Funny thing, I don't recall any differently. If nothing else, history teaches one to serously distrust ignorant, well-meaning, principled people who are out to "improve the world for others".
Inspired by the PEAR results, a research team, led by me, has discovered a revolutionary new phenomenon, which is strongly indicative of pstchic effects. We refer to this phenomenon as the "autological digital character entry effect" (ADCEE). In this preparation, subjects were presented with the PEAR patents along with a device consisting of 104 chiclets and a cathod ray tube (CRT). After reading these patents, subjects' digits were observed to twitch, causing the depression of said chiclets. To our amazement, this appeared to be strongly correlated with the modification of the scan pattern of CRT's electron guns. During debrief, subjects reported that they "thought about typing" and that they "moved their fingers" in order to "type".
This is clearly an instance of direct mental manipulation of matter. The precise mechanism underlying this effect is unknown, but we hypothesize that quantum diffusion of some kind of etheric vibration is at the root of this observable. We intend to pursue these studies vigorously, as soon as sufficient additional funding is acquired.
Who? William Randolph "You provide the pictures and I'll provide the war" Hearst? A warmonger? Promoted the Spanish-American War for profit? No. Really? I'm shocked, shocked, simply shocked.
It's like a lot of wars. There's no question that the US was looking for a reason to drive Spain out of Cuba. There was a recession on, and the Republicans were worried about losing their forty-year lock on the national government, and, besides, the presence of a Spanish colony just sixty miles of the US coast was a good pretext...I mean, a constant irritant.
All cynicism aside, tensions had been building for a long time, and, then as now, the presence of a potentially hostile government in Cuba posed a threat to the US national security. One might reasonably question the magnitude of the military threat posed by Cuba, but the perception of the threat at that time was what mattered, not its perception 102 years later. The Spanish colonial regime in Cuba and the Philippines was tyrannical; we were an anti-colonialist power; and the native peoples were oppressed and exploited. In my opinion, Hearst's newspapers triggered the war, but I rather expect that there would have been a war in Cuba within the following few years anyway.
IBM just maintains the database. The patent belongs to PEAR.
Why do people keep trying to blame the poor clowns at IBM for the nonsensical tripe in their patents database? That would be like blaming the Washington Post for Watergate or the Hearst newspapers for the Spanish-American War!
Why, sonny, in the old days, we used to be type "ping ef.be.ad.de" and be happy about it...although REAL men still typed "ping 0xdeadbeef" to do the same thing.
You know, what I never understood was why those two commands did the same thing, but "ftp ef.be.ad.de" and "ftp 0xdeadbeef" didn't...
First, Japan didn't start World War II as we know it. You know that it wouldn't have happened without Europe.
Of course not. Though Japan gave the US a convenient reason to get into the war, it was not responsible for starting the war itself.
You're joking, right? You do know that Imperial Japan had conquered a large portion of East Asia long before the US became involved, right? You do know of the Rape of Nanking, and the serial war crimes perpetrated by the Japanese army throughout Korea and the Philipines? Please tell me that you're just trolling...
Re:No, it's an "Insane IPO stock tumble"
on
Tech Stocks Tumble
·
· Score: 1
Not only is Yahoo not down for the year, but the Linux stock are down 78%. VA Linux is down 90% from its peak, and it is falling like a stone. Red Hat is flat from its IPO, down 75% from its peak, and, likewise, falling. Caldera is now down 67% from its opening day close, 75% from its peak. and falling. Corel...well, Corel doesn't count, it really is a lousy company run by a crook. The more it falls, the better.
But even investing in the legitimate Linux companies has been an exercise in catching knives.
No, actually, it isn't just a story. He confirmed it. Somebody actually posted the link on/. a few weeks ago.
And guess what? Real software needs to be rewritten from scratch every five years, because the assumptions you make about trade-offs become invalidated. Both sendmail and bind are *long* overdue for a rewrite.
Five years seems generous to me, actually; most software needs a complete rethink, if not a complete rewrite, on every major version. Of course bind, sendmail, and the Linux kernel need a rewrite! My point is that they won't ever get that rewrite. OSS projects that try to perform that rewrite fail -- look at Mozilla, if you need an example. Despite generous funding from its parent companies, Mozilla has already taken twice as long as was budgeted, and it's only just now released a beta. And that beta is slow and unstable.
[Raymond's] point remains true: that if you cannot audit the source, the executables are less trustworthy.
Russ, you know better. Even ignoring the famous Ken Thompson cc story, which happened with a small batch of open source, the whole idea that many eyes find all bugs is total nonsense.
Real software is just too complicated for many eyes to help. I've spent this weekend trying to fix a nasty bug in a Japanese enabled version of a program. Fact is, I doubt that it can be fixed. I'm going to keep trying, but I suspect that I'll have to give up soon, and simply release the program with a known, and fairly nasty, limitation. Why? No one person can possibly understand this code anymore. Many eyes have looked at it, and some of them have improved it...and most of them haven't. And that's what happens to any mature codebase. It reaches a point where addition of new functionality has made the code so baroque that you can't possibly understand all the ins and outs of it.
And don't tell me that doesn't happen in open source; it isn't true. Go look through any open source project that's actually made it to its second generation, and tell me about the cruft. Look at Sendmail. Look at Bind. Look at the fsck-ing Linux kernel, for God's sake. Think about the "fix" for the "thundering herd" select. Can you say "cruft", kiddies? If you weren't intimately familiar with the history of the Mindcraft benchmarks, those kernel patches would be totally incomprehensible.
And this same thing goes through at many levels. Things can be hidden in the code, and that's just the way it's always going to be, whether or not we can see the code itself. The way to avoid gotchas like this is to modularize code, and formalize all interfaces, and slow the code down by a factor of ten. That is never going to happen, because it will cost too much.
The point is that you have anonymity to use or abuse as you like. If I synthesize a nick to send out a single troll, then I've abused anonymity. If I post as an Anonymous Coward because the revelation of the source of my post will hurt me and the suppression of that content will hurt the community more, then I've used anonymity appropriately. We trust untrained members of our community to judge such postings without. We call those people "moderators".
That's exactly what Pinkerton's is doing. They accept anonymous information. You're right, there's a risk that such information will be malicious or just wrong. So they moderate it up or down. If it meets a reasonable set of standards, and that will need to include some double checking, then, hey, guess what, they pass it on to the people who should know about it. (Now, that's a devil's advocate argument: I want to see the standards. That's a reasonable request for the public.)
The basic premise of Katz's criticism is that too many people will abuse anonymity for that to be a workable solution. Slashdot shows that to be plausible in forums like this one, and, certainly, some cases of abuse will happen. The question before the house is whether that abuse will be frequent. That's an empirical question, and there's evidence speaking to it from other anonymous "snitching" services. For instance, the frequency of false reports to Date Rape hotlines is about the same as the frequency of false reports of stranger rape, approximately 3%. That tends to support the claim that WAVE will, on balance, do more good than harm.
You know, it interests me that Slashdot, a site which advocates and promotes anonymous discussion, objects to a corporate attempt to use a similar technique in the "real world".
Stop and think about it: who am I? All you know about me is a nick, and you don't know much about that. Posters here regularly talk, chests puffed out, about the importance of anonymity on the Web, and they have good reasons for those arguments. They talk about revolution and overthrowing governments as reasons to preserve anonymity. (I doubt that any of them really know much about revolution, but that's irrelevant; the principle is right.)
Now tell me why that isn't also important when you think that someone might pose a threat to you? If Mr. X has a gun, and he showed it to me at school, do I want him to ever be able to find out who turned him in? He's already shown that he's dangerous and unstable, for God's sake, by showing me the gun at school in the first place!
Jon, you're right about the other things: prizes, and vague rules, and the fact of safety in the schools. But you obviously didn't read the _Times_ piece very well: one of the other facts the authors pointed out was the number of times that everyone involved had ignored the threats offered by the rampage killers. Given the fact taht a potential rampage killer is, ipso facto, dangerous, what are you going to do to allow people to raise awareness of such a threat without providing anonymity?
Realistically, how many people need vast increases in capacity? Sure, there are uses for larger persistent random access storage devices, but the number of people who actually keep (legal) video on their machines is fairly small. Who else has a use for -- much less a need for -- 2^37 Byte platters? (2^40 bits = 2^37 bytes ~ 100 GBytes effective capacity.)
The thing which would be valuable to consumers would be a sharp increase in data throughput. It's true that disk drive capacity has grown faster than CPU speed over the past few decades -- but data transfer rate has not. The result? The CPU is data-starved, both by the bus and the swap speed.
For those of you not in the know, WaveLAN/AirPort uses DSS (Direct access Spread spectrum), which divides the 84 MHz of bandwidth in the 2.4 GHz reserved band into several channels. The information is distributed throughout the channel that's used. The default channel is ch. 3, but (at least w/ my home FBSD/W2K setup) you can move to any channel, so you aren't messed up by the other signal.
What worries me is what's going to happen to my networkthe first time that somebody walks into my house with a frequency hopping spread spectrum device (e.g. one that uses Bluetooth). That's going to be ugly. And we aren't going to talk about how bad the reception is next to the microwave oven...
[Since] these new CofL were pretty much pre-ordained by the FofF, folks haven't been waiting
Nonsense. The CofL are, among other things, dependent on Larry Lessig's construction of the previous appelate court ruling in US. vs Microsoft. If that construction is flawed, then Judge Jackson's finding of a section 1 violation is trashed. Given that opinion contained the juicy quotation concerning "allowing judges and incompent attorneys to design software", I would suggest that Jackson's CofL are exceedingly brittle.
This entire idea of consideration doesn't make sense from the betting point of view. Lawyers frequently offer this option: "I'll fight your case for no charge. But if you win, I get 50 percent of damages." Sounds like a good deal right? But where is the consideration to the lawyer when the case is lost? This contract is void! Yet people are doing this all the time, in the USA.
Ahh, but you're looking at the wrong side of the contract! The lawyer has already provided his consideration -- his time and expenses in fighting the suit. The bound party here is the client, not the attorney.
Consider -- if the client wins, then he provides half his settlement to the lawyer. If he loses...well, he owes him nothing. But he hasn't promised to make a gift to the lawyer for no consideration; the lawyer has provided him with (one hopes) valuable consideration.
Worst case (it seems) is that we'll have to start providing "written instruments" with each license. Even this may not be as bad as it seems, since much business is already done with legally binding signatures that are generated by machine. (E.g., the signatures on the gazillions of checks issued by big companies every week.) It should be possible to set up a web page for your GPL'd product, where people who want a license just click to get their written instrument. And it will only need to be done once, since whoever sets up the first one will GPL the CGI code for the rest of use to use.
This misses the point. If the copyright to the original source can be obtained, either by purchase or through a lawsuit, then the GPL itself can be revoked. The threat here is that the purchase culture is held to trump the gift culture. If a free-beer license cannot be irrevocable, then GPL, the BSD license, etc. are irrevocably toast.
Think about it. Suppose that Some Villainous Corporation (SVC) sues Linus Torvalds for infringement of their patents. (It's likely to be SCO, for technical reasons, but Microsoft probably makes a better villain.) I'd be stunned if there weren't legitimate infringements. What happens then? Torvalds loses, and has two choices: face the claim that SCO has suffered huge consequent damages based on his product. Oof -- Torvalds is certainly well off, but SVC can claim billions of dollars in damages. So, rather than face that, plus punitive damages, SVC sends an attorney to Torvalds, and says "Look, if you'll simply give us a piece of the copyright, you're fine." (And assignment to FSF won't help; the suit could be brought against them instead of Torvalds. Frankly, FSF is probably more vulnerable.)
You can guess the rest. What GPL? SCV is then free to change the license at will. Poof. No Linux 3.0.
No-one expired the certificates of pilots qualified on the 737 when the 757 came out. A similar situation would be if you were required to do a check exam each year to remain qualified as a MSCE.
That's quite true. But, when the 757 came out, every commercial pilot who wanted to fly one was required to qualify to fly it. The qualification process for any new airframe is quite stringent. So the appropriate parallel for the MCSE-NT4 not carrying over into the MCSE-W2K timeframe is whether a pilot who's qualified for the 737 should automatically be qualified for the 757. Obviously, he or she shouldn't be.
Notice that they talk very carefully about "up to eight user defined objects". This appears to be a front-end processor, using straightforward (and well-known) signal processing algorithms. For instance, background compensation is a well-known technology, as are techniques for tracking user-defined objects.
They aren't talking about doing object detection or segmentation, much less tracking, in a rich optical environment. Doing segmentation and tracking in a sparse and controlled visual environmant might useful in a factory environment, but it is going to be of very little value outside of that realm. That means that this chip is much less than it appears, and, frankly, isn't even all that new. Go look at Carver Mead's work in the early and middle eighties, if you don't believe me; it could do the same thing for the same price. Heck, go look at Eric Schwartz's work in the late eighties; it could do much more...for much less.
Well, on the one hand, you're right that Microsoft needs to publish the specs, but not for the reason you've been stating. I don't think that anybody here has recognized the clear perf and security consequences of the DCE approach to paired user/workstation authentication through Kerb. Microsoft's implementation may be ugly, but it solves two vexing problems with the DCE approach.
The DCE based system requires two calls to autheticate a workstation/user pair. The Microsoft version uses a single, self-contained call to present the ACL data. If the Kerb server is heavily loaded, as they almost always are, then this is a significant perf improvement on a real network.
More importantly, the standard DCE approach admits of a nasty race condition, where a second user uses a first user's workstation token to get at his or her user ACL entry. I have never believed the argument from "well, it doesn't happen any more". Like it or not, the MS hack completely resolves this problem, since the act of authenticating a workstation also, a fortiori, gives back a token which is labelled by its user, and this operation is atomic.
Here's the key point: Possessing lockpicks for the purpose of breaking into other people's houses is a crime, whether or not you ever successfully get into a house using them. If a cop catches you in my house, and the lock on the front door has been picked, and you have lock picks, he has a strong case to prove that not only have you broken into my house, but that you have those lock picks FOR THAT ILLEGAL PURPOSE, which is also a crime. But that's not the only conceivable way such a crime could have happened; if the cop catches you using those picks on my front door without my permission, the prosecutor could still charge you with possessing burglary tools, even though you weren't yet a burglar.
If some script kiddie with l0phtcrack ran it against a machine without any success, and it could be shown that his intent was to break into other people's accounts (say, by e-mail he sent out, or by recorded statments in an IRC), he'd still be guilty of possessing system cracking tools. And that would still be a crime, even though he never actually got into the system. Of course, the easiest way to show that the possession was criminal is quite simple: to catch him inside a system he's cracked. But that doesn't mean that the crime of possession is subsidiary to the crime of information theft; it merely means that the crime of information theft probably entails the crime of possession of cracking tools.
The prosecutor has a duty to charge a defend with all the crimes he could reasonably be believed to have committed, so that he has the best chance of convicting him. That's his job, even if you don't like it.
Actually, this is very close to what I do. I have a thrice-homed FreeBSD box connected to ADSL, and I run both a wired LAN (in the office itself) and a wireless LAN (WaveLAN Turbo using wi0). The FreeBSD box runs NAT for the rest of the network.
The subtlety is that F'BSD 3.3 and 3.4 don't support encryption for WEP. You'll need to wait for 4.0 before that support is available.
Re:Garbage Collection is for Incompetent Programme
on
The New Garbage Man
·
· Score: 1
So why did those non-incompetent programmers who know all about managing memory on their own bother with writing these tools for the weak-minded? Mainly, it's that good programmers understand the principles of program design. One important design principle is that when you want to know what your program is doing at the top level, you shouldn't have to care what is really going on at the lower levels.
I work for a large software company. One of my major jobs is training new developers in "best practices". What you've just said is one of the classic things that people in academic circles believe...but which is utterly and unforgivably wrong. I spend a lot of time teaching them that they do care very much about how the system works, and that they care a lot about how the memory allocator in their system works.
You see, most modern commercial software is well-designed. But efficient design will only take you so far. One of the key performance bottlenecks on an modern computer is simple and straightforward: page faulting. The basic joke we tell is that doubling the speed of your processor gets you to your next page fault twice as fast...and it takes as long with a newer machine as it did before. This means that the cheapest way to get a performance improvement is to avoid page faults -- and you do that by avoiding heap allocation at all costs. If you need to do a malloc, do it in two stages: allocate a small buffer on the stack, test the space needed, and use that buffer if its big enough. It's on the stack, so it won't get page faulted out. (But always check that it is big enough...there's this thing called a buffer overflow condition that you risk there.) Only fall back on the heap when you have to. Allocate several continguous pages of memory to handle list nodes, and use them. Only go outside that block when you must. Etc.
And this is for any program. Do you allow the user to undo things he's done? Then you maintain an undo stack. Allocate nodes for it efficiently. Are you doing searches in dynamic lists? Allocate your tree efficiently. It's not enough to know that you need to use a B-tree or an AVL-tree to keep your data around; you must also make sure that you keep stuff compact. Do you write in C++? The x86 code to support C++-style exceptions (or Java-style exceptions) is fantastically expensive, since the x86 doesn't support dynamic context unwinding in hardware. Maybe you're better off doing nothing in your constructor, and using an explicit initialization step afterwards. That way, you don't need to handle the throwing of exceptions.
Don't get me wrong: good design is the most important key to successful coding. A well designed piece of software can easily run ten times faster than a badly design piece of software that does the same thing. But implementation is also critical, and that often means knowing more than just how the algorithm works, but also means knowing how much operations cost on the target system.
Re:Unix programmers need to clean up after themsel
on
The New Garbage Man
·
· Score: 1
In my experience, the windows kernel is much worse at handling things like that than the unix kernel is. Sometimes you have to reboot to free the memory.
It's not a matter of "better or worse", but of "different". If you're porting to Windows 3.1 or Windows 9x in 16-bit mode, then yes, you can corrupt the global heap by failing to release memory after a process terminates. That's because all global memory in those models is shared among all processes, and so the system can't clean up without possibly destablizing some other process. Bizarre as it sounds, that not a bug, it's a feature. It's one of the ways that Windows 3.1 and 9x preserved backwards compatibility with DOS, in which all processes had relatively direct access to the physical memory.
If you're porting to Windows 9x in 32-bit mode, or if you're porting to Windows NT, then, no, processes clean up after themselves when they shut down. That's because Win32 uses an explicit shared memory model, with shared memory retained against the file system. Thus, if two processes share memory, and the memory that one of them holds is released by process termination, memory held by the other process doesn't get corrupted.
Intel has lost the low end of the Xeon market. For two or fewer processors, the Xeon is a waste of money. They have lost design-ins [sic] in that market segment. Everyone switched to single or dual Pentium IIIs
Hmm. Intel hasn't lost the market segment; Xeon has lost that market segment. Although I'm sure the Intel would be thrilled to be selling lots of processors for single- and dual-processor Xeon systems, they're almost as happy to be selling lots of processors for single- and dual-processor PIII systems.
Slashdot Mirror
Funny thing, I don't recall any differently. If nothing else, history teaches one to serously distrust ignorant, well-meaning, principled people who are out to "improve the world for others".
Inspired by the PEAR results, a research team, led by me, has discovered a revolutionary new phenomenon, which is strongly indicative of pstchic effects. We refer to this phenomenon as the "autological digital character entry effect" (ADCEE). In this preparation, subjects were presented with the PEAR patents along with a device consisting of 104 chiclets and a cathod ray tube (CRT). After reading these patents, subjects' digits were observed to twitch, causing the depression of said chiclets. To our amazement, this appeared to be strongly correlated with the modification of the scan pattern of CRT's electron guns. During debrief, subjects reported that they "thought about typing" and that they "moved their fingers" in order to "type".
This is clearly an instance of direct mental manipulation of matter. The precise mechanism underlying this effect is unknown, but we hypothesize that quantum diffusion of some kind of etheric vibration is at the root of this observable. We intend to pursue these studies vigorously, as soon as sufficient additional funding is acquired.
Who? William Randolph "You provide the pictures and I'll provide the war" Hearst? A warmonger? Promoted the Spanish-American War for profit? No. Really? I'm shocked, shocked, simply shocked.
It's like a lot of wars. There's no question that the US was looking for a reason to drive Spain out of Cuba. There was a recession on, and the Republicans were worried about losing their forty-year lock on the national government, and, besides, the presence of a Spanish colony just sixty miles of the US coast was a good pretext...I mean, a constant irritant.
All cynicism aside, tensions had been building for a long time, and, then as now, the presence of a potentially hostile government in Cuba posed a threat to the US national security. One might reasonably question the magnitude of the military threat posed by Cuba, but the perception of the threat at that time was what mattered, not its perception 102 years later. The Spanish colonial regime in Cuba and the Philippines was tyrannical; we were an anti-colonialist power; and the native peoples were oppressed and exploited. In my opinion, Hearst's newspapers triggered the war, but I rather expect that there would have been a war in Cuba within the following few years anyway.
IBM just maintains the database. The patent belongs to PEAR.
Why do people keep trying to blame the poor clowns at IBM for the nonsensical tripe in their patents database? That would be like blaming the Washington Post for Watergate or the Hearst newspapers for the Spanish-American War!
Why, sonny, in the old days, we used to be type "ping ef.be.ad.de" and be happy about it...although REAL men still typed "ping 0xdeadbeef" to do the same thing.
You know, what I never understood was why those two commands did the same thing, but "ftp ef.be.ad.de" and "ftp 0xdeadbeef" didn't...
Of course not. Though Japan gave the US a convenient reason to get into the war, it was not responsible for starting the war itself.
You're joking, right? You do know that Imperial Japan had conquered a large portion of East Asia long before the US became involved, right? You do know of the Rape of Nanking, and the serial war crimes perpetrated by the Japanese army throughout Korea and the Philipines? Please tell me that you're just trolling...
But even investing in the legitimate Linux companies has been an exercise in catching knives.
The Ken Thompson cc story is just a story.
No, actually, it isn't just a story. He confirmed it. Somebody actually posted the link on /. a few weeks ago.
And guess what? Real software needs to be rewritten from scratch every five years, because the assumptions you make about trade-offs become invalidated. Both sendmail and bind are *long* overdue for a rewrite.
Five years seems generous to me, actually; most software needs a complete rethink, if not a complete rewrite, on every major version. Of course bind, sendmail, and the Linux kernel need a rewrite! My point is that they won't ever get that rewrite. OSS projects that try to perform that rewrite fail -- look at Mozilla, if you need an example. Despite generous funding from its parent companies, Mozilla has already taken twice as long as was budgeted, and it's only just now released a beta. And that beta is slow and unstable.
Real software is just too complicated for many eyes to help. I've spent this weekend trying to fix a nasty bug in a Japanese enabled version of a program. Fact is, I doubt that it can be fixed. I'm going to keep trying, but I suspect that I'll have to give up soon, and simply release the program with a known, and fairly nasty, limitation. Why? No one person can possibly understand this code anymore. Many eyes have looked at it, and some of them have improved it...and most of them haven't. And that's what happens to any mature codebase. It reaches a point where addition of new functionality has made the code so baroque that you can't possibly understand all the ins and outs of it.
And don't tell me that doesn't happen in open source; it isn't true. Go look through any open source project that's actually made it to its second generation, and tell me about the cruft. Look at Sendmail. Look at Bind. Look at the fsck-ing Linux kernel, for God's sake. Think about the "fix" for the "thundering herd" select. Can you say "cruft", kiddies? If you weren't intimately familiar with the history of the Mindcraft benchmarks, those kernel patches would be totally incomprehensible.
And this same thing goes through at many levels. Things can be hidden in the code, and that's just the way it's always going to be, whether or not we can see the code itself. The way to avoid gotchas like this is to modularize code, and formalize all interfaces, and slow the code down by a factor of ten. That is never going to happen, because it will cost too much.
That's exactly what Pinkerton's is doing. They accept anonymous information. You're right, there's a risk that such information will be malicious or just wrong. So they moderate it up or down. If it meets a reasonable set of standards, and that will need to include some double checking, then, hey, guess what, they pass it on to the people who should know about it. (Now, that's a devil's advocate argument: I want to see the standards. That's a reasonable request for the public.)
The basic premise of Katz's criticism is that too many people will abuse anonymity for that to be a workable solution. Slashdot shows that to be plausible in forums like this one, and, certainly, some cases of abuse will happen. The question before the house is whether that abuse will be frequent. That's an empirical question, and there's evidence speaking to it from other anonymous "snitching" services. For instance, the frequency of false reports to Date Rape hotlines is about the same as the frequency of false reports of stranger rape, approximately 3%. That tends to support the claim that WAVE will, on balance, do more good than harm.
Stop and think about it: who am I? All you know about me is a nick, and you don't know much about that. Posters here regularly talk, chests puffed out, about the importance of anonymity on the Web, and they have good reasons for those arguments. They talk about revolution and overthrowing governments as reasons to preserve anonymity. (I doubt that any of them really know much about revolution, but that's irrelevant; the principle is right.)
Now tell me why that isn't also important when you think that someone might pose a threat to you? If Mr. X has a gun, and he showed it to me at school, do I want him to ever be able to find out who turned him in? He's already shown that he's dangerous and unstable, for God's sake, by showing me the gun at school in the first place!
Jon, you're right about the other things: prizes, and vague rules, and the fact of safety in the schools. But you obviously didn't read the _Times_ piece very well: one of the other facts the authors pointed out was the number of times that everyone involved had ignored the threats offered by the rampage killers. Given the fact taht a potential rampage killer is, ipso facto, dangerous, what are you going to do to allow people to raise awareness of such a threat without providing anonymity?
The thing which would be valuable to consumers would be a sharp increase in data throughput. It's true that disk drive capacity has grown faster than CPU speed over the past few decades -- but data transfer rate has not. The result? The CPU is data-starved, both by the bus and the swap speed.
for the clever parody post...
For those of you not in the know, WaveLAN/AirPort uses DSS (Direct access Spread spectrum), which divides the 84 MHz of bandwidth in the 2.4 GHz reserved band into several channels. The information is distributed throughout the channel that's used. The default channel is ch. 3, but (at least w/ my home FBSD/W2K setup) you can move to any channel, so you aren't messed up by the other signal.
What worries me is what's going to happen to my networkthe first time that somebody walks into my house with a frequency hopping spread spectrum device (e.g. one that uses Bluetooth). That's going to be ugly. And we aren't going to talk about how bad the reception is next to the microwave oven...
Nonsense. The CofL are, among other things, dependent on Larry Lessig's construction of the previous appelate court ruling in US. vs Microsoft. If that construction is flawed, then Judge Jackson's finding of a section 1 violation is trashed. Given that opinion contained the juicy quotation concerning "allowing judges and incompent attorneys to design software", I would suggest that Jackson's CofL are exceedingly brittle.
Ahh, but you're looking at the wrong side of the contract! The lawyer has already provided his consideration -- his time and expenses in fighting the suit. The bound party here is the client, not the attorney.
Consider -- if the client wins, then he provides half his settlement to the lawyer. If he loses...well, he owes him nothing. But he hasn't promised to make a gift to the lawyer for no consideration; the lawyer has provided him with (one hopes) valuable consideration.
This misses the point. If the copyright to the original source can be obtained, either by purchase or through a lawsuit, then the GPL itself can be revoked. The threat here is that the purchase culture is held to trump the gift culture. If a free-beer license cannot be irrevocable, then GPL, the BSD license, etc. are irrevocably toast.
Think about it. Suppose that Some Villainous Corporation (SVC) sues Linus Torvalds for infringement of their patents. (It's likely to be SCO, for technical reasons, but Microsoft probably makes a better villain.) I'd be stunned if there weren't legitimate infringements. What happens then? Torvalds loses, and has two choices: face the claim that SCO has suffered huge consequent damages based on his product. Oof -- Torvalds is certainly well off, but SVC can claim billions of dollars in damages. So, rather than face that, plus punitive damages, SVC sends an attorney to Torvalds, and says "Look, if you'll simply give us a piece of the copyright, you're fine." (And assignment to FSF won't help; the suit could be brought against them instead of Torvalds. Frankly, FSF is probably more vulnerable.)
You can guess the rest. What GPL? SCV is then free to change the license at will. Poof. No Linux 3.0.
That's quite true. But, when the 757 came out, every commercial pilot who wanted to fly one was required to qualify to fly it. The qualification process for any new airframe is quite stringent. So the appropriate parallel for the MCSE-NT4 not carrying over into the MCSE-W2K timeframe is whether a pilot who's qualified for the 737 should automatically be qualified for the 757. Obviously, he or she shouldn't be.
What's so hard about that?
They aren't talking about doing object detection or segmentation, much less tracking, in a rich optical environment. Doing segmentation and tracking in a sparse and controlled visual environmant might useful in a factory environment, but it is going to be of very little value outside of that realm. That means that this chip is much less than it appears, and, frankly, isn't even all that new. Go look at Carver Mead's work in the early and middle eighties, if you don't believe me; it could do the same thing for the same price. Heck, go look at Eric Schwartz's work in the late eighties; it could do much more...for much less.
The DCE based system requires two calls to autheticate a workstation/user pair. The Microsoft version uses a single, self-contained call to present the ACL data. If the Kerb server is heavily loaded, as they almost always are, then this is a significant perf improvement on a real network.
More importantly, the standard DCE approach admits of a nasty race condition, where a second user uses a first user's workstation token to get at his or her user ACL entry. I have never believed the argument from "well, it doesn't happen any more". Like it or not, the MS hack completely resolves this problem, since the act of authenticating a workstation also, a fortiori, gives back a token which is labelled by its user, and this operation is atomic.
Here's the key point: Possessing lockpicks for the purpose of breaking into other people's houses is a crime, whether or not you ever successfully get into a house using them. If a cop catches you in my house, and the lock on the front door has been picked, and you have lock picks, he has a strong case to prove that not only have you broken into my house, but that you have those lock picks FOR THAT ILLEGAL PURPOSE, which is also a crime. But that's not the only conceivable way such a crime could have happened; if the cop catches you using those picks on my front door without my permission, the prosecutor could still charge you with possessing burglary tools, even though you weren't yet a burglar.
If some script kiddie with l0phtcrack ran it against a machine without any success, and it could be shown that his intent was to break into other people's accounts (say, by e-mail he sent out, or by recorded statments in an IRC), he'd still be guilty of possessing system cracking tools. And that would still be a crime, even though he never actually got into the system. Of course, the easiest way to show that the possession was criminal is quite simple: to catch him inside a system he's cracked. But that doesn't mean that the crime of possession is subsidiary to the crime of information theft; it merely means that the crime of information theft probably entails the crime of possession of cracking tools.
The prosecutor has a duty to charge a defend with all the crimes he could reasonably be believed to have committed, so that he has the best chance of convicting him. That's his job, even if you don't like it.
The subtlety is that F'BSD 3.3 and 3.4 don't support encryption for WEP. You'll need to wait for 4.0 before that support is available.
I work for a large software company. One of my major jobs is training new developers in "best practices". What you've just said is one of the classic things that people in academic circles believe...but which is utterly and unforgivably wrong. I spend a lot of time teaching them that they do care very much about how the system works, and that they care a lot about how the memory allocator in their system works.
You see, most modern commercial software is well-designed. But efficient design will only take you so far. One of the key performance bottlenecks on an modern computer is simple and straightforward: page faulting. The basic joke we tell is that doubling the speed of your processor gets you to your next page fault twice as fast...and it takes as long with a newer machine as it did before. This means that the cheapest way to get a performance improvement is to avoid page faults -- and you do that by avoiding heap allocation at all costs. If you need to do a malloc, do it in two stages: allocate a small buffer on the stack, test the space needed, and use that buffer if its big enough. It's on the stack, so it won't get page faulted out. (But always check that it is big enough...there's this thing called a buffer overflow condition that you risk there.) Only fall back on the heap when you have to. Allocate several continguous pages of memory to handle list nodes, and use them. Only go outside that block when you must. Etc.
And this is for any program. Do you allow the user to undo things he's done? Then you maintain an undo stack. Allocate nodes for it efficiently. Are you doing searches in dynamic lists? Allocate your tree efficiently. It's not enough to know that you need to use a B-tree or an AVL-tree to keep your data around; you must also make sure that you keep stuff compact. Do you write in C++? The x86 code to support C++-style exceptions (or Java-style exceptions) is fantastically expensive, since the x86 doesn't support dynamic context unwinding in hardware. Maybe you're better off doing nothing in your constructor, and using an explicit initialization step afterwards. That way, you don't need to handle the throwing of exceptions.
Don't get me wrong: good design is the most important key to successful coding. A well designed piece of software can easily run ten times faster than a badly design piece of software that does the same thing. But implementation is also critical, and that often means knowing more than just how the algorithm works, but also means knowing how much operations cost on the target system.
It's not a matter of "better or worse", but of "different". If you're porting to Windows 3.1 or Windows 9x in 16-bit mode, then yes, you can corrupt the global heap by failing to release memory after a process terminates. That's because all global memory in those models is shared among all processes, and so the system can't clean up without possibly destablizing some other process. Bizarre as it sounds, that not a bug, it's a feature. It's one of the ways that Windows 3.1 and 9x preserved backwards compatibility with DOS, in which all processes had relatively direct access to the physical memory.
If you're porting to Windows 9x in 32-bit mode, or if you're porting to Windows NT, then, no, processes clean up after themselves when they shut down. That's because Win32 uses an explicit shared memory model, with shared memory retained against the file system. Thus, if two processes share memory, and the memory that one of them holds is released by process termination, memory held by the other process doesn't get corrupted.
Once in December of last year (by CmdrTaco), and once even earlier. It was shot down then, and I don't see why it'd fly now.
Hmm. Intel hasn't lost the market segment; Xeon has lost that market segment. Although I'm sure the Intel would be thrilled to be selling lots of processors for single- and dual-processor Xeon systems, they're almost as happy to be selling lots of processors for single- and dual-processor PIII systems.