't detect many issues which tend to kill drugs.
I'm not sure what kind of a bidding process was used to sell Taxol, but if it was competitive, then there must have been a reason that nobody offered big money for it. If it wasn't competitive then it was just an example of government corruption.
In any case, I'm sure BMS still invested more money into producing Taxol than is spent on the typical movie or record album, which was the whole point of the comparison. The marginal cost of production doesn't cover those costs. If a new drug like Taxol were allowed to be sold completely royalty-free it would be difficult to even recoup the costs that were paid just to license it from the government in the first place.
Drugs and movies also have other things in common. The cost of failures is amortized into the profits of successes. With drugs it is almost impossible to predict whether a given molecule is going to turn out to be safe, and it isn't uncommon for drug programs to get canceled even after hundreds of millions of dollars are spent on them. For movies you'd think that success would be a lot easier to predict, but there are still lots of flops out there where investors make no money, or little money. So, when doing the accounting for a successful production you do need to consider the risk of failure. A $100M investment that makes $2B seems like a huge success, but if the risk of failure was 99% then it was a foolish investment. It is like saying that we need to do something about all those rich lottery winners getting away with murder - they only have costs of $1 and yet they make $50M. Sure, if you only look at the winners it seems that way, but lottery winners are just lucky fools. The comparison breaks down since no rational investor would ever buy lottery tickets, but drugs and movies are areas where profits can currently be made (but not if the product is priced at marginal cost plus a few percent).
Agreed.
For music this shouldn't be that hard of a problem to solve, as the cost of making music is fairly low. If you want big salaries for the artists there are still some issues, but at least the problem is in the realms of possibility.
For movies the problems are much larger, because the costs are a whole lot higher. You need hundreds of people to make a summer blockbuster, even if you trim out ALL of the fat. You can't get by on a couple of cents here or there in revenue.
The situation isn't unlike that faced by drug companies - the pills cost cents to make, but determining of they are safe costs hundreds of millions of dollars, plus lots of waste on stuff that doesn't pan out.
In a society where the marginal cost of reproduction approaches zero, you need to find new models of paying for the creation of creative works in the first place.
Here is the thing I don't get.
Why do ANY two processes share the same address space? Shouldn't it all be virtualized? So, two different processes could be assigned a block of memory with the same address, but in reality neither corresponds to the same region in physical memory?
Ok, maybe as some kind of an optimization you do give processes non-colliding address spaces. Well, even then there should be some kind of privilege assignments so that processes can't access each other's memories.
Otherwise, stuff like this can happen.
I'll be the first to admit that I'm not an expert in x86 protected mode, but it just seems like a multi-user OS should completely isolate the memory of its processes from each other. If a process wants to mmap something into address page zero, no big deal - 100 processes can all do it and they'll all be happy and none of them will tread over each other. If you spawn a new process it should get a clean address space - ESPECIALLY if it is suid.