In most systems I've encountered where it is practical to codify the formal requirements up front either does so, or the design of the product has taken it so deeply into account it becomes easy to forget the product implements just one approach to a larger and messier problem.
The systems where it is possible to write down the formal requirements, but where the architecture doesn't reflect this, tend to the ones where the formal requirements are worthless in practice. There is more to a product than satisfying a formal integrity constraints: it must also conduct a practical computation to get there. When the number of approaches to practical computation explodes, you typically end up with a system where the user is given a lot of control over how the computation unfolds.
Databases are hideously complex in this aspect. Performance characteristics have many dimensions: memory use, disk throughput, disk seek rate, transaction throughput (amortized, worst case, under normal load, under extreme load, for common queries, uncommon queries, on past hardware, on present hardware, on possible future hardware, WRT serialization bottlenecks, etc.) How soon do you need the integrity checks to take place? That can impact a multitude of metrics.
If you had HAL, and you told HAL all this information about your desired performance profile, as well as your formal integrity requirements, then I'm sure HAL could automatically make all these choices for you. But if you already had HAL, no one would be working at that level anymore. Yes, it's an intellectual salt mine, but until computation begins to grow faster than the size of data warehouses / complexity of the queries desired, it's probably going to stay that way.
Now if we had a quantum dot where we could multiplex a terabyte of data into the state of a single quantum particle, almost any query you desired to run could be returned in constant time. In this regime, you might be able to get away with a pure formal specificiation, without any decoherence of your Platonic ideal from requirements in the time domain.
I can't believe how few posts here grasp the central legal issue. The case against CC is a side show. Up until the officer verified that nothing was stolen, he probably had the law, or at least the sympathies of the judiciary, behind him. *After* he verified that no crime had originally been committed, it was his snotty-nosed follow-up charge of impeding police procedure that is going to get him into some deep legal hot water, because at that point in time he suspected no crime at all, other than the refusal to show a driver's license, which it's doubtful he had any right to demand, and furthermore, the officer neglected to ask for other information he was entitled to that would have enabled him to conduct those duties without needing the DL in the first place.
This is a case of an officer issuing a "screw you" charge against a citizen, at a point in time where he is suspected of no original crime, for sticking up legal rights he actually holds.
What is it about this that's hard to grasp? For that matter, why don't the police just get it over and done with by charging the constitution for obstruction of law-enforcement activities. It absolutely does obstruct law-enforcement. There's no question about that whatsoever. It turns out that law-enforcement is not the highest ideal of constitutional society, a mundane and disagreeable detail which the police occasionally forget.
Imagine you go to a bank to protest a $50 banking fee that was charged by mistake. The bank manager agrees that the $50 charge was in error. Then you return home and check your bank statement electronically and it now shows a $500 fee for "irregular statement review request".
The cop had an opportunity to drop the matter once the reality of no original crime was apparent to all involved. He didn't. He chose to go snotty. That's the issue here. Not Circuit City groping people's bagaloons. Like, duh.
Regardless of what JJ accomplishes with this movie, it will take him his entire future life to redeem himself for the two hours I spent stuck on a airplane watching Regarding Henry. That movie was a study in cynicism: his own, regarding his audience. It does sound like a good formula, though. A mature, polished JJ pitted against the cultural icon of dorky fan-bases: Trekkies. Can the good gollum prevail over the bad gollum? "We hates them, we hates them. Filthy Trekkies." No, I think not.
I thought I might repurpose two paragraphs from Wikipedia, under the taking-is-giving license:
Shylock refuses Bassanio's offer, despite Bassanio increasing the repayment to 6000 ducats (twice the specified loan). He demands the pound of flesh from Antonio. The Duke, wishing to save Antonio but unwilling to set a dangerous legal precedent of nullifying a contract, refers the case to Balthasar, a young male "doctor of the law" who is actually Portia in disguise, with "his" lawyer's clerk, who is Nerissa in disguise. Portia asks Shylock to show mercy in a famous speech (The quality of mercy is not strained--IV,i,185), but Shylock refuses. Thus the court allows Shylock to extract the pound of flesh.
At the very moment Shylock is about to cut Antonio with his knife, Portia points out a flaw in the contract (see Quibble (plot device)). The bond only allows Shylock to remove the flesh, not blood, of Antonio. If Shylock were to shed any drop of Antonio's blood in doing so, his "lands and goods" will be forfeited under Venetian laws. You can't strip a license without also taking the blood. With the license stripped, the code is rendered dead to the purpose under debate. See Quibble (I'm-smarter-than-you device).
If this is truly the case, come work for me for no paycheck.
Card carrying member of the all-or-nothing crowd? Some of us still value shades of grey. But I can suspend that momentarily.
While we're in the process of doing a root-canal on human dignity, what is it about human nature that connects such a worthless rejoinder directly to the kneecap? A lot of people in history have worked for no paycheck, and there's a name for that, although it's hard to get creative work out of people under those conditions.
Fast forward history from the primal snarl, the dilemma arises where one profit-oriented sweatshop industrialist finds himself undercut by an even more ruthless profit-oriented sweatshop industrialist. He needs an edge. Maybe even an idea. But where to get such a thing? He certainly can't produce one himself, that might cut into his ego-maintenance time. No, his only recourse is to shackle himself a golden goose, one of those notorious flakes who has not fully and properly internalized the value that life is all about money.
Or if not money, honor. For example, if I work a machine shop and lose my hand, I get compensation. If I work for the military and I lose my life, I get a flag. The military has a similar never-ending connundrum: how to recruit without paying people commensurate to the risk and sacrifice involved. Amp up the service and loyalty and nation-under-threat rhetoric. It works for business too. Just amp up the "it's all about money", or bare a fang while leering "come work for free", and play it up as a fair rejoinder. The rhetoric "it's all about money" does not speak to money, it speaks to subordination, and primal greeds satisfied by one person controlling another. Any person who goes around reminding others of their primal needs is all about control. I once witnesses a person purporting to be an angel investor who came into the meeting room and filled an entire white-board with the two words: FEAR and GREED. That was on there the whole time he spoke, and another week afterwards. We were too intimidated to erase it.
DLink 530TX: Via Rhine chipset. 530TX+: Realtek 8139. Apparently the + sign meant "more sucky". After validating that one has worked well, and then ending up with the other, and not having it work well (it crippled a basement closet NFS server), I can get a little choked about these small distinctions for a very long time. Bought more expensive Intel fxp cards for a long time afterward.
Here's the thing. If I order a 530TX from my favorite rock-solid discount house, they will fill the order with a 530TX+ without even asking, because of the suggestion conveyed by the product name of "small highly-compatible improvement".
It's a flagrant violation of the social contract.
However, in the case of Seagate, I don't see the performance delta between these drives as being much to cry over. With the 8139 I was seeing a 50% packet loss on certain protocols. Wake me up when the AAK is reporting a 50% seek failure rate on selected workloads.
Look at the craziness. Seagate might be making these drives on two different continents, and obviously that could involve some significant differences in component supply chains. In some cases, it might turn out that the drives produced by one supply chain are a bit more jiggy than the other supply chain. For reasons no-one fully understands, the exact composition of the bearing lubricant and bearing steel reduce spindle vibration by 1% So you tune the BIOSes a little bit different to bring out the best in both production series. Both series meet the specified performance target. No animals were harmed. Yet the small performance profile difference incites a wave of entitlement lust throughout the page-view-for-pennies-fan-boy-wanker-cult, who quite blithely accept the presence of their cycle-stealing Realtek engineered-for-mediocrity embedded networking chip. You go, Realtek. The wanker boys have spotted a mouse.
The GPL is "more free" than the BSD because the only thing it restricts is the kind of restrictions that can be put on the code. Its restrictions are much like the restrictions found in the bill of rights, they ensure no one can take your freedom. Now tell me, is the US more or less free with the 1st amendment?
Well, now, that's a truly fabulous slight-of-mind. Kudos. But tell me, is 1st amendment a reference to first post? Under this theory of law, you really want to get there first.
The GPL actually functions like a 1st amendment which states that America grants certain rights to Americans (some in the guise of restrictions on restrictions), and we're going to bomb any country which doesn't follow suit back to the stone age.
Tell me, wouldn't the 1st amendment be better if it was more honest about the carpet bombing of contrary views back to the stone age? Subject only to available funds?
I admire the GPL empire, but personally I choose not to live there. I think the GPL is the right choice for systems and the wrong choice for technologies. If I invented a new network protocol (Internet 3, since Internet 2 is already registered) I would license the implementation under BSD, and the compliance suite under GPL. A protocol is worthless if people don't steal enough of the original code to make everything work together. Likewise, the protocol is worthless if people Balkanize compliance with the edge cases. I see two different purposes, and two different licences with respect to those purposes, neither of which involves any recursive viral calculus to comprehend.
The most valuable the employee, the less that employee functions machine-like. If you job function is perfectly suited to machine-like productivity, soon a machine will do it. The deep reason corporations dislike the natural human rhythm of attention and distraction is that work tasks demanding bursts of intellectually non-sustainable heavy-lifting can't be automated.
Any job where you can precisely quantify the waste of time, is itself a waste of time.
Much deep intellectual work resembles throwing pasta at the wall. Boil vigorously after second coffee, then huck the starch at the nearest paint. While you wait to see what sticks, you play a little Freecell.
What I'd like to see is a study on the magnitude of learned helplessness promoting "non-productive" stress outlets. Specically those persistent, magic, inexplicable, and unpreventable random font and layout changes that define Microsoft Word. Innocently press the tab key, your page layout morphs through a wormhole. The truth of the matter is that I ever find myself in an employment situation that requires use of Microsoft Word, it will also require heavy use of Freecell, yet I've never seen that term reported in any Microsoft-backed TCO.
I have to say what makes me laugh about this entire thread, is that people complain it doesn't "just work" and then go out and buy themselves a "lappy". A lappy with wifi. Buying a lappy is like marrying five people simultaneously and not being able to divorce any of them. Wifi is like also having a mistress on the side, in case the harem is too harmonious. So it becomes Ubuntu's problem to function as the go-between and keep the dysfunctional union ticking along nicely. Oh joy to Ubuntu.
We should study wifi client deployment for deep insight into multi-dimensional packing problems. Any three distinct wifi clients wishing to connect to a single AP will force the AP to be configured into the least secure possible mode. What should we call that? The Hamming effect?
If I filled my desktop PC with two gallons of epoxy resin, then I could also join the just-work killer bee swarm. By definition, once I've deprived myself of the latitude to fix it myself, it instantly becomes the problem of the upsteam provider, and since I'm not otherwise busy with fixing anything (the brokenness is welded, baked, and boxed), I've nothing else to do but complain about it.
Maybe Ubuntu needs to emit a more forceful installer message: Sorry, lappy dude, your APIC is fucked. Patch available at ebay.com. Batteries not included.
I read TFA and MOFA (many other fine articles) on this subject, and they all stink to some degree, though most not so much as this one, which I could barely even keep straight. The formula is simple: what did they find, what did they observe, what are the working conclusions, and what assumptions does this challenge. Not one of the articles I found explains how the ape diet differs from the common ape ancestor. More roughage. More roughage than what? I guess they don't know, since they haven't discovered those teeth yet, but nevertheless, the consensus is firm, and unspoken to the unwashed reader.
And somehow, if you find eight teeth in Ethiopia, it instantly proves that no other segment of the ancestor lineage drifted into Eurasia on a small vacation for any length of time. If I spot in Spokane an RV with a Massachusetts license plate heading west on the I90, then instantly I conclude "they be taking the direct route to Seattle". Not possible they visited Florida. No chance. But apparently, you can fill in the gap from 25 Ma to 10 Ma just like that on the basis of eight teeth.
I'll come back in a year or two when the dust settles and the reporting improves.
As was explained in A Beautiful Mind, there's no point everyone hitting up the hot blond who is going to reject you anyway, when your failure to achieve your first objective then compromises your chances to succeed with a second objective. Does anyone here think that a total anhilation of Open XML is in the cards? The point of that scene is not that it's a particularly good expression of Nash equilibrium, but that even a blond can understand it, which indirectly serves as a good example of settling for second best, when best is not in the cards to begin with.
Maybe a better plan would be to hit on the brunette. Microsoft has a long and sordid legal history concerning the display of scare boxes. This gratuitious bonking behaviour goes back at least as far as DRDOS. I've lost count of the number of cases since. It's not like they have much cred to say "we don't do that" or even "we don't do that anymore". Microsoft is like a drug addict where everyone has come to accept that "not any more" translates to "not since the last time".
How about as a condition for accepting Open XML, Microsoft is required to provide BSD licenced source code that scans an Open XML document for every possible defect that any version of any Microsoft software might display to the user with even the faintest whiff that anything is not entirely right with the OOXML document being processed.
Before Open XML can be regarded as an open standard in any significant sense, Microsoft needs to be deprived of their privileged position with regard to validation of Open XML and ability to taint the minds of the users with fear, uncertainty, and doubt.
The complexity of the open source validation suite would itself raise eyebrows concerning the purported openness of the Open XML standard. It is quite likely that Microsoft would refuse to go along with this proposal. That wouldn't reflect well on their intent either.
I'm putting this forward as an example of negotiating by contract: the open source implementation of the validation suite would serve as the contract by which Microsoft agrees on what manipulations of an Open XML might potentially sterilize your children, and what will certainly not, nor be purported as such, except to Microsoft's corporate liability. Personally, I could live with that.
Did I just eat too many syrup coated waffles? He's telling me the rounding error is 10^-4 or 10^-5 on values with more trailing nines than I can count between sugary blinks. Not long ago I came across a slide presentation from a HEP lab concerning C++0x with a slide proclaiming that decimal floating point in hardware was the wave of the future. Now while I don't see any numerical advantage to this change, it will probably reduce the number of floating point gurus who gouge their own eyes out after rubbing shoulders with the ULP-retarded hoi pollio.
The people responding here do realize, almost by definition, that "refocussing" Linux involves a hypothetical mode of intervention that stands at complete contrast to everything Linux has so far represented. There is no vagrant pool of talent at this level, nor is there a mechanism to confine this pool of talent to pie-in-the-sky wishlish thinking.
As a point of reference, this text has been in the OpenBSD dhcpd man page for as long as I can remember:
We realize that it would be nice if one could send a SIGHUP to the server and have it reload the database. This is not technically impossible, but it would require a great deal of work, our resources are extremely limited, and they can be better spent elsewhere. So please don't complain about this on the mailing list unless you're prepared to fund a project to implement this feature, or prepared to do it yourself.
So far it has been extremely effective in scaring off any useful contribution. There are plenty of dirty jobs that need doing. Many small dirty jobs is worth more than a heaping serving of pie.
I can supply prior art on predictive Chinese input (pinyin/bopomofo to simplified/classical hanzi) dating back to the mid-eighties on a 4.77MHz PC, becoming quite sophisticated by the early 1990s: also hanzi back to pinyin (with disambiguation), pinyin to pinyin with correct tone markings, and romaji/hiragana to kanji. The first version involved a hardware expansion card to supply the Chinese character fonts, which were too large to handle in PC memory at the time. The version from the early 1990s was set up like a Soduko puzzle: you could constrain the character options for any syllable by any character property (pinyin syllable, tone, stroke count, principle radical, any arbitrary subset) and it would solve for the most likely hanzi sequence matching those constraints. It had a moderately large dictionary of common hanzi phrases, and a generative grammar to handle number-measure expressions. At the time I last worked on it, the statistical model wasn't fully fleshed out to the extent of a word model in a fully developed speech recognition system, but it was headed that direction. It was also specifically tuned to resolve the most common ambiguities present in standard pinyin transcription, with a couple of unsolvable problems left to squirt out the sides: most users created a custom annotation to distinguish ta=he from ta=she. After the first conversion, if the hanzi sequence was incorrect, the user could select any incorrect character and ask for the next most likely conversion (at isolated character or full phrase levels), select the correct character manually and have the rest of the phrase updated for congruence, or just eliminate that character as undesired, and have the conversion resolved with the addition of a new constraint: e.g. position three does not resolve to character X.
We never went to the next level with the negative inference that if the user conventionally writes ta=she as taa, the syllable ta becomes unlikely to resolve to the female pronoun, but even that refinement was obvious within the framework.
Not all of this was brought out to the user interface at the time. We were trying not to scare people with too many options.
A related problem was automatically recognizing the text sort of a document the user was composing, and constraining the word model accordingly. That's a very difficult problem. If the program adapts too much, it throws the user off of their regular expectations. There are some clear cut examples, however. A government statistician working on demographic models is rarely called upon to discuss pubic issues (but few spell checkers are configured to flag this embarrassment, as it's also a perfectly legal spelling).
Good textual discretion might be original. This patent isn't.
The same game plays out in the food processing industry where the loudness of a food item is defined by its salt content. Drunk people staggering homeward pay more attention to a loud selection on the Jukebox the same way they choose their late night snack selections: pretzels, pizza, poutine, peanuts. Common ingredient? Hint: it's not the letter p. Recently I tried a Greek ewe's chees, Myzithra, that caused my salt detectors to compress my sense of taste into a square wave. Would have been a toss-up against anchovie paste straight out of the tube.
The whole stupid FA never once gives the percentage improvement when exposed to a sunlight spectrum. Not cool. Not cool at all. I'm sure it was an easy oversight. Sunlight is an obscure point of reference in this debate. After all, sunlight is nowhere near as common as water or air.
If CVS isn't broken, I have a three-legged ladder I'd like to sell you. I'll even set it up in the parking lot and, with but a modest presence of mind, balance myself motionlessly on the very top step to prove how very not-broken it really is. On one foot. And I'll juggle, too.
CVS is what happens when you've roped yourself up into some high, awkward, inaccessible place, then you discover you brought along the wrong toolbox. Subversion is a fancy pair of vise-grips with rubber handles: doesn't hurt your hand so much when you have to grasp with extreme force the bolt head with no remaining flat edges, because you're too damn lazy to rope yourself back down to get the tool you should have used in the first place.
Yes, they need to block Firefox without users of other browsers noting the fact, or the fact that they blocked Firefox will become a talisman for anyone with a clue, regardless of what user agent they possess, decided that visiting somewhere else is a better use of their time and energy. Kind of like those strings of coloured plastic triangles they hang over used car lots. Obviously, there are plenty of people scraping change together who fail to heed the warning signs, but there are also many affluent people who value calm business relations who immediately head elsewhere.
The notion there is a social contract to imbibe the brain-washing impressions amuses me, but the lifelong pursuit of unnecessary wealth does eventually lead the unwary to think in these terms. Simple question: point at the moment in time where this social contract came into force. Even if money interests succeed in brainwashing the higher court, a group of people with even longer careers centered around never forgetting your premises (regardless of whether your strategy is to reveal or conceal), the decision won't state at what point between the first televised (or otherwise electronically mediated) ad impression and the present that this social contract gained binding force; it will merely be a verdict that money interests can operate as if that contract now exists, as expressed through interpretations of the DMCA and related legislation. If this comes to pass, it would be an interesting social experiment, as the brainwashing has so far always operated with the brainwashee of the belief that participation was voluntary. An explicit legal decision that the brainwashee can be punished for consuming the content without also consuming the ad impressions would partially break the spell. I have no doubt that the discipline of psychological engineering would prove up to the challenge within another generation or so: look what the dementors have already done to pop music.
Well, yes, your copy might not include the plot spoilers for Deathly Hallows or the latest exploit by Lindsay Lohan, which excludes all but the narrowest academic use. Once the cellulosic ethanol process is a bit more mature, we can recover some prime real estate where our public libraries used to reside. The children's books alone will fuel a Hummer H1 for almost a year.
I once calculated that a 3 gallon-per-minute shower involving a 40 degree C temp. boost (the intake temp. in the Canadian winter is not high) draws about 32kW (mostly extracted from the giant heat resevoire known as a hot water tank). If my math was correct, this car provides enough power to heat ten shower stalls on an instantaneous basis.
Lot's of people care about the difference between decimal SI and the binary expropriation of SI, and I'm a card carrying member of the MB=1024 society.
If I'm in a conversation with people who purport to know something (some level of computer industry insider) then they had better demonstrate two things: that binary units are assumed in such a context unless otherwise stated, and that enumeration begins with zero, or I'm likely to flip the bozo-bit on that person.
In a more rarified context, the other parties had better understand that short-circuit evaluation is assumed for predicate expressions (consider: if (p != NULL && *p == c) which is otherwise illegal), that semi-open intervals eliminate fence-post errors, and the only sane definition of r = x mod y for integer x and positive integer y satisfies the post-condition that r is on the interval [0..y), and that malloc(0) returns a valid pointer *distinct* from any other currently allocated memory block. Apparently, bozos have occassionally infiltrated the language standardization committee of Your Favorite Language(TM).
In a truly rarified context (perhaps myself only), it would also be understood that if you can rewrite the condition tree:
if (x == a) {
}
else
if (x == b) {
}
else
if (x == c) {
}
As the following:
if (x == a) {
}
if (x == b) {
}
if (x == c) {
}
Without changing the *logic* of your program, the original construct might be syntactically nested in the view of your compiler, but it sure as heck isn't semantically nested. If your compiler recognizes the predicates as fully disjoint (without side-effects), it probably rewrites the later as the former internally.
How much of that should I assume from the average visitor to Wikipedia?
In mixed contexts, which includes both insiders and outsiders, the safe bet is to be specific about the difference between MiB and MB. Wikipedia is written to the ultimate mixed crowd: any person on the planet who can puzzle out some text in one of the world's major languages. It's total arrogance to assume that internal industry norms should prevail in such a context. That said, I'm quite comfortable if the article states its conventions in a footnote, or parathetically in each instance, or just spells it out each time.
I have zero tolerance for making light of the distinction, wide latitude in how the distinction is honored.
Yes, as per usual, the tribalism reflex on this thread has shut down useful brain circuits in most of the posters. Gorged with tribalistic lust, the average post here seems to be able to consider only the issue of getting pwned, or the paranoid dichotomy between useful and secure, or the purportedly paltry size of the OpenBSD user base, or the irritating slogan at the top of the OpenBSD home page (which hardly negates their contributions to the security ecology since the inception of the project no matter how sour the partisan).
The useful insight here is that systrace implemented on top of clone is a relatively pathetic defense againt malicious adversaries. So be it. That's not the only scenario to consider. It could possibly be applied to remove setuid from a long running daemon process that only requires the secure call during startup, which could be arranged to occur with the single in single-user mode. It can be used as a research tool with an unfamiliar application to discover how many system calls it makes which require elevated security. It can be used to configure a systrace profile for an application that will eventually be ported to a secure, kernel-based implementation of syscall. Or it can be used as suggested, to help protect the sytem from unwitting blunders that might otherwise have remained possible.
Given a choice, I think the OpenBSD would prefer to see secure applications written to drop privs for the client threads. In general terms, applying systrace to impose security after-the-fact is doomed to fall far short of deploying an application architected from a security standpoint to begin with.
The BUGS entry at the bottom of the OpenBSD systrace man page makes it quite clear that, as presently implemented, this little piggy lives in a house of straw. Not the best foundation for retrofitting iron bars on the windows and doors, as the sysjail project has attempted to do.
Ah yes, the "hosts file" tweaker. Ever an important advertising demographic. It doesn't pay to piss people off who have any means of doing something about it. In all other cases, in inculcates learned helplessness, the wet-dream of pseudo-democracies everywhere. Can't uninstall or disable or live without Java? And the banks are involved? Ah yes, the wet-dream of monopolistic capitalism. Strange how many countries wake up on a map of France every morning.
You have a short little span of attention. When Intel first hit 60W with the original Pentium there was a huge outcry about its outrageous power consumption, and it hardly performed any better than a 100MHz 486, either. After a quick die shrink, the next version wasn't so bad. Now Intel sells the Core Duo at 65W as a major innovation in power management. After Intel's Prescott, it's almost impossible for anything else to look bad. But really, should a product that never deserved to be made in the first place define the frame of reference moving forward? If you factor the environment into the picture, a TDP of 35W would look far more responsible.
Slashdot Mirror
In most systems I've encountered where it is practical to codify the formal requirements up front either does so, or the design of the product has taken it so deeply into account it becomes easy to forget the product implements just one approach to a larger and messier problem.
The systems where it is possible to write down the formal requirements, but where the architecture doesn't reflect this, tend to the ones where the formal requirements are worthless in practice. There is more to a product than satisfying a formal integrity constraints: it must also conduct a practical computation to get there. When the number of approaches to practical computation explodes, you typically end up with a system where the user is given a lot of control over how the computation unfolds.
Databases are hideously complex in this aspect. Performance characteristics have many dimensions: memory use, disk throughput, disk seek rate, transaction throughput (amortized, worst case, under normal load, under extreme load, for common queries, uncommon queries, on past hardware, on present hardware, on possible future hardware, WRT serialization bottlenecks, etc.) How soon do you need the integrity checks to take place? That can impact a multitude of metrics.
If you had HAL, and you told HAL all this information about your desired performance profile, as well as your formal integrity requirements, then I'm sure HAL could automatically make all these choices for you. But if you already had HAL, no one would be working at that level anymore. Yes, it's an intellectual salt mine, but until computation begins to grow faster than the size of data warehouses / complexity of the queries desired, it's probably going to stay that way.
Now if we had a quantum dot where we could multiplex a terabyte of data into the state of a single quantum particle, almost any query you desired to run could be returned in constant time. In this regime, you might be able to get away with a pure formal specificiation, without any decoherence of your Platonic ideal from requirements in the time domain.
I can't believe how few posts here grasp the central legal issue. The case against CC is a side show. Up until the officer verified that nothing was stolen, he probably had the law, or at least the sympathies of the judiciary, behind him. *After* he verified that no crime had originally been committed, it was his snotty-nosed follow-up charge of impeding police procedure that is going to get him into some deep legal hot water, because at that point in time he suspected no crime at all, other than the refusal to show a driver's license, which it's doubtful he had any right to demand, and furthermore, the officer neglected to ask for other information he was entitled to that would have enabled him to conduct those duties without needing the DL in the first place.
This is a case of an officer issuing a "screw you" charge against a citizen, at a point in time where he is suspected of no original crime, for sticking up legal rights he actually holds.
What is it about this that's hard to grasp? For that matter, why don't the police just get it over and done with by charging the constitution for obstruction of law-enforcement activities. It absolutely does obstruct law-enforcement. There's no question about that whatsoever. It turns out that law-enforcement is not the highest ideal of constitutional society, a mundane and disagreeable detail which the police occasionally forget.
Imagine you go to a bank to protest a $50 banking fee that was charged by mistake. The bank manager agrees that the $50 charge was in error. Then you return home and check your bank statement electronically and it now shows a $500 fee for "irregular statement review request".
The cop had an opportunity to drop the matter once the reality of no original crime was apparent to all involved. He didn't. He chose to go snotty. That's the issue here. Not Circuit City groping people's bagaloons. Like, duh.
Regardless of what JJ accomplishes with this movie, it will take him his entire future life to redeem himself for the two hours I spent stuck on a airplane watching Regarding Henry. That movie was a study in cynicism: his own, regarding his audience. It does sound like a good formula, though. A mature, polished JJ pitted against the cultural icon of dorky fan-bases: Trekkies. Can the good gollum prevail over the bad gollum? "We hates them, we hates them. Filthy Trekkies." No, I think not.
At the very moment Shylock is about to cut Antonio with his knife, Portia points out a flaw in the contract (see Quibble (plot device)). The bond only allows Shylock to remove the flesh, not blood, of Antonio. If Shylock were to shed any drop of Antonio's blood in doing so, his "lands and goods" will be forfeited under Venetian laws. You can't strip a license without also taking the blood. With the license stripped, the code is rendered dead to the purpose under debate. See Quibble (I'm-smarter-than-you device).
If this is truly the case, come work for me for no paycheck.
Card carrying member of the all-or-nothing crowd? Some of us still value shades of grey. But I can suspend that momentarily.
While we're in the process of doing a root-canal on human dignity, what is it about human nature that connects such a worthless rejoinder directly to the kneecap? A lot of people in history have worked for no paycheck, and there's a name for that, although it's hard to get creative work out of people under those conditions.
Fast forward history from the primal snarl, the dilemma arises where one profit-oriented sweatshop industrialist finds himself undercut by an even more ruthless profit-oriented sweatshop industrialist. He needs an edge. Maybe even an idea. But where to get such a thing? He certainly can't produce one himself, that might cut into his ego-maintenance time. No, his only recourse is to shackle himself a golden goose, one of those notorious flakes who has not fully and properly internalized the value that life is all about money.
Or if not money, honor. For example, if I work a machine shop and lose my hand, I get compensation. If I work for the military and I lose my life, I get a flag. The military has a similar never-ending connundrum: how to recruit without paying people commensurate to the risk and sacrifice involved. Amp up the service and loyalty and nation-under-threat rhetoric. It works for business too. Just amp up the "it's all about money", or bare a fang while leering "come work for free", and play it up as a fair rejoinder. The rhetoric "it's all about money" does not speak to money, it speaks to subordination, and primal greeds satisfied by one person controlling another. Any person who goes around reminding others of their primal needs is all about control. I once witnesses a person purporting to be an angel investor who came into the meeting room and filled an entire white-board with the two words: FEAR and GREED. That was on there the whole time he spoke, and another week afterwards. We were too intimidated to erase it.
One does not cross paths with 13 standard deviations in an average universe.
http://math2.org/math/stat/distributions/z-dist.h
This calculator computes the area from -inf to 6.5 as 1.0000000016. Wow! Only half-way there and we've already spotted some dark matter.
Open Office also buckles under the strain.
=NORMDIST(7.96;0;1;1) reports
9.9999999999999900000000000000000000]E-001
=NORMDIST(8;0;1;1) reports
10.0000000000000000000000000000000000]E+000
I didn't think ordinary matter was supposed to get all the way to 10. Cool. We're now on the way to 11, boys and girls.
DLink 530TX: Via Rhine chipset. 530TX+: Realtek 8139. Apparently the + sign meant "more sucky". After validating that one has worked well, and then ending up with the other, and not having it work well (it crippled a basement closet NFS server), I can get a little choked about these small distinctions for a very long time. Bought more expensive Intel fxp cards for a long time afterward.
Here's the thing. If I order a 530TX from my favorite rock-solid discount house, they will fill the order with a 530TX+ without even asking, because of the suggestion conveyed by the product name of "small highly-compatible improvement".
It's a flagrant violation of the social contract.
However, in the case of Seagate, I don't see the performance delta between these drives as being much to cry over. With the 8139 I was seeing a 50% packet loss on certain protocols. Wake me up when the AAK is reporting a 50% seek failure rate on selected workloads.
Look at the craziness. Seagate might be making these drives on two different continents, and obviously that could involve some significant differences in component supply chains. In some cases, it might turn out that the drives produced by one supply chain are a bit more jiggy than the other supply chain. For reasons no-one fully understands, the exact composition of the bearing lubricant and bearing steel reduce spindle vibration by 1% So you tune the BIOSes a little bit different to bring out the best in both production series. Both series meet the specified performance target. No animals were harmed. Yet the small performance profile difference incites a wave of entitlement lust throughout the page-view-for-pennies-fan-boy-wanker-cult, who quite blithely accept the presence of their cycle-stealing Realtek engineered-for-mediocrity embedded networking chip. You go, Realtek. The wanker boys have spotted a mouse.
Well, now, that's a truly fabulous slight-of-mind. Kudos. But tell me, is 1st amendment a reference to first post? Under this theory of law, you really want to get there first.
The GPL actually functions like a 1st amendment which states that America grants certain rights to Americans (some in the guise of restrictions on restrictions), and we're going to bomb any country which doesn't follow suit back to the stone age.
Tell me, wouldn't the 1st amendment be better if it was more honest about the carpet bombing of contrary views back to the stone age? Subject only to available funds?
I admire the GPL empire, but personally I choose not to live there. I think the GPL is the right choice for systems and the wrong choice for technologies. If I invented a new network protocol (Internet 3, since Internet 2 is already registered) I would license the implementation under BSD, and the compliance suite under GPL. A protocol is worthless if people don't steal enough of the original code to make everything work together. Likewise, the protocol is worthless if people Balkanize compliance with the edge cases. I see two different purposes, and two different licences with respect to those purposes, neither of which involves any recursive viral calculus to comprehend.
The most valuable the employee, the less that employee functions machine-like. If you job function is perfectly suited to machine-like productivity, soon a machine will do it. The deep reason corporations dislike the natural human rhythm of attention and distraction is that work tasks demanding bursts of intellectually non-sustainable heavy-lifting can't be automated.
Any job where you can precisely quantify the waste of time, is itself a waste of time.
Much deep intellectual work resembles throwing pasta at the wall. Boil vigorously after second coffee, then huck the starch at the nearest paint. While you wait to see what sticks, you play a little Freecell.
What I'd like to see is a study on the magnitude of learned helplessness promoting "non-productive" stress outlets. Specically those persistent, magic, inexplicable, and unpreventable random font and layout changes that define Microsoft Word. Innocently press the tab key, your page layout morphs through a wormhole. The truth of the matter is that I ever find myself in an employment situation that requires use of Microsoft Word, it will also require heavy use of Freecell, yet I've never seen that term reported in any Microsoft-backed TCO.
I have to say what makes me laugh about this entire thread, is that people complain it doesn't "just work" and then go out and buy themselves a "lappy". A lappy with wifi. Buying a lappy is like marrying five people simultaneously and not being able to divorce any of them. Wifi is like also having a mistress on the side, in case the harem is too harmonious. So it becomes Ubuntu's problem to function as the go-between and keep the dysfunctional union ticking along nicely. Oh joy to Ubuntu.
We should study wifi client deployment for deep insight into multi-dimensional packing problems. Any three distinct wifi clients wishing to connect to a single AP will force the AP to be configured into the least secure possible mode. What should we call that? The Hamming effect?
If I filled my desktop PC with two gallons of epoxy resin, then I could also join the just-work killer bee swarm. By definition, once I've deprived myself of the latitude to fix it myself, it instantly becomes the problem of the upsteam provider, and since I'm not otherwise busy with fixing anything (the brokenness is welded, baked, and boxed), I've nothing else to do but complain about it.
Maybe Ubuntu needs to emit a more forceful installer message: Sorry, lappy dude, your APIC is fucked. Patch available at ebay.com. Batteries not included.
I read TFA and MOFA (many other fine articles) on this subject, and they all stink to some degree, though most not so much as this one, which I could barely even keep straight. The formula is simple: what did they find, what did they observe, what are the working conclusions, and what assumptions does this challenge. Not one of the articles I found explains how the ape diet differs from the common ape ancestor. More roughage. More roughage than what? I guess they don't know, since they haven't discovered those teeth yet, but nevertheless, the consensus is firm, and unspoken to the unwashed reader.
And somehow, if you find eight teeth in Ethiopia, it instantly proves that no other segment of the ancestor lineage drifted into Eurasia on a small vacation for any length of time. If I spot in Spokane an RV with a Massachusetts license plate heading west on the I90, then instantly I conclude "they be taking the direct route to Seattle". Not possible they visited Florida. No chance. But apparently, you can fill in the gap from 25 Ma to 10 Ma just like that on the basis of eight teeth.
I'll come back in a year or two when the dust settles and the reporting improves.
As was explained in A Beautiful Mind, there's no point everyone hitting up the hot blond who is going to reject you anyway, when your failure to achieve your first objective then compromises your chances to succeed with a second objective. Does anyone here think that a total anhilation of Open XML is in the cards? The point of that scene is not that it's a particularly good expression of Nash equilibrium, but that even a blond can understand it, which indirectly serves as a good example of settling for second best, when best is not in the cards to begin with.
Maybe a better plan would be to hit on the brunette. Microsoft has a long and sordid legal history concerning the display of scare boxes. This gratuitious bonking behaviour goes back at least as far as DRDOS. I've lost count of the number of cases since. It's not like they have much cred to say "we don't do that" or even "we don't do that anymore". Microsoft is like a drug addict where everyone has come to accept that "not any more" translates to "not since the last time".
How about as a condition for accepting Open XML, Microsoft is required to provide BSD licenced source code that scans an Open XML document for every possible defect that any version of any Microsoft software might display to the user with even the faintest whiff that anything is not entirely right with the OOXML document being processed.
Before Open XML can be regarded as an open standard in any significant sense, Microsoft needs to be deprived of their privileged position with regard to validation of Open XML and ability to taint the minds of the users with fear, uncertainty, and doubt.
The complexity of the open source validation suite would itself raise eyebrows concerning the purported openness of the Open XML standard. It is quite likely that Microsoft would refuse to go along with this proposal. That wouldn't reflect well on their intent either.
I'm putting this forward as an example of negotiating by contract: the open source implementation of the validation suite would serve as the contract by which Microsoft agrees on what manipulations of an Open XML might potentially sterilize your children, and what will certainly not, nor be purported as such, except to Microsoft's corporate liability. Personally, I could live with that.
Did I just eat too many syrup coated waffles? He's telling me the rounding error is 10^-4 or 10^-5 on values with more trailing nines than I can count between sugary blinks. Not long ago I came across a slide presentation from a HEP lab concerning C++0x with a slide proclaiming that decimal floating point in hardware was the wave of the future. Now while I don't see any numerical advantage to this change, it will probably reduce the number of floating point gurus who gouge their own eyes out after rubbing shoulders with the ULP-retarded hoi pollio.
The people responding here do realize, almost by definition, that "refocussing" Linux involves a hypothetical mode of intervention that stands at complete contrast to everything Linux has so far represented. There is no vagrant pool of talent at this level, nor is there a mechanism to confine this pool of talent to pie-in-the-sky wishlish thinking.
As a point of reference, this text has been in the OpenBSD dhcpd man page for as long as I can remember:
So far it has been extremely effective in scaring off any useful contribution. There are plenty of dirty jobs that need doing. Many small dirty jobs is worth more than a heaping serving of pie.
I can supply prior art on predictive Chinese input (pinyin/bopomofo to simplified/classical hanzi) dating back to the mid-eighties on a 4.77MHz PC, becoming quite sophisticated by the early 1990s: also hanzi back to pinyin (with disambiguation), pinyin to pinyin with correct tone markings, and romaji/hiragana to kanji. The first version involved a hardware expansion card to supply the Chinese character fonts, which were too large to handle in PC memory at the time. The version from the early 1990s was set up like a Soduko puzzle: you could constrain the character options for any syllable by any character property (pinyin syllable, tone, stroke count, principle radical, any arbitrary subset) and it would solve for the most likely hanzi sequence matching those constraints. It had a moderately large dictionary of common hanzi phrases, and a generative grammar to handle number-measure expressions. At the time I last worked on it, the statistical model wasn't fully fleshed out to the extent of a word model in a fully developed speech recognition system, but it was headed that direction. It was also specifically tuned to resolve the most common ambiguities present in standard pinyin transcription, with a couple of unsolvable problems left to squirt out the sides: most users created a custom annotation to distinguish ta=he from ta=she. After the first conversion, if the hanzi sequence was incorrect, the user could select any incorrect character and ask for the next most likely conversion (at isolated character or full phrase levels), select the correct character manually and have the rest of the phrase updated for congruence, or just eliminate that character as undesired, and have the conversion resolved with the addition of a new constraint: e.g. position three does not resolve to character X.
We never went to the next level with the negative inference that if the user conventionally writes ta=she as taa, the syllable ta becomes unlikely to resolve to the female pronoun, but even that refinement was obvious within the framework.
Not all of this was brought out to the user interface at the time. We were trying not to scare people with too many options.
A related problem was automatically recognizing the text sort of a document the user was composing, and constraining the word model accordingly. That's a very difficult problem. If the program adapts too much, it throws the user off of their regular expectations. There are some clear cut examples, however. A government statistician working on demographic models is rarely called upon to discuss pubic issues (but few spell checkers are configured to flag this embarrassment, as it's also a perfectly legal spelling).
Good textual discretion might be original. This patent isn't.
The same game plays out in the food processing industry where the loudness of a food item is defined by its salt content. Drunk people staggering homeward pay more attention to a loud selection on the Jukebox the same way they choose their late night snack selections: pretzels, pizza, poutine, peanuts. Common ingredient? Hint: it's not the letter p. Recently I tried a Greek ewe's chees, Myzithra, that caused my salt detectors to compress my sense of taste into a square wave. Would have been a toss-up against anchovie paste straight out of the tube.
The whole stupid FA never once gives the percentage improvement when exposed to a sunlight spectrum. Not cool. Not cool at all. I'm sure it was an easy oversight. Sunlight is an obscure point of reference in this debate. After all, sunlight is nowhere near as common as water or air.
If CVS isn't broken, I have a three-legged ladder I'd like to sell you. I'll even set it up in the parking lot and, with but a modest presence of mind, balance myself motionlessly on the very top step to prove how very not-broken it really is. On one foot. And I'll juggle, too.
CVS is what happens when you've roped yourself up into some high, awkward, inaccessible place, then you discover you brought along the wrong toolbox. Subversion is a fancy pair of vise-grips with rubber handles: doesn't hurt your hand so much when you have to grasp with extreme force the bolt head with no remaining flat edges, because you're too damn lazy to rope yourself back down to get the tool you should have used in the first place.
Yes, they need to block Firefox without users of other browsers noting the fact, or the fact that they blocked Firefox will become a talisman for anyone with a clue, regardless of what user agent they possess, decided that visiting somewhere else is a better use of their time and energy. Kind of like those strings of coloured plastic triangles they hang over used car lots. Obviously, there are plenty of people scraping change together who fail to heed the warning signs, but there are also many affluent people who value calm business relations who immediately head elsewhere.
The notion there is a social contract to imbibe the brain-washing impressions amuses me, but the lifelong pursuit of unnecessary wealth does eventually lead the unwary to think in these terms. Simple question: point at the moment in time where this social contract came into force. Even if money interests succeed in brainwashing the higher court, a group of people with even longer careers centered around never forgetting your premises (regardless of whether your strategy is to reveal or conceal), the decision won't state at what point between the first televised (or otherwise electronically mediated) ad impression and the present that this social contract gained binding force; it will merely be a verdict that money interests can operate as if that contract now exists, as expressed through interpretations of the DMCA and related legislation. If this comes to pass, it would be an interesting social experiment, as the brainwashing has so far always operated with the brainwashee of the belief that participation was voluntary. An explicit legal decision that the brainwashee can be punished for consuming the content without also consuming the ad impressions would partially break the spell. I have no doubt that the discipline of psychological engineering would prove up to the challenge within another generation or so: look what the dementors have already done to pop music.
Well, yes, your copy might not include the plot spoilers for Deathly Hallows or the latest exploit by Lindsay Lohan, which excludes all but the narrowest academic use. Once the cellulosic ethanol process is a bit more mature, we can recover some prime real estate where our public libraries used to reside. The children's books alone will fuel a Hummer H1 for almost a year.
I once calculated that a 3 gallon-per-minute shower involving a 40 degree C temp. boost (the intake temp. in the Canadian winter is not high) draws about 32kW (mostly extracted from the giant heat resevoire known as a hot water tank). If my math was correct, this car provides enough power to heat ten shower stalls on an instantaneous basis.
% 2F+gram+*+3+gallons+%2F+minute+*+1+kg%2Fliter+*+40 &btnG=Search&meta=
Hells bells, I got the same answer again:
http://www.google.ca/search?hl=en&q=4.184+joules+
Lot's of people care about the difference between decimal SI and the binary expropriation of SI, and I'm a card carrying member of the MB=1024 society.
If I'm in a conversation with people who purport to know something (some level of computer industry insider) then they had better demonstrate two things: that binary units are assumed in such a context unless otherwise stated, and that enumeration begins with zero, or I'm likely to flip the bozo-bit on that person.
In a more rarified context, the other parties had better understand that short-circuit evaluation is assumed for predicate expressions (consider: if (p != NULL && *p == c) which is otherwise illegal), that semi-open intervals eliminate fence-post errors, and the only sane definition of r = x mod y for integer x and positive integer y satisfies the post-condition that r is on the interval [0..y), and that malloc(0) returns a valid pointer *distinct* from any other currently allocated memory block. Apparently, bozos have occassionally infiltrated the language standardization committee of Your Favorite Language(TM).
In a truly rarified context (perhaps myself only), it would also be understood that if you can rewrite the condition tree:
if (x == a) {
}
else
if (x == b) {
}
else
if (x == c) {
}
As the following:
if (x == a) {
}
if (x == b) {
}
if (x == c) {
}
Without changing the *logic* of your program, the original construct might be syntactically nested in the view of your compiler, but it sure as heck isn't semantically nested. If your compiler recognizes the predicates as fully disjoint (without side-effects), it probably rewrites the later as the former internally.
How much of that should I assume from the average visitor to Wikipedia?
In mixed contexts, which includes both insiders and outsiders, the safe bet is to be specific about the difference between MiB and MB. Wikipedia is written to the ultimate mixed crowd: any person on the planet who can puzzle out some text in one of the world's major languages. It's total arrogance to assume that internal industry norms should prevail in such a context. That said, I'm quite comfortable if the article states its conventions in a footnote, or parathetically in each instance, or just spells it out each time.
I have zero tolerance for making light of the distinction, wide latitude in how the distinction is honored.
Yes, as per usual, the tribalism reflex on this thread has shut down useful brain circuits in most of the posters. Gorged with tribalistic lust, the average post here seems to be able to consider only the issue of getting pwned, or the paranoid dichotomy between useful and secure, or the purportedly paltry size of the OpenBSD user base, or the irritating slogan at the top of the OpenBSD home page (which hardly negates their contributions to the security ecology since the inception of the project no matter how sour the partisan).
The useful insight here is that systrace implemented on top of clone is a relatively pathetic defense againt malicious adversaries. So be it. That's not the only scenario to consider. It could possibly be applied to remove setuid from a long running daemon process that only requires the secure call during startup, which could be arranged to occur with the single in single-user mode. It can be used as a research tool with an unfamiliar application to discover how many system calls it makes which require elevated security. It can be used to configure a systrace profile for an application that will eventually be ported to a secure, kernel-based implementation of syscall. Or it can be used as suggested, to help protect the sytem from unwitting blunders that might otherwise have remained possible.
Given a choice, I think the OpenBSD would prefer to see secure applications written to drop privs for the client threads. In general terms, applying systrace to impose security after-the-fact is doomed to fall far short of deploying an application architected from a security standpoint to begin with.
The BUGS entry at the bottom of the OpenBSD systrace man page makes it quite clear that, as presently implemented, this little piggy lives in a house of straw. Not the best foundation for retrofitting iron bars on the windows and doors, as the sysjail project has attempted to do.
Ah yes, the "hosts file" tweaker. Ever an important advertising demographic. It doesn't pay to piss people off who have any means of doing something about it. In all other cases, in inculcates learned helplessness, the wet-dream of pseudo-democracies everywhere. Can't uninstall or disable or live without Java? And the banks are involved? Ah yes, the wet-dream of monopolistic capitalism. Strange how many countries wake up on a map of France every morning.
You have a short little span of attention. When Intel first hit 60W with the original Pentium there was a huge outcry about its outrageous power consumption, and it hardly performed any better than a 100MHz 486, either. After a quick die shrink, the next version wasn't so bad. Now Intel sells the Core Duo at 65W as a major innovation in power management. After Intel's Prescott, it's almost impossible for anything else to look bad. But really, should a product that never deserved to be made in the first place define the frame of reference moving forward? If you factor the environment into the picture, a TDP of 35W would look far more responsible.