bidi is not everything. If you have a 33k modem connection to a 256-node beowulf cluster, do you claim it's useless? AGP cards have pretty beefy serial processing chips, that can be programmed with any, generic tasks just like CPUs, and for some of these tasks they will suck a big time (but still work) and for some they will rule (stuff like lots of similar rather simple calculations on lots and lots of data - they are unbeatable.) Statistics, rendering, filtering, encoding/decoding, all such stuff is really fast. Now the downstream is pretty slow so it hurts that -very- simple calculations can't be done en masse (the GPU can do them great but they get stuck at sending them back to the PC), and hard calculations with lots of decision-making are better handled in the CPU but there is a class of tasks where the GPU is unbeatable.
I don't know of any non-gfx cards that would use the CPU but there was a C compiler released that would use the GPU instead of CPU for your generic computations (instead of 3d gfx) and for certain kinds of calculations/programs it would be equivalent of 10GHZ P4 class CPU in the means of speed. Look up archives of Slashdot for it.
calling bluff, you can't quad KOTR. Hard-coded limitation. (who would want to wait through the goddamn 3-minute video 4 times? A single attack taking over 10 minutes to display?) You can quad Behemoth Zero though, and you can double KOTR (and if you get it to split you can double it twice a round, with 2 characters, but that's about it, and getting it to split is nearly impossible too.)
The last battle was quite easy for me because the characters had lots of med-to-high power stuff accumulated. (Counter-attack)=(Slash-All), )(Counter-Attack)=(4x-cut), and for one enemy blow I was doing 8 or so cuts.
BTW, ever played against weaker enemies with (added effect)=(Hades) ? Whoa, that's fun: mini, frog, sleep, curse, confuse and generally -everything- Hades -might- cast when summoned gets cast on the enemy you hit, at no MP cost. Add slash-all and see 5 enemies turning into sleeping frogs in one hit:) Not too good against bosses but FUN in normal battles:)
Well, 10^5 or 10^7 is very little considering that a (any) mutation happens once in 1000 specimens or so. Therefore two unrelated mutations in one individual in about 1,000,000, three in 10^9 etc... Here we're getting a combination of of about 6-7 totally unrelated and completely unprofitable when taken alone mutations. So the chance of a single random wasp mutating into a zombie driver wasp would be like 10^20 or so. Add to that the element of dumb luck (zombie driver wasp isn't killed by a bird or doesn't freeze on a cold day before giving offspring), and you suddenly see how amazingly rare this is. Building the first cell-like organism basing on DNA that was randomly assembled in the ancient "life soup" was even less likely, but back then you had whole planet covered with oceans filled with this soup and a thousands of random DNA strings assembling inside every single drop of water from these oceans. Wasps certainly outnumber humans on Earth, but still the population size, even when multiplied by several millions of years doesn't really make it close to the probability figure. Unless of course my "spherical cow" approximations are way off.
Wl, the first phase is obvious. It's the second phase that seems damn improbable. A wasp mutates to inject neurotoxin into roach's ass. No effect. A wasp mutates to inject neurotoxin into roach's leg. No effect. (a hundred more variations till the wasp injects neurotoxin into the brain) A wasp mutates to inject neurotoxin into roach's brain. Roach dies. A wasp mutates to inject neurotoxin into roach's brain, other area. Roach dies. A wasp mutates to inject neurotoxin into roach's brain, other area. Roach dies. A wasp mutates to inject neurotoxin into roach's brain, other area. Roach goes on a wild rampage and kills the wasp. A wasp mutates to inject neurotoxin into roach's brain, other area. Roach flips and goes numb. A wasp mutates to inject neurotoxin into roach's brain, other area. Roach breaks into spasms. (a thousand more variations till the wasp gets to the right point of the roach brain, with invariable repeatablity) A wasp mutates to inject neurotoxin into roach's brain, changing roach into a zombie. Then lays eggs and a bird eats the roach. A wasp mutates to inject neurotoxin into roach's brain, changing roach into a zombie. Then rides the roach around in random direction. A wasp mutates to inject neurotoxin into roach's brain, changing roach into a zombie. Then rides the roach towards the Sun. A wasp mutates to inject neurotoxin into roach's brain, changing roach into a zombie. Then rides the roach out in the open. (another several thousands of variations where the wasp rides the roach in random unprofitable direction) A wasp mutates to inject neurotoxin into roach's brain, changing roach into a zombie. Then rides the roach towards he nest. The roach dies halfway to the nest. (and a new wasp must evolve life-sustaining additions to the neurotoxin and injecting them in the right place of roach brain)
All the above are pointless from evolutionary point of view, and (as we know) evolution doesn't take long strides through unprofitable behaviours until it reaches some "higher plan", a profitable sophisticated behaviour. There's no evolutionary advancement from a wasp that doesn't inject any neurotoxin and one that injects the neurotoxin and drives the roach in a circle. Only fully developed set of behaviors, from the initial paralysing to settling the roach in the nest and laying the egg is evolutionarily profitable for the species. And the combinatory explosion resulting from all the possible UNPROFITABLE behaviours between plain "kills with poison" and "drives to nest" make me sometimes really doubt plain evolution (though I discard ID as explaination. I just assume "unexplained".) It's just that the change wouldn't be evolutionary but revolutionary - the difference between one and the other behaviour is TOO big and anything inbetween doesn't make sense, so how did the jump happen?
Showers show stock quotes. Mirrors with tickers instead of stickers. Jacuzzi with hotline to Yakuza. Stock of news-toilet-paper. You will know that MSFT dropped by 0.02 points but you won't know where your towel is. Ah, the modern world where even sanitation devices can drive you to insanity.
Well, both ways. The application of the "ns" extension is the same in both cases: to separate entries/functions from the standard namespace. So while most likely the w3c suffix means "namespace", and the prefix in Mozilla originates from NetScape, the conclusion would be Mozilla should abandon its Netscapish NS prefix and replace it with something like... NameSpace, short: NS?
Especially that current standards are ridden with it:) Yeppers, w3c-blessed DOM standard has most functions duplicated with "NS" version, like createAttribute and createAttributeNS.
Well, possibly it was too alarmist. As for me, 34 flaws are few. I'd expect more like 80-200. In two years. Even 50 flaws on the day of release, as results of known bugs found thorough internal tests, "we found it in the ready product. Correcting this would be too expensive, they aren't important enough." - fine. But 34 flaws in 20 days since release, many reported by outside entities - that's some outrage. Errors are unavoidable. You will make them, no matter what. You can only do your best to detect and fix them. Intel failed at the "error detection" part all the way, and they refuse to do the fixing part. And with 30 serious flaws in two weeks, what's the chance that one REALLY critical will be found in the next months? The problem isn't the "34" but the "20".
There are bugs and bugs. If you browse bugzilla.mozilla.org, you'll find minor bugs untouched for years. Nobody bothers, nobody worries. They are mapped. they aren't in the way, they aren't dangerous. Hundreds of them. But if somebody finds a security-related bug in "stable" version, what an outrage!
Would you rather fly a plane running software that was very thoroughly tested but not fixed afterwards, having the bugreport list in hand, knowing well which option is broken, which ones to avoid and which ones work unreliably? Sure, you need to think twice before pressing a button, and consult the errata sheet, but it's there for you. Or a plane that runs software that passed all the basic alpha tests and all the obvious bugs were fixed, but you have no idea what bugs lurk anywhere deeper? You just sat in the pilot chair and during the first 15 minutes you discovered 6 critical bugs. You write them down below 9 others written down by another would-be pilot who sat there before you but gave up.
...is to yield all the way. Pay compensation for losses in any desired amount. Submit themselves to FBI for violating DMCA and paying whatever fines it would cost them. Take the direst consequences against employees involved (at least officially. What they really do with them behind the scenes is a different matter.)
The more damage they take, the better for them, as long as they don't fight back, just swallow it. Take all the blame and "redeem" it. They CAN afford it.
And then use their own court case as example/precedent in any lawsuit they start. Small losses? Insignificantly small violation? So what? We paid. You must pay too.
I am torn between not liking the film maker for trying to find something to sue for and not liking the MPAA because they're cockroaches
I'm not. One thing if generic law forbids you something you want to do, no harm done, everyone does it and so on. You get a good chance you won't get caught, you game the system, sometimes you lose, more often you win. The law is just there as the weapon of your greedy opponents. But this one is a deliberate trap, Blatantly obvious like huge stinking beartrap in the middle of the sidewalk. With a tasty morsel inside but so obvious, that only a cretin would try to pick it. And MPAA pushes its dirty paws in the very middle. It's not about gaming the system, luck or not. Not about protecting your rights or profits. Not about greed. It's a fucking deliberate revenge, about showing the true face of the bastards and letting them taste their own poison. Unlike the assholes with "submarine patents" or obnoxious copyright holders, this guy didn't do this for profit. Most likely if he found his movie on a warez server, he would just grin. He did it just to purposedly turn MPAA's weapon against them. They created a monster called DMCA, which nobody really can control. Now he gamed the system enough so now he can direct that monster against them.
In short... MPAA got hacked. An elaborate hack, mix of social engineering, legaleese and media. I love that guy.
Heh. Nice one. Unfortunately there is a reason why the upgrade costs 1500 Euro: the software is far beyond reach of a single programmer or a small group of freelancers in the means of complexity, and it's niche enough that getting support of wider audience is nearly impossible. Besides, it just shows that it was written by a varied group of programmers, with most skilled ones assigned to more difficult tasks, weaker ones working on easier parts. The core computational part is a marvelous heavy wizardry, It has a small share of bugs, but they are catchable and avoidable, and besides, given difficulty of the problems I find it amazing that it contains so few and so minor ones. I doubt if I could do it myself. Very unlikely I could do it better. Then there are some features that seem like the users and programmers agreed nobody needs them, but the manager insisted so programmers wrote them so they are there but you're strongly discouraged from using them. And there's the basic stuff. Like the user interface, the system interaction, screen redrawing and such. Easy stuff. And it sucks. A big time. The easy part was given to novice programmers and they screwed it up totally. Don't try to swithch to Firefox while long computations are running. They will finish and they will be there, but the moment you restore focus to the window, the program will reset them. A dialog window gets open. Click anywhere not on a button and it will be taken as "cancel". Middle-click on a text entry box, and you'll be taken to some essential submenu not accessible by other ways. Acknowledge the fact you aborted computations by clicking OK four times in four requesters stating the same thing in a row. I wish I had access to the sources of that program. The core is good, strong and out of my grasp, but it begs for a good UI.
With regard to your 'dual core' argument. It is true that each core is the same, but the memory controller will be more complex (among other things) so your still assuming more complexity.
Yes, that's why the "1.5". It is there, it is complex, but it's not nearly as complex as each of the cores. Of course dual core will have more bugs than single core, but not twice as much.
At 65 nm tunneling electrons and short gate effects can cause otherwise functional blocks of logic to malfunction Still they are a probability-based stuff that should be catchable by tests. In modern technology a big percent (30-90%, varies over time and technology) of manufactured chips is faulty and discarded because they don't pass the tests. Of course some faults will slip through the tests, but I honestly doubt they would make it into erratas, because they are "instance" bugs, characteristic to given single chip (not series, not revision, I mean physical device - two PCs with two identical CPUs may behave differently because one of them has given bug and the other doesn't.) They result from impurities of material, glitches and wear of equipment, just plain dumb random distribution of atoms (luck) etc, but they are usually unique - a chip with given bug is manufactured once. The way of dealing with them if they are located is extremally simple: the bug test is added to the suite of tests, and any CPU failing the test is discarded, together with all the other faulty ones - so no reason for "won't fix" flag in errata! Unless of course given condition occurs in a bigger percent of the CPUs (like, 10%?) and Intel decides it's cheaper to keep selling broken chips and note that "it happens" than to discard them and sell only flawless units.
4x would be a good upper bound on the number of bugs a chip would have with twice the complexity and half the feature sizes of another chip. In one hand - the theoretical minimal radius of turn of a car is square root of sum of the distance between the axis and half the width of an axis, squared. (real definition from some book). Of course all cars can turn front wheels 90 degrees and pivot on one back wheel in place, yeah. Strongly theoretical boundary;)
In the other hand: Not really. Halving feature sizes doesn't double the amount of problems - they are already working on threshold of failure. It's more like halving the thickness of beams of a bridge, corresponding to probability the bridge collapses. A chip of given design, in given technology, with paths half as thick, will not have twice the bugs. It will be completely FUBAR. That's why the whole "new technology" together with algorithms of safe laying tracks at reliable non-tunnelling distances, methodology of finding faults etc - it may be less reliable, or more reliable - but not related in terms of reliablity to the old one. It just carries its own, completely new range of problems - and errors. Not related to core design too.
So why I still say 4x is bullshit? Getting good Bohr Bugs you can put in errata from the technology problems is very unlikely. If a problem related to technology persists to happen in one place, it will appear just the same in a thousand other places, because designs get recycled and there's really little unique stuff that could develop unique problems. A 32-bit register in silicon is just the same 32-bit register thorough the whole chip, no matter what it does, and if a fault sets bit 18 of one of the 32-bit registers at random in given core, it will tend to do so thorough all the 32-bit registers in the chip - not necessarily the same instance, but the same series. If one byte of cache fails, any byte can fail just the same way. Not an acceptable bug, must be fixed. So either the bug is too common to be ignored (multiple locations in the same CPU) or too uncommon to be noticed (obscure unique feature, obscure conditions and one in a million produced chips has it.)
To be "erratable" the technology-related bug must happen in unique place so
> While software bugs have (almost) always a solution,
Start thinking outside the Open Source box.;) I have some essential custom proprietary software at work, and it is ridden by bugs. At times working with it is like an arcade game: save, restore, try again. I know I could fix at least some of these bugs myself. But there's no source code. Bummer. Save, restore, try again, wait for update, spend 1500 Euro to buy it. Just to find all the old bugs are still there and new ones got introduced with new features.
Broken closed-source software leaves you in just as bad situation as broken hardware.
Therefore its expected that a chip fabricated on a substrate whose minimum feature sizes are half those of the other chip and whose complexity is double the other chip would have 4x the errata items of the other chip.
Complexity of the CPU contributes some to the amount of bugs - more project work = more bugs, though only in cases of introducing new algorithms, not in case of adding "more of the same" - dual core CPU is NOT supposed to have twice as many bugs as single-core counterpart, because the two cores are identical, contain the same flaws as the single core, and new ones are introduced only by the extra glue logic that makes it "dual". Twice the complexity usually means twice the number of gates, not twice the difficulty of design - stuff like cache memory swallows a major part of available space but 64KB of cache is associated with the same number of bugs as 4MB of it. So not x2 by complexity. At most x1.5 or so.
And thet errors are not manufacturing flaws, they are design flaws / software (VHDL) bugs. If I write a program twice as long as original and save it to a harddrive of double the capacity, am I expected to have four times as many bugs? The new technology has its own share of problems but they are to be caught before releasing the chip from the factory, and chip that has a technology-related fault is just faulty and should be replaced. It has nothing to do with what appears in errata.
So - the new CPU can have more bugs than the old one. But not four times as many!
Roland Piquepaille's article not linked to his blog outside of the "credit link"? I wonder, did Roland make a mistake or maybe Scuttlemonkey read my post?
Also, while Galileo receivers in general may be more accurate than, say, the GPS receiver in your PDA, high-grade GPS receivers used in military and commerical research applications can get centimeter or finer resolution
Over what period of time?
Commercial Galileo appliances are to perform the 1m measurements within one readout, almost instantly. And personally I've talked with someone who uses pro GPS for civilian mapmaking (while I was installing the dedicated GPS data processing software for them...) and it seems GPS can give you milimeters of accuracy thanks to repeating measurements over a couple of hours and averaging the results. They were taking about 3-6h (depending on desired accuracy, usually ~1cm) on their equipment to collect the data of a single measurement point from one receiver and then quite some time for crunching the collected results in the PC.
I don't know, maybe the situation has changed since then (some 5 years ago) but generally seems that for military use GPS in current shape can't get much better. You can't afford a hour of averaging readouts to guide a missile. You can drastically improve the measurement by reading more satellites at the same time and averaging few last readouts, but soon the improvement quality drops and time needed for it grows. With addition to better devices you need better satellites with more precise signal.
How precise can you get by averaging hours of measurements of Galileo reader?
I didn't RTFA. I don't think I need. All I needed to see is "Linux", "Microsoft", "patches", "legacy systems". With emphasis on the last one. Take my three legacy systems: Mom's Pentium MMX 166 webbrowsing machine, my 486 firewall and my work machine, P2 300, 256M RAM. Or something around these lines, somewhere up to 64MB RAM... WHAT systems run on these machines? Mom's computer runs Win98. Dumbed down interface plus low system requirements. (Sorry: Easy, Lightweight, Stable, pick any two.) My job machine runs NT. It could run 2k but it would slow down so much that the it would risk stalling machine it drives. My 486 runs Debian. Now which one is most secure? Seems the 486 Debian box, firewall with automatic security updates. Update quality/Legacyness ratio: very high. Neither 98 nor NT are supported anymore. No security updates for them at all. Zero divided by old/medium equipment.
What kind of "legacy" hardware do you need to benefit from the "quality patches" issued by Microsoft? How soon your current hardware will become "legacy", your OS "unsupported", your software "obsolete"? In my case the firewall fulfills its role at 100% efficiency, running some extra services, allowing remote login, being rock-stable and secure. NT quality: 90%. Behind dedicated firewall, running antivirus, crashes less than once a month, provides all I need. Win 98: some 60%. Still somewhat slow, security in hands of Firefox, antivirus and the 486 firewall, crashes on regular basis despite clean system.
Legacy systems are dead for Microsoft. Talking about quality patches for them is laughable.
Why not just stop the hipocrisy and plainly create a new section: "Slashvertisment"? Subscribed users would be able to disable it in profiles, the rest would have to play with Greasemonkey pretty hard to switch them off and it would be certainly more efficient than (already adblocked) banner ads. Make it clear a given story IS sponsored, IS biased, aims at promoting given product and was submitted for profit. I'd swallow it.
Just replace the article text. Leave the attribution and attribution link (under the nickname, rarely followed by users) but rewrite the summary and skip the middleman, linking directly to the article. So Roland posts in his blog a piece of some other site and links to it. Write "[Roland] wrote about [this cool site], which is about..." instead of "[Roland] wrote: I've put a short blurb [in my blog] about that cool site..." He gets the nickname attribution link. Not all the slashdot effect hits.
Slashdot Mirror
bidi is not everything. If you have a 33k modem connection to a 256-node beowulf cluster, do you claim it's useless? AGP cards have pretty beefy serial processing chips, that can be programmed with any, generic tasks just like CPUs, and for some of these tasks they will suck a big time (but still work) and for some they will rule (stuff like lots of similar rather simple calculations on lots and lots of data - they are unbeatable.) Statistics, rendering, filtering, encoding/decoding, all such stuff is really fast. Now the downstream is pretty slow so it hurts that -very- simple calculations can't be done en masse (the GPU can do them great but they get stuck at sending them back to the PC), and hard calculations with lots of decision-making are better handled in the CPU but there is a class of tasks where the GPU is unbeatable.
There were quite a few non-gfx devices using the VESA (Video Electronics Standards Association) slots.
It's been 1 minute since you last successfully posted a comment. Yeah, I know, fuck you slashdot.
I don't know of any non-gfx cards that would use the CPU but there was a C compiler released that would use the GPU instead of CPU for your generic computations (instead of 3d gfx) and for certain kinds of calculations/programs it would be equivalent of 10GHZ P4 class CPU in the means of speed. Look up archives of Slashdot for it.
calling bluff, you can't quad KOTR. Hard-coded limitation. (who would want to wait through the goddamn 3-minute video 4 times? A single attack taking over 10 minutes to display?)
:) Not too good against bosses but FUN in normal battles :)
You can quad Behemoth Zero though, and you can double KOTR (and if you get it to split you can double it twice a round, with 2 characters, but that's about it, and getting it to split is nearly impossible too.)
The last battle was quite easy for me because the characters had lots of med-to-high power stuff accumulated.
(Counter-attack)=(Slash-All), )(Counter-Attack)=(4x-cut), and for one enemy blow I was doing 8 or so cuts.
BTW, ever played against weaker enemies with (added effect)=(Hades) ? Whoa, that's fun: mini, frog, sleep, curse, confuse and generally -everything- Hades -might- cast when summoned gets cast on the enemy you hit, at no MP cost. Add slash-all and see 5 enemies turning into sleeping frogs in one hit
Well, 10^5 or 10^7 is very little considering that a (any) mutation happens once in 1000 specimens or so.
Therefore two unrelated mutations in one individual in about 1,000,000, three in 10^9 etc... Here we're getting a combination of of about 6-7 totally unrelated and completely unprofitable when taken alone mutations. So the chance of a single random wasp mutating into a zombie driver wasp would be like 10^20 or so. Add to that the element of dumb luck (zombie driver wasp isn't killed by a bird or doesn't freeze on a cold day before giving offspring), and you suddenly see how amazingly rare this is.
Building the first cell-like organism basing on DNA that was randomly assembled in the ancient "life soup" was even less likely, but back then you had whole planet covered with oceans filled with this soup and a thousands of random DNA strings assembling inside every single drop of water from these oceans. Wasps certainly outnumber humans on Earth, but still the population size, even when multiplied by several millions of years doesn't really make it close to the probability figure. Unless of course my "spherical cow" approximations are way off.
And to think people get so worked up about Jesus being born in a stable.
Wl, the first phase is obvious. It's the second phase that seems damn improbable.
A wasp mutates to inject neurotoxin into roach's ass. No effect.
A wasp mutates to inject neurotoxin into roach's leg. No effect.
(a hundred more variations till the wasp injects neurotoxin into the brain)
A wasp mutates to inject neurotoxin into roach's brain. Roach dies.
A wasp mutates to inject neurotoxin into roach's brain, other area. Roach dies.
A wasp mutates to inject neurotoxin into roach's brain, other area. Roach dies.
A wasp mutates to inject neurotoxin into roach's brain, other area. Roach goes on a wild rampage and kills the wasp.
A wasp mutates to inject neurotoxin into roach's brain, other area. Roach flips and goes numb.
A wasp mutates to inject neurotoxin into roach's brain, other area. Roach breaks into spasms.
(a thousand more variations till the wasp gets to the right point of the roach brain, with invariable repeatablity)
A wasp mutates to inject neurotoxin into roach's brain, changing roach into a zombie. Then lays eggs and a bird eats the roach.
A wasp mutates to inject neurotoxin into roach's brain, changing roach into a zombie. Then rides the roach around in random direction.
A wasp mutates to inject neurotoxin into roach's brain, changing roach into a zombie. Then rides the roach towards the Sun.
A wasp mutates to inject neurotoxin into roach's brain, changing roach into a zombie. Then rides the roach out in the open.
(another several thousands of variations where the wasp rides the roach in random unprofitable direction)
A wasp mutates to inject neurotoxin into roach's brain, changing roach into a zombie. Then rides the roach towards he nest. The roach dies halfway to the nest.
(and a new wasp must evolve life-sustaining additions to the neurotoxin and injecting them in the right place of roach brain)
All the above are pointless from evolutionary point of view, and (as we know) evolution doesn't take long strides through unprofitable behaviours until it reaches some "higher plan", a profitable sophisticated behaviour. There's no evolutionary advancement from a wasp that doesn't inject any neurotoxin and one that injects the neurotoxin and drives the roach in a circle. Only fully developed set of behaviors, from the initial paralysing to settling the roach in the nest and laying the egg is evolutionarily profitable for the species. And the combinatory explosion resulting from all the possible UNPROFITABLE behaviours between plain "kills with poison" and "drives to nest" make me sometimes really doubt plain evolution (though I discard ID as explaination. I just assume "unexplained".) It's just that the change wouldn't be evolutionary but revolutionary - the difference between one and the other behaviour is TOO big and anything inbetween doesn't make sense, so how did the jump happen?
Showers show stock quotes.
Mirrors with tickers instead of stickers.
Jacuzzi with hotline to Yakuza.
Stock of news-toilet-paper.
You will know that MSFT dropped by 0.02 points but you won't know where your towel is.
Ah, the modern world where even sanitation devices can drive you to insanity.
Well, both ways. The application of the "ns" extension is the same in both cases: to separate entries/functions from the standard namespace. So while most likely the w3c suffix means "namespace", and the prefix in Mozilla originates from NetScape, the conclusion would be Mozilla should abandon its Netscapish NS prefix and replace it with something like... NameSpace, short: NS?
Simply change the way you read it...
Especially that current standards are ridden with it :)
Yeppers, w3c-blessed DOM standard has most functions duplicated with
"NS" version, like createAttribute and createAttributeNS.
Well, possibly it was too alarmist. As for me, 34 flaws are few. I'd expect more like 80-200. In two years. Even 50 flaws on the day of release, as results of known bugs found thorough internal tests, "we found it in the ready product. Correcting this would be too expensive, they aren't important enough." - fine. But 34 flaws in 20 days since release, many reported by outside entities - that's some outrage. Errors are unavoidable. You will make them, no matter what. You can only do your best to detect and fix them. Intel failed at the "error detection" part all the way, and they refuse to do the fixing part. And with 30 serious flaws in two weeks, what's the chance that one REALLY critical will be found in the next months? The problem isn't the "34" but the "20".
There are bugs and bugs. If you browse bugzilla.mozilla.org, you'll find minor bugs untouched for years. Nobody bothers, nobody worries. They are mapped. they aren't in the way, they aren't dangerous. Hundreds of them. But if somebody finds a security-related bug in "stable" version, what an outrage!
Would you rather fly a plane running software that was very thoroughly tested but not fixed afterwards, having the bugreport list in hand, knowing well which option is broken, which ones to avoid and which ones work unreliably? Sure, you need to think twice before pressing a button, and consult the errata sheet, but it's there for you.
Or a plane that runs software that passed all the basic alpha tests and all the obvious bugs were fixed, but you have no idea what bugs lurk anywhere deeper? You just sat in the pilot chair and during the first 15 minutes you discovered 6 critical bugs. You write them down below 9 others written down by another would-be pilot who sat there before you but gave up.
How much is that list to grow yet?
...is to yield all the way.
Pay compensation for losses in any desired amount.
Submit themselves to FBI for violating DMCA and paying whatever fines it would cost them.
Take the direst consequences against employees involved (at least officially. What they really do with them behind the scenes is a different matter.)
The more damage they take, the better for them, as long as they don't fight back, just swallow it. Take all the blame and "redeem" it. They CAN afford it.
And then use their own court case as example/precedent in any lawsuit they start. Small losses? Insignificantly small violation? So what? We paid. You must pay too.
I am torn between not liking the film maker for trying to find something to sue for and not liking the MPAA because they're cockroaches
I'm not.
One thing if generic law forbids you something you want to do, no harm done, everyone does it and so on. You get a good chance you won't get caught, you game the system, sometimes you lose, more often you win. The law is just there as the weapon of your greedy opponents.
But this one is a deliberate trap, Blatantly obvious like huge stinking beartrap in the middle of the sidewalk. With a tasty morsel inside but so obvious, that only a cretin would try to pick it. And MPAA pushes its dirty paws in the very middle. It's not about gaming the system, luck or not. Not about protecting your rights or profits. Not about greed. It's a fucking deliberate revenge, about showing the true face of the bastards and letting them taste their own poison. Unlike the assholes with "submarine patents" or obnoxious copyright holders, this guy didn't do this for profit. Most likely if he found his movie on a warez server, he would just grin. He did it just to purposedly turn MPAA's weapon against them. They created a monster called DMCA, which nobody really can control. Now he gamed the system enough so now he can direct that monster against them.
In short... MPAA got hacked. An elaborate hack, mix of social engineering, legaleese and media. I love that guy.
Seems like the Police giving drugs to a drug dealer so he could sell them to kids, some of which they could then arrest. :D
Nice
Heh. Nice one. Unfortunately there is a reason why the upgrade costs 1500 Euro: the software is far beyond reach of a single programmer or a small group of freelancers in the means of complexity, and it's niche enough that getting support of wider audience is nearly impossible. Besides, it just shows that it was written by a varied group of programmers, with most skilled ones assigned to more difficult tasks, weaker ones working on easier parts. The core computational part is a marvelous heavy wizardry, It has a small share of bugs, but they are catchable and avoidable, and besides, given difficulty of the problems I find it amazing that it contains so few and so minor ones. I doubt if I could do it myself. Very unlikely I could do it better. Then there are some features that seem like the users and programmers agreed nobody needs them, but the manager insisted so programmers wrote them so they are there but you're strongly discouraged from using them. And there's the basic stuff. Like the user interface, the system interaction, screen redrawing and such. Easy stuff. And it sucks. A big time. The easy part was given to novice programmers and they screwed it up totally. Don't try to swithch to Firefox while long computations are running. They will finish and they will be there, but the moment you restore focus to the window, the program will reset them. A dialog window gets open. Click anywhere not on a button and it will be taken as "cancel". Middle-click on a text entry box, and you'll be taken to some essential submenu not accessible by other ways. Acknowledge the fact you aborted computations by clicking OK four times in four requesters stating the same thing in a row.
I wish I had access to the sources of that program. The core is good, strong and out of my grasp, but it begs for a good UI.
With regard to your 'dual core' argument. It is true that each core is the same, but the memory controller will be more complex (among other things) so your still assuming more complexity.
;)
Yes, that's why the "1.5". It is there, it is complex, but it's not nearly as complex as each of the cores. Of course dual core will have more bugs than single core, but not twice as much.
At 65 nm tunneling electrons and short gate effects can cause otherwise functional blocks of logic to malfunction
Still they are a probability-based stuff that should be catchable by tests. In modern technology a big percent (30-90%, varies over time and technology) of manufactured chips is faulty and discarded because they don't pass the tests. Of course some faults will slip through the tests, but I honestly doubt they would make it into erratas, because they are "instance" bugs, characteristic to given single chip (not series, not revision, I mean physical device - two PCs with two identical CPUs may behave differently because one of them has given bug and the other doesn't.) They result from impurities of material, glitches and wear of equipment, just plain dumb random distribution of atoms (luck) etc, but they are usually unique - a chip with given bug is manufactured once. The way of dealing with them if they are located is extremally simple: the bug test is added to the suite of tests, and any CPU failing the test is discarded, together with all the other faulty ones - so no reason for "won't fix" flag in errata!
Unless of course given condition occurs in a bigger percent of the CPUs (like, 10%?) and Intel decides it's cheaper to keep selling broken chips and note that "it happens" than to discard them and sell only flawless units.
4x would be a good upper bound on the number of bugs a chip would have with twice the complexity and half the feature sizes of another chip.
In one hand - the theoretical minimal radius of turn of a car is square root of sum of the distance between the axis and half the width of an axis, squared. (real definition from some book). Of course all cars can turn front wheels 90 degrees and pivot on one back wheel in place, yeah. Strongly theoretical boundary
In the other hand: Not really. Halving feature sizes doesn't double the amount of problems - they are already working on threshold of failure. It's more like halving the thickness of beams of a bridge, corresponding to probability the bridge collapses. A chip of given design, in given technology, with paths half as thick, will not have twice the bugs. It will be completely FUBAR. That's why the whole "new technology" together with algorithms of safe laying tracks at reliable non-tunnelling distances, methodology of finding faults etc - it may be less reliable, or more reliable - but not related in terms of reliablity to the old one. It just carries its own, completely new range of problems - and errors. Not related to core design too.
So why I still say 4x is bullshit?
Getting good Bohr Bugs you can put in errata from the technology problems is very unlikely. If a problem related to technology persists to happen in one place, it will appear just the same in a thousand other places, because designs get recycled and there's really little unique stuff that could develop unique problems. A 32-bit register in silicon is just the same 32-bit register thorough the whole chip, no matter what it does, and if a fault sets bit 18 of one of the 32-bit registers at random in given core, it will tend to do so thorough all the 32-bit registers in the chip - not necessarily the same instance, but the same series. If one byte of cache fails, any byte can fail just the same way. Not an acceptable bug, must be fixed. So either the bug is too common to be ignored (multiple locations in the same CPU) or too uncommon to be noticed (obscure unique feature, obscure conditions and one in a million produced chips has it.)
To be "erratable" the technology-related bug must happen in unique place so
> While software bugs have (almost) always a solution,
;)
Start thinking outside the Open Source box.
I have some essential custom proprietary software at work, and it is ridden by bugs. At times working with it is like an arcade game: save, restore, try again. I know I could fix at least some of these bugs myself. But there's no source code. Bummer. Save, restore, try again, wait for update, spend 1500 Euro to buy it. Just to find all the old bugs are still there and new ones got introduced with new features.
Broken closed-source software leaves you in just as bad situation as broken hardware.
Therefore its expected that a chip fabricated on a substrate whose minimum feature sizes are half those of the other chip and whose complexity is double the other chip would have 4x the errata items of the other chip.
Complexity of the CPU contributes some to the amount of bugs - more project work = more bugs, though only in cases of introducing new algorithms, not in case of adding "more of the same" - dual core CPU is NOT supposed to have twice as many bugs as single-core counterpart, because the two cores are identical, contain the same flaws as the single core, and new ones are introduced only by the extra glue logic that makes it "dual". Twice the complexity usually means twice the number of gates, not twice the difficulty of design - stuff like cache memory swallows a major part of available space but 64KB of cache is associated with the same number of bugs as 4MB of it. So not x2 by complexity. At most x1.5 or so.
And thet errors are not manufacturing flaws, they are design flaws / software (VHDL) bugs. If I write a program twice as long as original and save it to a harddrive of double the capacity, am I expected to have four times as many bugs? The new technology has its own share of problems but they are to be caught before releasing the chip from the factory, and chip that has a technology-related fault is just faulty and should be replaced. It has nothing to do with what appears in errata.
So - the new CPU can have more bugs than the old one. But not four times as many!
Just make sure everyone in the prison knows what the guy did, and the sentence will get extended with lifetime requirement to use diapers.
Roland Piquepaille's article not linked to his blog outside of the "credit link"? I wonder, did Roland make a mistake or maybe Scuttlemonkey read my post?
Also, while Galileo receivers in general may be more accurate than, say, the GPS receiver in your PDA, high-grade GPS receivers used in military and commerical research applications can get centimeter or finer resolution
Over what period of time?
Commercial Galileo appliances are to perform the 1m measurements within one readout, almost instantly. And personally I've talked with someone who uses pro GPS for civilian mapmaking (while I was installing the dedicated GPS data processing software for them...) and it seems GPS can give you milimeters of accuracy thanks to repeating measurements over a couple of hours and averaging the results. They were taking about 3-6h (depending on desired accuracy, usually ~1cm) on their equipment to collect the data of a single measurement point from one receiver and then quite some time for crunching the collected results in the PC.
I don't know, maybe the situation has changed since then (some 5 years ago) but generally seems that for military use GPS in current shape can't get much better. You can't afford a hour of averaging readouts to guide a missile. You can drastically improve the measurement by reading more satellites at the same time and averaging few last readouts, but soon the improvement quality drops and time needed for it grows. With addition to better devices you need better satellites with more precise signal.
How precise can you get by averaging hours of measurements of Galileo reader?
I didn't RTFA. I don't think I need. All I needed to see is "Linux", "Microsoft", "patches", "legacy systems". With emphasis on the last one.
Take my three legacy systems: Mom's Pentium MMX 166 webbrowsing machine, my 486 firewall and my work machine, P2 300, 256M RAM. Or something around these lines, somewhere up to 64MB RAM... WHAT systems run on these machines?
Mom's computer runs Win98. Dumbed down interface plus low system requirements. (Sorry: Easy, Lightweight, Stable, pick any two.) My job machine runs NT. It could run 2k but it would slow down so much that the it would risk stalling machine it drives. My 486 runs Debian.
Now which one is most secure? Seems the 486 Debian box, firewall with automatic security updates. Update quality/Legacyness ratio: very high. Neither 98 nor NT are supported anymore. No security updates for them at all. Zero divided by old/medium equipment.
What kind of "legacy" hardware do you need to benefit from the "quality patches" issued by Microsoft? How soon your current hardware will become "legacy", your OS "unsupported", your software "obsolete"? In my case the firewall fulfills its role at 100% efficiency, running some extra services, allowing remote login, being rock-stable and secure. NT quality: 90%. Behind dedicated firewall, running antivirus, crashes less than once a month, provides all I need. Win 98: some 60%. Still somewhat slow, security in hands of Firefox, antivirus and the 486 firewall, crashes on regular basis despite clean system.
Legacy systems are dead for Microsoft. Talking about quality patches for them is laughable.
Why not just stop the hipocrisy and plainly create a new section: "Slashvertisment"? Subscribed users would be able to disable it in profiles, the rest would have to play with Greasemonkey pretty hard to switch them off and it would be certainly more efficient than (already adblocked) banner ads. Make it clear a given story IS sponsored, IS biased, aims at promoting given product and was submitted for profit. I'd swallow it.
npwm.
Just replace the article text. Leave the attribution and attribution link (under the nickname, rarely followed by users) but rewrite the summary and skip the middleman, linking directly to the article. So Roland posts in his blog a piece of some other site and links to it. Write "[Roland] wrote about [this cool site], which is about..." instead of "[Roland] wrote: I've put a short blurb [in my blog] about that cool site..." He gets the nickname attribution link. Not all the slashdot effect hits.