Sorry, you guys got it right on the metric thing, but this one's silly: a million is a thousand thousands, a billion is a thousand thousand thousands, a trillion is a thousand thousand thousand thousands, and so on. The European version is just unnatural.
Honey lamb, I don't know what planet you're posting from, but here on Earth a thousand million is a billion. We used to call 1024 bytes '1 kilobyte' and 1024 kilobytes '1 megabyte' because memory came in natural sizes that were powers of two.
We thereby confused the bejeezus out of everyone who wasn't used to talking in powers of two.
This is going to turn out to be one of those class-action suits were some people get a coupon and the lawyers collection $10 million in fees, you mark my words.
It doesn't mitigate the damage... but it skews the numbers. Consider, eg, if there was a Cisco router load-balancing the 10,000 sites on 100 servers, and someone hacked the router: then there would have been only one exploit, but 10,000 web sites would have been affected. Does it make sense to claim that the attack was 10,000 times as bad as hitting another router that was in front of a single site with 10,000 pages balanced across 100 servers?
The underlying point is a good old engineering school point: you don't know what a measurement means until you understand the units.
I don't know what their methodology was, but from looking at the results from ethereal, it's clear that there were more than 20 Windows boxes that were successfuly attacked on my broadband provider's local NAT domain alone. I doubt the proportion of clueless Windows users in this subnet is unusually high (if anything, it's likely low) so it seems very probable that many tens of thousands of windows bozes were attcked by SoBig alone.
It seems therefore extremely unlikely that only 4000-odd Windows boxes were hacked total in their study. This makes me suspect that they are playing fast and loose with their counting methods.
It's always hard to get this across to my clients, but you need first of all to answer the question "how much reliability/availability do you need?"
Think of it this way: if what you need is no better than 99 percent, you need to be able to fail over fast enough to only have 864 minutes of downtime a week. Of course, that's about 14 minutes, so you can practically handle it by doing a hand cutover.
On the other hand, if you need availability of 99.999 percent ("five nines") you can only afford to have about 40 seconds a week total downtime.
If you need true five-nines, you need to look at some of those nasty commerical apps. If you can slack off from there, the PostGreSQL and nySQL replication schemes work just fine, and you can use DNS remapping to do the failover. (I've done it with mySQL, it worked; think about needing a heartbeat to detect if the master server has gone down.)
Java becoming stagnant?
on
Java vs .NET
·
· Score: 1
Frankly, my complaint is that Java isn't stagnant enough. I've beeb writing Java for as long as there's been a Java, and I've been a Java architect for both Sun and IBM -- so, to the extent that one can grow old and gray in Java, I've done so. And for me, the biggest annoyance is that that damned language keeps changing. New keywords are added, new facilities are added -- not that I object to new class libraries, but there's an annoying pattern to it: someone comes up with something really useful, like log4j; Javasoft decides to adopt it; but when they do it's not compatible with the original notion. So now instead of a uniform, popular technology like log4j, we've got log4j AND java.util.logging. (And then to try to solve that, someone develops a wrapper class so that you can use either one, viz Apache. Now we've got THREE logging schemes.)
Similarly things have happened in 1.5 with, eg, automatic coercion between primitives and the primitive wrapper types, so you no longer have to write code that does things like
HashMap map = new HashMap(); map.put("One", new Integer(1)); ... int val = ((Integer)map.get("One")).intValue();
when what you wanted was a dictionary with integer values and string keys. (And yes, you can write a class MapStringsToInts for this, but then you're adding something else incompatible, that will be obviated in 1.5.)
The point is that the language -- not just the implementation, the language -- changes fairly significantly at every.X version. This means that I, as a consultant, have to keep many versions of the JDK to deal with different customers, and fairly often I write code that fails because I wrote a 1.4-ism in a customer's 1.3.1-compatible code. It makes the language harder to use, and deters people from adopting it.
What we ought to do is declare a feature freeze at 1.5; new things that can't be implemented without changing the language don't get in, and everything that requires new class libraries should be made an extension, not part of the base language.
Sympathize (or at least empathize) yes --and more so now, since I'm one of those laid off IT folks. Although it doesn't seem like the notion of jobs going overseas (or to Canada) works all that well, because what I'm finding is that all the Markov modeling and architect skills I've developed aren't selling, but writing SWING GUIs is. If it were jobs overseas that were the issue, you'd imagine that the high-level skills that require customer interaction would be the ones most in demand. But that's anechdotal.
We've also got to recognize that we did a good bit of this to ourselves. I still see the same damn dumb mistakes as I did 30 years ago. In theory I've spent my 35 years in the industry -- and a PhD in software engineering -- learning how not to make those same damn dumb mistakes, but none the less I go to customers and have to start with the same damn lectures: "this is called configuration management" and "you don't have to have a big process, but you've got to have some process" and "if you don't know where you're going when you start, you won't probably get there." By contrast, the first firm to have a CMM level 5 (or in other words, a well-measured, repeatable, self-correcting process) was a big software company in Bangalore. You can argue whether CMM5 certification is important, but you can be damn sure that if a customer wants CMM5, that Indian firm is going to be the one that gets the job. Not because they're cheaper -- because they're better.
In any case, though, the answer (as unpleasant as it sometimes seems) is that things change. The only solutions are: (1) win the lottery, make a zillion dollars, and spend your life being a pain in the ass to everyone; or (2) be prepared to change too. Don't expect your next job to be like your last one, and don't expect all change to look like progress -- sometimes you have to take a pay cut to get the job you want.
In the mean time, though, I've gone from getting around 20 hits a day on dice.com to getting 60-100. My guess is that the market is turning up, and that people are also beginning to notice that it can be an incredible pain in the ass to try to do a project 12,000 miles and 16 hours of time-zone difference away.
Frankly, the notion that the linear no-threshold model would hold is, biologically, the more extrodinary claim: it's tantamount to suggesting that there is no biological mechanism that deals with radiation damage in naturally-occuring dose rates. Non-radiological toxins that don't occur in nature are another thing, but it's damned hard to find one that doesn't have a close biological analogue.
No, no, I don't think you're wrong at all -- I suspect that you're on the right track. The underlying mechanism seems to be activation of a cell-repair (and probably an accelerated cell-death) mechanism that either uses some kind of error-correction to clean up ionization damage, or that causes a cell to "suicide" if it's irrepairable. (Now, kids, don't try this at home -- it's been 20 years since medical school and I was just auditing anyway.) Your dose of a couple of centriGray probably gets the mechanism up and going like a fire alarm, while 300 mR/yr is more like a faintly irritating dripping faucet -- but enough to keep the repair mechanism awake.
Sorry, make that two quibbles: Cohen's study was overall incidence of cancer, not just melanoma -- which we have an unfortunate lot of up here, due to high UV rates and an outdoors lifestyle.
Which proves hormesis can only go so far.
But it wouldn't surprise me to find that skin cancer versus lifetime UV exposure isn't linear no-threshold either.
One quibble: hormesis is observed epidemiologically in populations that are exposed to relatively high background exposures without the 'priming' dose characteristic. I believe the experiments you're talking about have a relatively higher dose rate, if only because you could die of boredom waiting for the effects of 300 mR/year to show up in a population of experimental subjects, but that's fuzzy memory and I'd love to be corrected.
A vent fan isn't the only thing needed, at least in some of the amelioration attempts. As well as the vent fan, some places out here have had to have the basement specially sealed, and in a few cases they've had to actually have hot stuff excavated from around the house and replaced with clean sand and concrete. The price of this is in the thousands or tens of thousands.
You bet, although it'll be short (a Google for radiation, epidemiology, and "Bernard Cohen" will get you more, as will "radiation" and "hormesis".
Idaho State has a site on radiaiton effects and radiation protection. They've also got some stuff on depleted uranium that looks good, although I've not read it in detail.
There is a journal devoted to low-dose radiation studies.
This page has about a zillion references. (Some of them disagree, by the way. This makes the page very useful.)
This paper has a nice discussion of hormesis and the linear no-threshold model.
BELLE Online is a good source in general. Also, the names "Bernard Cohen" and "Edward Calabrese" show up quite often.
One point about the "suspiciously low" cancer rates that I didn't say is that the rates in, eg, Cohen's epi paper are smoking-adjusted.
There are many possible reasons for this; maybe Colorado just has a better public health system or healthier lifestyles.
Read the sources before you respond to them. It makes you look less foolish later.
It is quite ironic that people like you often call themselves "conservatives", but then want to subject the US population to historically unprecedented exposures to largely unstudied chemicals and radiation.
I beg your pardon: all I said about "conservative" was that EPA claimed the linear no-threshold model they used was "conservative". Maybe you should read my article before you respond to it. As far as "historically unprecedented" exposure to radiation, you're making, by accident, my point for me: in the case of low doses of radiation, the attempt to eliminate exposure to low doses of radiation is what's unprecedented.
The only news is that ignorant politicians with a corporatist agenda use such obscure scientific tidbits out of context to argue that pollution is harmless.
Your compaint about taking things out of context, after this note, is quite amusing indeed.
It's sorta off topic, but what Freddie baby actually said is "Was mich nicht umbringt, macht mich noch staerker". The important verb "umbringen" means "to kill, to murder, to liquidate" and it's got sort of the sense of "bringing down" that we'd have if we spoke of bringing down a deer with hounds.
So it's really something like "That which can't catch me and kill me makes me even faster and stronger."
Slashdot Mirror
... but not both?
That's okay -- we've got them outnumbered now.
Just by the way -- what do you call a thousand billion?
(And you guys treat collectives as plural, too. "Billions". "The BBC have...." Just wrong.)
I just knew someone was going to bring this up.
Sorry, you guys got it right on the metric thing, but this one's silly: a million is a thousand thousands, a billion is a thousand thousand thousands, a trillion is a thousand thousand thousand thousands, and so on. The European version is just unnatural.
Honey lamb, I don't know what planet you're posting from, but here on Earth a thousand million is a billion. We used to call 1024 bytes '1 kilobyte' and 1024 kilobytes '1 megabyte' because memory came in natural sizes that were powers of two.
We thereby confused the bejeezus out of everyone who wasn't used to talking in powers of two.
This is going to turn out to be one of those class-action suits were some people get a coupon and the lawyers collection $10 million in fees, you mark my words.
It doesn't mitigate the damage ... but it skews the numbers. Consider, eg, if there was a Cisco router load-balancing the 10,000 sites on 100 servers, and someone hacked the router: then there would have been only one exploit, but 10,000 web sites would have been affected. Does it make sense to claim that the attack was 10,000 times as bad as hitting another router that was in front of a single site with 10,000 pages balanced across 100 servers?
The underlying point is a good old engineering school point: you don't know what a measurement means until you understand the units.
You mean to say it's all just a phase?
Cripes, what time was it when I was posting that?
... and on from there. Gack.
86,400 sec/day x 0.01 -- 14.4 minutes.
But the point is still a good one: you have to ask the question, even if you should do your own division.
... he said gently.
I don't know what their methodology was, but from looking at the results from ethereal, it's clear that there were more than 20 Windows boxes that were successfuly attacked on my broadband provider's local NAT domain alone. I doubt the proportion of clueless Windows users in this subnet is unusually high (if anything, it's likely low) so it seems very probable that many tens of thousands of windows bozes were attcked by SoBig alone.
It seems therefore extremely unlikely that only 4000-odd Windows boxes were hacked total in their study. This makes me suspect that they are playing fast and loose with their counting methods.
It's always hard to get this across to my clients, but you need first of all to answer the question "how much reliability/availability do you need?"
Think of it this way: if what you need is no better than 99 percent, you need to be able to fail over fast enough to only have 864 minutes of downtime a week. Of course, that's about 14 minutes, so you can practically handle it by doing a hand cutover.
On the other hand, if you need availability of 99.999 percent ("five nines") you can only afford to have about 40 seconds a week total downtime.
If you need true five-nines, you need to look at some of those nasty commerical apps. If you can slack off from there, the PostGreSQL and nySQL replication schemes work just fine, and you can use DNS remapping to do the failover. (I've done it with mySQL, it worked; think about needing a heartbeat to detect if the master server has gone down.)
Similarly things have happened in 1.5 with, eg, automatic coercion between primitives and the primitive wrapper types, so you no longer have to write code that does things likewhen what you wanted was a dictionary with integer values and string keys. (And yes, you can write a class MapStringsToInts for this, but then you're adding something else incompatible, that will be obviated in 1.5.)
The point is that the language -- not just the implementation, the language -- changes fairly significantly at every
What we ought to do is declare a feature freeze at 1.5; new things that can't be implemented without changing the language don't get in, and everything that requires new class libraries should be made an extension, not part of the base language.
Great, thank you.
This being slashdot, please don't mistake this as argument, but I'd love to see some more extensive sources on this.
Sympathize (or at least empathize) yes --and more so now, since I'm one of those laid off IT folks. Although it doesn't seem like the notion of jobs going overseas (or to Canada) works all that well, because what I'm finding is that all the Markov modeling and architect skills I've developed aren't selling, but writing SWING GUIs is. If it were jobs overseas that were the issue, you'd imagine that the high-level skills that require customer interaction would be the ones most in demand. But that's anechdotal.
We've also got to recognize that we did a good bit of this to ourselves. I still see the same damn dumb mistakes as I did 30 years ago. In theory I've spent my 35 years in the industry -- and a PhD in software engineering -- learning how not to make those same damn dumb mistakes, but none the less I go to customers and have to start with the same damn lectures: "this is called configuration management" and "you don't have to have a big process, but you've got to have some process" and "if you don't know where you're going when you start, you won't probably get there." By contrast, the first firm to have a CMM level 5 (or in other words, a well-measured, repeatable, self-correcting process) was a big software company in Bangalore. You can argue whether CMM5 certification is important, but you can be damn sure that if a customer wants CMM5, that Indian firm is going to be the one that gets the job. Not because they're cheaper -- because they're better.
In any case, though, the answer (as unpleasant as it sometimes seems) is that things change. The only solutions are: (1) win the lottery, make a zillion dollars, and spend your life being a pain in the ass to everyone; or (2) be prepared to change too. Don't expect your next job to be like your last one, and don't expect all change to look like progress -- sometimes you have to take a pay cut to get the job you want.
In the mean time, though, I've gone from getting around 20 hits a day on dice.com to getting 60-100. My guess is that the market is turning up, and that people are also beginning to notice that it can be an incredible pain in the ass to try to do a project 12,000 miles and 16 hours of time-zone difference away.
Really? Gee, I wonder where I was working, then? I sure thought it was IT.
Well, really, we called it "data processing" then, but same thing.
I think you need to read the references.
Frankly, the notion that the linear no-threshold model would hold is, biologically, the more extrodinary claim: it's tantamount to suggesting that there is no biological mechanism that deals with radiation damage in naturally-occuring dose rates. Non-radiological toxins that don't occur in nature are another thing, but it's damned hard to find one that doesn't have a close biological analogue.
Yeah, leave it to you white people to take a perfectly good Big Medicine plant and turn it into a habit.
No, no, I don't think you're wrong at all -- I suspect that you're on the right track. The underlying mechanism seems to be activation of a cell-repair (and probably an accelerated cell-death) mechanism that either uses some kind of error-correction to clean up ionization damage, or that causes a cell to "suicide" if it's irrepairable. (Now, kids, don't try this at home -- it's been 20 years since medical school and I was just auditing anyway.) Your dose of a couple of centriGray probably gets the mechanism up and going like a fire alarm, while 300 mR/yr is more like a faintly irritating dripping faucet -- but enough to keep the repair mechanism awake.
Sorry, make that two quibbles: Cohen's study was overall incidence of cancer, not just melanoma -- which we have an unfortunate lot of up here, due to high UV rates and an outdoors lifestyle.
Which proves hormesis can only go so far.
But it wouldn't surprise me to find that skin cancer versus lifetime UV exposure isn't linear no-threshold either.
Yes, exactly. I was just hoping to karma whore for a giggle or two.
One quibble: hormesis is observed epidemiologically in populations that are exposed to relatively high background exposures without the 'priming' dose characteristic. I believe the experiments you're talking about have a relatively higher dose rate, if only because you could die of boredom waiting for the effects of 300 mR/year to show up in a population of experimental subjects, but that's fuzzy memory and I'd love to be corrected.
A vent fan isn't the only thing needed, at least in some of the amelioration attempts. As well as the vent fan, some places out here have had to have the basement specially sealed, and in a few cases they've had to actually have hot stuff excavated from around the house and replaced with clean sand and concrete. The price of this is in the thousands or tens of thousands.
You bet, although it'll be short (a Google for radiation, epidemiology, and "Bernard Cohen" will get you more, as will "radiation" and "hormesis".
Idaho State has a site on radiaiton effects and radiation protection. They've also got some stuff on depleted uranium that looks good, although I've not read it in detail.
There is a journal devoted to low-dose radiation studies.
This page has about a zillion references. (Some of them disagree, by the way. This makes the page very useful.)
This paper has a nice discussion of hormesis and the linear no-threshold model.
BELLE Online is a good source in general. Also, the names "Bernard Cohen" and "Edward Calabrese" show up quite often.
One point about the "suspiciously low" cancer rates that I didn't say is that the rates in, eg, Cohen's epi paper are smoking-adjusted.
There are many possible reasons for this; maybe Colorado just has a better public health system or healthier lifestyles.
Read the sources before you respond to them. It makes you look less foolish later.
It is quite ironic that people like you often call themselves "conservatives", but then want to subject the US population to historically unprecedented exposures to largely unstudied chemicals and radiation.
I beg your pardon: all I said about "conservative" was that EPA claimed the linear no-threshold model they used was "conservative". Maybe you should read my article before you respond to it. As far as "historically unprecedented" exposure to radiation, you're making, by accident, my point for me: in the case of low doses of radiation, the attempt to eliminate exposure to low doses of radiation is what's unprecedented.
The only news is that ignorant politicians with a corporatist agenda use such obscure scientific tidbits out of context to argue that pollution is harmless.
Your compaint about taking things out of context, after this note, is quite amusing indeed.
It's sorta off topic, but what Freddie baby actually said is "Was mich nicht umbringt, macht mich noch staerker". The important verb "umbringen" means "to kill, to murder, to liquidate" and it's got sort of the sense of "bringing down" that we'd have if we spoke of bringing down a deer with hounds.
So it's really something like "That which can't catch me and kill me makes me even faster and stronger."