My last experience with HP, for much longer yet, was helping a friend install an HP printer driver under Windows 2000 a couple of years back. The HP driver borked the OS so badly the only recourse was to boot the system in safe-mode. HP provided some software support to remove the pesky printer driver. I tried to run the HP uninstaller, but it reported:
Your screen resolution is insufficient to run the borked HP printer driver uninstall program.
I'm in safe-mode here, my options for screen resolution are 640x480 and 800x600. Neither of those modes were acceptable to the HP uninstall software. Quelle foresight.
OK, I'll report the issue to HP. Spent fifteen minutes recounted the sequence of events in an HP web support form. Press submit. What comes up?
404 not found
Click the "back" button. Nope, my entire problem report has funneled down the Mariana trench. Another fifteen minutes of my life hopelessly extinguished.
At that point, I did what all good techies do in that situation. I decided to hold a grudge. For a long, long, long time. The perfect storm: incompetence (in the printer driver itself), stupidity (an uninstaller that refused to run in safe-mode), and insolence (a web support form that vaporises my problem report into the Mariana trench).
What's not to like about the new HP? Utterly, utterly, utterly despicable. I couldn't have said it better myself.
The new layout just became active for my account and I can't stand it. Can't even find "new layout goes live" to properly grouse about it. When I enlarge the fonts as I like to do in Firefox, text is mashing over top of other text. I hate the first column. How do I get rid of the whole darn thing? And two of the sections in the first column refuse to remain closed. Light layout is no improvement. I want my serif font back, boxes that look like boxes, and a paucity of useless columns. Don't moderate me -1 troll, moderate me -10 sour beyond all recognition.
Ad skipping remains perfectly legal. No one is forced to own a television in the first place. I don't see the point in getting all hung up over removing the mind control substance from a meal that was never very nutricious in the first place. Once a month I go to my local video store, rent five movies for a week for $10. I can barely *find* five movies a month worth watching. With some help from my pickiest friends I have another five movies on deck.
Seems to be a common theme here: some of these might be a little depressing. There's the problem. Most people who moan about skipping the ads aren't prepared to go all the way.
The proponents of evolution must obey the laws of physics, while the proponents of intelligent design are no so constrained. You can't blame email for the crazy ideas, scams, belief systems foisted by one party upon another. It seems to be endemic wherever humans gather to communicate.
Email has no end of faults that could have been mitigated in some measure by a superior design. All the same, the battle has hardly been lost. Email bears no responsibility for the emergence of botfarms, it's just the unlucky target. If the argument is that we'll never eliminate botfarms, that the botfarms will always be with us, then I think we have far bigger fish to fry.
After we win the battle against the botfarms, perhaps our problems with email will no longer appear quite so dire.
Cities of significance emerged three or four thousand years ago. It wasn't until circa 1850 that sanitation and clean water were fully addressed, and even that victory has many loose ends remaining.
On my monitor which has never had good gamma at the black levels I didn't even *see* the black text on the green bar until I read your comment. I've tried no end of adjustments but eventually I realized that any site worth reading does create this problem to begin with.
Strangely, those of us who prefer it this way appear to be a very small minority. I once read an ergonomics guide that suggesting sitting far enough away from the screen to comfortably page down with your big toe. The comfortable majority of squiggy-lovers seem to be the same crowd who claim that reading long articles on screen is more tiring than printing them out. I'm planning to keep my crisp 19" monitors until a very large, affordable LCD panel comes along featuring 200dpi resolution or better, so I can finally use a slightly smaller font and still enjoy the well-resolved letter forms that make extending on-screen reading an effortless activity.
And yes, typing so fast that entire words go missing and inflections are mangled is an accurate reflection on what a post to slashdot is worth anymore. All praise the brown paper bag.
While I'm on the subject, one of the best typographic makeovers in my experience was when the Economist adopted their new typeface in the early 1990s IIRC which included a provision for small-caps to typeset increasingly prevalent accronyms without rendering the overall paragraph composition garish. But then, the Economist felt they had some worth saying, and they branded themselves more through the style of their prose (and its power to convey ideas in certain rhetorical modality which they favour) rather than the freshness of their masthead.
As a general rule, sans serif fonts are a triumph of style over substance. In a discussion oriented web site, the font should be chosen for readability over appearance. The great thing about Helvetica is that it makes your spelling mistakes 15% less perceptible. Huge appeal to the slashdot editors there. Not long ago the NYTimes adopted a misguided makeover incorporating bamboo-slat column widths crowding their primary asset--their journalism--into a claustrophobic prison window. I've cut my page views from daily to weekly as a direct result.
I didn't regard that post as a troll whatsoever. Heck, with the money he saves buying a generic domain name, he can afford to buy himself a nice Herman Miller Aeron chair.
As far as squatting goes, you can't maintain an empty ecological niche that has proven this profitable. Blaming the squatters is hugely misguided. It's like blaming teenagers for having sex. When a giant new chunk of realestate becomes available (e.g. wild west) you tend to end up with one of two models: land grab, or public auction. The public auction is hard to set up until a market already exists. And people heap just as much vitriole on the FCC for attempting this as they do on the squatters for exploiting what was given to them.
Traditionally the government ends up applying a property tax on assets of this nature. A mil rate assessment process is devised, and anyone setting up shop on a street corner in Manhattan had better plan to operate a viable business or find themselves taxed into bankruptcy.
Anyone care to hazard a guess whether the people presently screaming out about squatters will be first in line to yell their leather lungs out when the government comes along and established a domain-name mil rate assessment bureau?
I'm not sure I agree with recent FSF positions (haven't tracked them much recently), but I agree overall with the FSF taking the long view of free software. There are enormous latent risks that DRM or shifts in the IP landscape (patents) could poison the well ten or twenty years down the road, by which point the crucial battles have already been lost. It's easy to come off as radical crusaders fighting battles that won't play out over a span of decades. Our short little span of attention is our worst enemy in these matters. The fact that they are alone in their extreme urgency doesn't prove much directly: they might be equally alone in a correct analysis of the risks at hand. Just because Chicken Little is squawking, that doesn't mean the sky isn't falling. Glib comments about Chicken Little behaving like Chicken Little have add nothing of any use to the larger debate. My comments add nothing of any use, either, but at least I know the difference.
That smacks of the same cleverness that exempted light trucks and SUVs from mandated fuel economy guidelines, and we all know where that went: the automakers went to town pushing demand into a more profitable vehicle category simultaneously driving net fuel consumption upward. Not to worry. We've since sent a crack team of negotiators to the middle east to ensure our supply. A few of them come home in boxes. Small price to pay.
Big pharma is going to love these terms. Choice is a powerful weapon in highly contested spheres. Do you ever see a major sports league offer the team finishing first in the regular season the choice to pick their first round opponent of the other teams making the cut? Of course not. It would open up a gaping wound for manipulation. Side deals to rig which teams end up matched together.
I could see this working if the choice of patent extension were restricted to the cosmetic/vanity drugs (hair restoration creams, erectile aids, wrinkle creams, etc.) Applied to the only drug in existence that treats a certain form of cancer, or cures childhood malaria? That's giving up far too much.
Easy to offer that now, because few people are clever enough to recognize the end game. How many people saw the California fuel-economy act as a way to upsell consumers into a pricier vehicle class?
Twenty years later the legacy of that loophole has become painfully obvious.
I've read T-bomb's side of the issue whenever this debate has flared up, and I've never come away impressed with his ability to fully enter into the debate.
In effect, Linus is saying the past 20 years of work on object-oriented programming is misguided. I don't buy that.
The comments Linux offered up were from the perspective that there is more than one way to implement information hiding. T-bomb believes that only hardward enforced information-hiding (exploiting MMU protections) is valid. Many critical data structures in Linux enforce information hiding through a C language macro interface which encapsules all the extremely tricky logic that is terribly hard to get right.
Both forms of information-hiding break down if employed incorrectly. In the micro-kernel instance, the break-down is caught (usually) by the MMU which traps out one (or more) of the participants in the failed initiative. I fail to see how Minix can self-heal itself from incorrect program code. What it can do is resume or limp along more gracefully. The fix to the program logic waits upon a developer to come along and identify and correct the faulty program logic.
In Linux, mistakes in the use of the critical primitives will likely take down the entire kernel. It's well understood within the core community of kernel hackers what kinds of misuse or mistakes most often lead to this unfortunate result. If a chunk of code does not pass a visual sanity check, most likely it never becomes part of a sancitified kernel in the first place. The coders who consistently contribute code which does pass the visual sanity review tend to deeply understand the issues at stake. In practice, the dire consequence doesn't leak through the early stages of the development process nearly as often as T-bomb insinuates. I diagnosed one kernel hang myself concerning an embedded 2.4.19 ARM kernel where the semantics of an ARM timer register were subtly abused in concert with sleep/wake cycles. If this same programming mistake had been made in MINUX, it wouldn't have been of any benefit. We still could not ship the product to the end customer until fixing the mistake.
Linus further stated that there are development costs associated with the micro-kernel design, such as achieving the same degree of efficiency while copying stack frames from one protected address frame into another. This leaks out far more often than most people admit. fgetc() doesn't call the OS (as a true u-kern would) for every character processed. That would be too costly for a primitive operation at that scale. Instead the C library buffers character data, a small detail which adds much complexity to things like the C++ streams library. We had some hideous problems with a PHP module until we realized it was buffering old results from fstat() calls (even after closing and reopening the file IIRC). Both of these practices stem from trying to eliminate unnecessary context switches for performance critical operations.
u-kern APIs tend to involve more of this than monolithic kernels because the context switch is more frequent or more expensive. I think in the 3.51 NT u-kern (before they abandoned many of their u-kern principles) they went to a lot of trouble to batch up GDI calls when possible on the application side of the fence. This kind of thing sucks developer attention and can introduce strange artifacts such as the PHP artifact we observed: open file, write data to file, close file, fstat file, observe length is sill zero.
What Linus was pointing out about Linux kernel development is that once you've paid the price to establish good cooperative information-hiding through macro layers (and fully debuged the difficult semantics) in many ways it takes less effort to review the code for correct use of these abstraction layers than it does to deal with the ongoing headaches introduced by crossing MMU barriers for every trivial operation.
I happened to despise the cooperative tasking model from Windows 3.1 and OS 9. In t
I watched "Bowling for Columbine" the other night and "The Corporation" and then "The Civil War" (first episode). Despite all that stimulation, it wasn't until I read your post that the penny finally dropped.
In the American mind, slavery was not originally a race issue. The whites were the people who had guns, and the blacks were the people who didn't. Gun = slave owner, no gun = slave. Now if there were more guns in China, the Chinese could have themselves a Chinese civil war and 20 million Chinese could die in the cause of mutual freedom, if it takes the same 2% of the population as died in the American civil war to settle their small difference of opinion.
I never did figure out why Heston poo-pooed the idea that everyone in America having their own little lump of plutonium. If everyone having handguns is the great catalyst of social freedom, why wouldn't little lumps of plutonium work 1000 times better? I mean, we'd become so concerned about not pissing each other off, we might even turn on an evening news source with some content about the world, not just those stupid sound bites that even Bush can muster up.
Or maybe I'm thinking in the wrong direction. In America, with guns, it took four years for Americans to kill 600,000 other Americans. In Rwanda, with cheap machetes, it only took 100 days to settle 800,000 differences of opinion.
Wang made a PC clone with the 8086 for a while. It was lacking compatibility on a number of fronts, so it died of a not-so-slow terminal illness. It ran about 60% faster than an 8088 PC at the same rated clock speed.
It's amazing that IBM looked at the 8086 and decided it needed an additional whack to the kneecap. As if it wasn't hobbled badly enough already.
Here's one small defect that greatly enhanced the Propecia sales among a generation of early adoptors. The segment registers were designed with a four bit offset from the address registers, yielding a 1MB address space.
If that offset had been six bits instead, the machine would have had a 4MB address space. Each of the eight ISA slots could have had a dedicated 64K address mapped I/O space (consuming 512KB). A 4MB address space would have nicely bridged the evolution to 32-bit multitasking operating systems, which become viable somewhere between 4MB and 8MB.
That four bit offset cost the industry billions of dollars that could have been far better invested elsewhere. There was precious little advantage to a 16 byte segment alignment granularity that I ever noticed.
There was some added efficiency, but MOST of the gain was from the basic fact that a process shrink (and a core adjusted to make use of it) allowed higher frequencies.
This is one of the most bizarre intellectual shell games to ever make common currency. The reason this new rocket goes twice as fast as this old rocket is that we burn the fuel twice as quickly. All the rest of the expensive engineering was to ensure that the rocket doesn't blow itself up on the launch pad coping with all that thrust. But the real technology was burning the fuel twice as fast.
90% of a modern CPU design consists of circuit elements to conceal latency from the frequency obsessed execution units: the branch prediction unit, the parallel instruction decoder, the L1 cache, the L2 cache, and the out-of-order execution engine. It costs electrical power to feed all those circuits to hide those latencies. Which is why cycles/watt have increased far more slowly than cycles/second, despite the fact that power is reduced to a first order as a function of voltage squared.
Hey stop that! I trademarked that clue long before *you* got to it.
What's interesting is how the converse stupidity was established in the first place. Proponents of the RISC camp were so busy announcing their imminent victory they forget to extrapolate more than one or two die shrinks into the future. A design parameter that makes you pull your hair out today can evolve into a subtle advantage a few shrinks later.
I'd rip apart how x86 handles the flag register (partial register stalls and truly Byzantine retirement logic) long before I'd replace the x86 instruction encoding format (except perhaps the prefix bytes, which are super annoying for decode alignment). The stack based floating point register file and instruction set is one of the few x86 features that was beyond redemption at any scale (do I sound like Ralph Nader or what?)
The bottom line is that the flawed nature of the x86 instruction set had a much bigger impact on thermal performance (MIPS/watt) than it ever had on top speed. Now that thermal performance determines top speed, that old shoe is beginning to pinch in new places.
But, hey, the grammar Nazi Nazi is still a good old boy. Which makes me a grammar Nazi Nazi Nazi. I'm not telling where I live.
How did outsourcing get such a bad name in the first place? Reproduction requires outsourcing. I thought most people liked it that way. I mean, insourcing is a great just-in-time tactic for a few years, but it starts to seem, I don't know, a little bit xenophobic and self absorbed?
Very clever of the litcrit community to notice that with multitudinous overlaying of semantic fields over top of a single symbol set (words in a language) renders it impossible to construct any symbol sequence that is non-zero only once (in a single semantic field).
To a physicist, however, black means "absence of photons" and carries no connotation whatsoever of "melanin enhanced".
Those among us who get to the point where they refuse to peel away and discard these multitudinous (and often dubious) semantic overlays render themselves incapable of action. Water is also black if one refuses to filter out the crap. I for one refuse to drink from the Ganges of litcrit.
Walmart also happens to be an icon of the American economy during an era where the division of wealth in America returned to the same historical parameters as America experienced when Rockefeller was the world's richest tycoon, and his Standard Oil empire mired in controversy.
Yes, it was bound to happen, and so was the terrible technical analogy to spam gone wild. That doesn't make it true.
Spam has no permanent "revision history", nor a community consensus model to evolve its protections. The Wikipedia is fundamentally based on a centralized control model and I don't see that there is much organized PR flacks can do about this while that model remains in place. I've seen some extreme criticism directed against Jimbo Wales precisely because he *does* have that power and people resent it. His primary controls are not at the level of locking articles or unlocking articles, but in the control of what users can run bots and what users can't, and what kinds of bots are approved, and what kinds aren't. There won't be any spambots, yet there will be many policebots if those are ultimately required.
The primary avenue of dissent built into the Wikipedia stems directly from the GPL licence they adopted: anyone can fork.
If the search engines elect to rank search results higher from corporate PR sanitized forks of the Wikipedia then the contents of the primary Wikipedia itself will lose sway among the unthinking masses.
Contrary to most mental skills, the human brain is wired quite well to keep track of who is dishing the scoop with greater or lessor partisan emphasis. Unfortunately, our innate talents in this department tend to drown out our other critical faculties, except in those of us labeled as geeks. A geek can almost be defined as a person who attempts to put rationality ahead of reputation.
The Wikipedia is fundamentally a more social technology than spam. Anyone who doesn't notice this is geek impaired. That's not always a good thing.
Dvorak is a green sandwich. You know, the sandwich you find at the back of your school locker on the last day of school in your grade three year. I could begin by listing his conceptual mistakes, but let's not, I'd rather save my fingers and forearms for my second career.
I'll just pick one. The ActiveX play was part of Microsoft's crusade to queer Sun's Java as the "write once, run anywhere" network-enabled desktop platform of the future. Highly successful, too. Now Java is better known as "write once, debug everywhere". The MS ploy to "embrace and extend" Java was more of a bear hug than it could handle.
Does Dvorak mention either Sun or Java in his childish tirade? Not once. Does Dvorak metion the seven billion dollars venture capital (IIRC) that gathered behind the Netscape storm cloud? Not once. No, Microsoft was too busy overreacting to butt nuggets like Dvorak, to hear him as he retells it.
OK, after getting over my snit that this constitutes a discovery, how about somebody wake me up again once someone bothers to quantify the entropy of the response distribution? Generally, the outliers amount a substantial swack of most distributions, even power law distributions.
Slashdot Mirror
I'm in safe-mode here, my options for screen resolution are 640x480 and 800x600. Neither of those modes were acceptable to the HP uninstall software. Quelle foresight.
OK, I'll report the issue to HP. Spent fifteen minutes recounted the sequence of events in an HP web support form. Press submit. What comes up?
Click the "back" button. Nope, my entire problem report has funneled down the Mariana trench. Another fifteen minutes of my life hopelessly extinguished.
At that point, I did what all good techies do in that situation. I decided to hold a grudge. For a long, long, long time. The perfect storm: incompetence (in the printer driver itself), stupidity (an uninstaller that refused to run in safe-mode), and insolence (a web support form that vaporises my problem report into the Mariana trench).
What's not to like about the new HP? Utterly, utterly, utterly despicable. I couldn't have said it better myself.
The new layout just became active for my account and I can't stand it. Can't even find "new layout goes live" to properly grouse about it. When I enlarge the fonts as I like to do in Firefox, text is mashing over top of other text. I hate the first column. How do I get rid of the whole darn thing? And two of the sections in the first column refuse to remain closed. Light layout is no improvement. I want my serif font back, boxes that look like boxes, and a paucity of useless columns. Don't moderate me -1 troll, moderate me -10 sour beyond all recognition.
Ad skipping remains perfectly legal. No one is forced to own a television in the first place. I don't see the point in getting all hung up over removing the mind control substance from a meal that was never very nutricious in the first place. Once a month I go to my local video store, rent five movies for a week for $10. I can barely *find* five movies a month worth watching. With some help from my pickiest friends I have another five movies on deck.
n ightmare/t _guys_in_the_room/
http://www.rottentomatoes.com/m/gaza_strip/
http://www.rottentomatoes.com/m/10005064-darwins_
http://www.rottentomatoes.com/m/syriana/
http://www.rottentomatoes.com/m/enron_the_smartes
http://www.rottentomatoes.com/m/mondovino/
Seems to be a common theme here: some of these might be a little depressing. There's the problem. Most people who moan about skipping the ads aren't prepared to go all the way.
The proponents of evolution must obey the laws of physics, while the proponents of intelligent design are no so constrained. You can't blame email for the crazy ideas, scams, belief systems foisted by one party upon another. It seems to be endemic wherever humans gather to communicate.
Email has no end of faults that could have been mitigated in some measure by a superior design. All the same, the battle has hardly been lost. Email bears no responsibility for the emergence of botfarms, it's just the unlucky target. If the argument is that we'll never eliminate botfarms, that the botfarms will always be with us, then I think we have far bigger fish to fry.
After we win the battle against the botfarms, perhaps our problems with email will no longer appear quite so dire.
Cities of significance emerged three or four thousand years ago. It wasn't until circa 1850 that sanitation and clean water were fully addressed, and even that victory has many loose ends remaining.
On my monitor which has never had good gamma at the black levels I didn't even *see* the black text on the green bar until I read your comment. I've tried no end of adjustments but eventually I realized that any site worth reading does create this problem to begin with.
Strangely, those of us who prefer it this way appear to be a very small minority. I once read an ergonomics guide that suggesting sitting far enough away from the screen to comfortably page down with your big toe. The comfortable majority of squiggy-lovers seem to be the same crowd who claim that reading long articles on screen is more tiring than printing them out. I'm planning to keep my crisp 19" monitors until a very large, affordable LCD panel comes along featuring 200dpi resolution or better, so I can finally use a slightly smaller font and still enjoy the well-resolved letter forms that make extending on-screen reading an effortless activity.
Anyone know a cure for carpal-toe syndrome?
And yes, typing so fast that entire words go missing and inflections are mangled is an accurate reflection on what a post to slashdot is worth anymore. All praise the brown paper bag.
While I'm on the subject, one of the best typographic makeovers in my experience was when the Economist adopted their new typeface in the early 1990s IIRC which included a provision for small-caps to typeset increasingly prevalent accronyms without rendering the overall paragraph composition garish. But then, the Economist felt they had some worth saying, and they branded themselves more through the style of their prose (and its power to convey ideas in certain rhetorical modality which they favour) rather than the freshness of their masthead.
As a general rule, sans serif fonts are a triumph of style over substance. In a discussion oriented web site, the font should be chosen for readability over appearance. The great thing about Helvetica is that it makes your spelling mistakes 15% less perceptible. Huge appeal to the slashdot editors there. Not long ago the NYTimes adopted a misguided makeover incorporating bamboo-slat column widths crowding their primary asset--their journalism--into a claustrophobic prison window. I've cut my page views from daily to weekly as a direct result.
I didn't regard that post as a troll whatsoever. Heck, with the money he saves buying a generic domain name, he can afford to buy himself a nice Herman Miller Aeron chair.
As far as squatting goes, you can't maintain an empty ecological niche that has proven this profitable. Blaming the squatters is hugely misguided. It's like blaming teenagers for having sex. When a giant new chunk of realestate becomes available (e.g. wild west) you tend to end up with one of two models: land grab, or public auction. The public auction is hard to set up until a market already exists. And people heap just as much vitriole on the FCC for attempting this as they do on the squatters for exploiting what was given to them.
Traditionally the government ends up applying a property tax on assets of this nature. A mil rate assessment process is devised, and anyone setting up shop on a street corner in Manhattan had better plan to operate a viable business or find themselves taxed into bankruptcy.
Anyone care to hazard a guess whether the people presently screaming out about squatters will be first in line to yell their leather lungs out when the government comes along and established a domain-name mil rate assessment bureau?
I'm not sure I agree with recent FSF positions (haven't tracked them much recently), but I agree overall with the FSF taking the long view of free software. There are enormous latent risks that DRM or shifts in the IP landscape (patents) could poison the well ten or twenty years down the road, by which point the crucial battles have already been lost. It's easy to come off as radical crusaders fighting battles that won't play out over a span of decades. Our short little span of attention is our worst enemy in these matters. The fact that they are alone in their extreme urgency doesn't prove much directly: they might be equally alone in a correct analysis of the risks at hand. Just because Chicken Little is squawking, that doesn't mean the sky isn't falling. Glib comments about Chicken Little behaving like Chicken Little have add nothing of any use to the larger debate. My comments add nothing of any use, either, but at least I know the difference.
That smacks of the same cleverness that exempted light trucks and SUVs from mandated fuel economy guidelines, and we all know where that went: the automakers went to town pushing demand into a more profitable vehicle category simultaneously driving net fuel consumption upward. Not to worry. We've since sent a crack team of negotiators to the middle east to ensure our supply. A few of them come home in boxes. Small price to pay.
Big pharma is going to love these terms. Choice is a powerful weapon in highly contested spheres. Do you ever see a major sports league offer the team finishing first in the regular season the choice to pick their first round opponent of the other teams making the cut? Of course not. It would open up a gaping wound for manipulation. Side deals to rig which teams end up matched together.
I could see this working if the choice of patent extension were restricted to the cosmetic/vanity drugs (hair restoration creams, erectile aids, wrinkle creams, etc.) Applied to the only drug in existence that treats a certain form of cancer, or cures childhood malaria? That's giving up far too much.
Easy to offer that now, because few people are clever enough to recognize the end game. How many people saw the California fuel-economy act as a way to upsell consumers into a pricier vehicle class?
Twenty years later the legacy of that loophole has become painfully obvious.
I've read T-bomb's side of the issue whenever this debate has flared up, and I've never come away impressed with his ability to fully enter into the debate.
In effect, Linus is saying the past 20 years of work on object-oriented programming is misguided. I don't buy that.
The comments Linux offered up were from the perspective that there is more than one way to implement information hiding. T-bomb believes that only hardward enforced information-hiding (exploiting MMU protections) is valid. Many critical data structures in Linux enforce information hiding through a C language macro interface which encapsules all the extremely tricky logic that is terribly hard to get right.
Both forms of information-hiding break down if employed incorrectly. In the micro-kernel instance, the break-down is caught (usually) by the MMU which traps out one (or more) of the participants in the failed initiative. I fail to see how Minix can self-heal itself from incorrect program code. What it can do is resume or limp along more gracefully. The fix to the program logic waits upon a developer to come along and identify and correct the faulty program logic.
In Linux, mistakes in the use of the critical primitives will likely take down the entire kernel. It's well understood within the core community of kernel hackers what kinds of misuse or mistakes most often lead to this unfortunate result. If a chunk of code does not pass a visual sanity check, most likely it never becomes part of a sancitified kernel in the first place. The coders who consistently contribute code which does pass the visual sanity review tend to deeply understand the issues at stake. In practice, the dire consequence doesn't leak through the early stages of the development process nearly as often as T-bomb insinuates. I diagnosed one kernel hang myself concerning an embedded 2.4.19 ARM kernel where the semantics of an ARM timer register were subtly abused in concert with sleep/wake cycles. If this same programming mistake had been made in MINUX, it wouldn't have been of any benefit. We still could not ship the product to the end customer until fixing the mistake.
Linus further stated that there are development costs associated with the micro-kernel design, such as achieving the same degree of efficiency while copying stack frames from one protected address frame into another. This leaks out far more often than most people admit. fgetc() doesn't call the OS (as a true u-kern would) for every character processed. That would be too costly for a primitive operation at that scale. Instead the C library buffers character data, a small detail which adds much complexity to things like the C++ streams library. We had some hideous problems with a PHP module until we realized it was buffering old results from fstat() calls (even after closing and reopening the file IIRC). Both of these practices stem from trying to eliminate unnecessary context switches for performance critical operations.
u-kern APIs tend to involve more of this than monolithic kernels because the context switch is more frequent or more expensive. I think in the 3.51 NT u-kern (before they abandoned many of their u-kern principles) they went to a lot of trouble to batch up GDI calls when possible on the application side of the fence. This kind of thing sucks developer attention and can introduce strange artifacts such as the PHP artifact we observed: open file, write data to file, close file, fstat file, observe length is sill zero.
What Linus was pointing out about Linux kernel development is that once you've paid the price to establish good cooperative information-hiding through macro layers (and fully debuged the difficult semantics) in many ways it takes less effort to review the code for correct use of these abstraction layers than it does to deal with the ongoing headaches introduced by crossing MMU barriers for every trivial operation.
I happened to despise the cooperative tasking model from Windows 3.1 and OS 9. In t
I watched "Bowling for Columbine" the other night and "The Corporation" and then "The Civil War" (first episode). Despite all that stimulation, it wasn't until I read your post that the penny finally dropped.
In the American mind, slavery was not originally a race issue. The whites were the people who had guns, and the blacks were the people who didn't. Gun = slave owner, no gun = slave. Now if there were more guns in China, the Chinese could have themselves a Chinese civil war and 20 million Chinese could die in the cause of mutual freedom, if it takes the same 2% of the population as died in the American civil war to settle their small difference of opinion.
I never did figure out why Heston poo-pooed the idea that everyone in America having their own little lump of plutonium. If everyone having handguns is the great catalyst of social freedom, why wouldn't little lumps of plutonium work 1000 times better? I mean, we'd become so concerned about not pissing each other off, we might even turn on an evening news source with some content about the world, not just those stupid sound bites that even Bush can muster up.
Or maybe I'm thinking in the wrong direction. In America, with guns, it took four years for Americans to kill 600,000 other Americans. In Rwanda, with cheap machetes, it only took 100 days to settle 800,000 differences of opinion.
You are so backwards. The 8088 was the cripple-chip used in the first IBM PC.
http://en.wikipedia.org/wiki/Intel_8088
Wang made a PC clone with the 8086 for a while. It was lacking compatibility on a number of fronts, so it died of a not-so-slow terminal illness. It ran about 60% faster than an 8088 PC at the same rated clock speed.
It's amazing that IBM looked at the 8086 and decided it needed an additional whack to the kneecap. As if it wasn't hobbled badly enough already.
Here's one small defect that greatly enhanced the Propecia sales among a generation of early adoptors. The segment registers were designed with a four bit offset from the address registers, yielding a 1MB address space.
If that offset had been six bits instead, the machine would have had a 4MB address space. Each of the eight ISA slots could have had a dedicated 64K address mapped I/O space (consuming 512KB). A 4MB address space would have nicely bridged the evolution to 32-bit multitasking operating systems, which become viable somewhere between 4MB and 8MB.
That four bit offset cost the industry billions of dollars that could have been far better invested elsewhere. There was precious little advantage to a 16 byte segment alignment granularity that I ever noticed.
There was some added efficiency, but MOST of the gain was from the basic fact that a process shrink (and a core adjusted to make use of it) allowed higher frequencies.
This is one of the most bizarre intellectual shell games to ever make common currency. The reason this new rocket goes twice as fast as this old rocket is that we burn the fuel twice as quickly. All the rest of the expensive engineering was to ensure that the rocket doesn't blow itself up on the launch pad coping with all that thrust. But the real technology was burning the fuel twice as fast.
90% of a modern CPU design consists of circuit elements to conceal latency from the frequency obsessed execution units: the branch prediction unit, the parallel instruction decoder, the L1 cache, the L2 cache, and the out-of-order execution engine. It costs electrical power to feed all those circuits to hide those latencies. Which is why cycles/watt have increased far more slowly than cycles/second, despite the fact that power is reduced to a first order as a function of voltage squared.
Hey stop that! I trademarked that clue long before *you* got to it.
What's interesting is how the converse stupidity was established in the first place. Proponents of the RISC camp were so busy announcing their imminent victory they forget to extrapolate more than one or two die shrinks into the future. A design parameter that makes you pull your hair out today can evolve into a subtle advantage a few shrinks later.
I'd rip apart how x86 handles the flag register (partial register stalls and truly Byzantine retirement logic) long before I'd replace the x86 instruction encoding format (except perhaps the prefix bytes, which are super annoying for decode alignment). The stack based floating point register file and instruction set is one of the few x86 features that was beyond redemption at any scale (do I sound like Ralph Nader or what?)
The bottom line is that the flawed nature of the x86 instruction set had a much bigger impact on thermal performance (MIPS/watt) than it ever had on top speed. Now that thermal performance determines top speed, that old shoe is beginning to pinch in new places.
Usually, that's not what happens for us. Generally, when our performance goes up, so goes the power.
Oh yes, quite a claim there. You could limbo under the Prescott power budget on stilts.
Even Grammar Nazis have been outsourced.
But, hey, the grammar Nazi Nazi is still a good old boy. Which makes me a grammar Nazi Nazi Nazi. I'm not telling where I live.
How did outsourcing get such a bad name in the first place? Reproduction requires outsourcing. I thought most people liked it that way. I mean, insourcing is a great just-in-time tactic for a few years, but it starts to seem, I don't know, a little bit xenophobic and self absorbed?
Why drink bourbon when you have pure ethanol?
Why eat cheese when you have lard?
Very clever of the litcrit community to notice that with multitudinous overlaying of semantic fields over top of a single symbol set (words in a language) renders it impossible to construct any symbol sequence that is non-zero only once (in a single semantic field).
To a physicist, however, black means "absence of photons" and carries no connotation whatsoever of "melanin enhanced".
Those among us who get to the point where they refuse to peel away and discard these multitudinous (and often dubious) semantic overlays render themselves incapable of action. Water is also black if one refuses to filter out the crap. I for one refuse to drink from the Ganges of litcrit.
Walmart also happens to be an icon of the American economy during an era where the division of wealth in America returned to the same historical parameters as America experienced when Rockefeller was the world's richest tycoon, and his Standard Oil empire mired in controversy.
Yes, it was bound to happen, and so was the terrible technical analogy to spam gone wild. That doesn't make it true.
Spam has no permanent "revision history", nor a community consensus model to evolve its protections. The Wikipedia is fundamentally based on a centralized control model and I don't see that there is much organized PR flacks can do about this while that model remains in place. I've seen some extreme criticism directed against Jimbo Wales precisely because he *does* have that power and people resent it. His primary controls are not at the level of locking articles or unlocking articles, but in the control of what users can run bots and what users can't, and what kinds of bots are approved, and what kinds aren't. There won't be any spambots, yet there will be many policebots if those are ultimately required.
The primary avenue of dissent built into the Wikipedia stems directly from the GPL licence they adopted: anyone can fork.
If the search engines elect to rank search results higher from corporate PR sanitized forks of the Wikipedia then the contents of the primary Wikipedia itself will lose sway among the unthinking masses.
Contrary to most mental skills, the human brain is wired quite well to keep track of who is dishing the scoop with greater or lessor partisan emphasis. Unfortunately, our innate talents in this department tend to drown out our other critical faculties, except in those of us labeled as geeks. A geek can almost be defined as a person who attempts to put rationality ahead of reputation.
The Wikipedia is fundamentally a more social technology than spam. Anyone who doesn't notice this is geek impaired. That's not always a good thing.
Dvorak is a green sandwich. You know, the sandwich you find at the back of your school locker on the last day of school in your grade three year. I could begin by listing his conceptual mistakes, but let's not, I'd rather save my fingers and forearms for my second career.
I'll just pick one. The ActiveX play was part of Microsoft's crusade to queer Sun's Java as the "write once, run anywhere" network-enabled desktop platform of the future. Highly successful, too. Now Java is better known as "write once, debug everywhere". The MS ploy to "embrace and extend" Java was more of a bear hug than it could handle.
Does Dvorak mention either Sun or Java in his childish tirade? Not once. Does Dvorak metion the seven billion dollars venture capital (IIRC) that gathered behind the Netscape storm cloud? Not once. No, Microsoft was too busy overreacting to butt nuggets like Dvorak, to hear him as he retells it.
OK, after getting over my snit that this constitutes a discovery, how about somebody wake me up again once someone bothers to quantify the entropy of the response distribution? Generally, the outliers amount a substantial swack of most distributions, even power law distributions.