The same reason system administrators are reluctant to reboot their production servers in order to apply a kernel patch that fixes serious performance issues. Government downtime is costly. And in this case we have a fairly invasive patch involving a new algorithm, and although the algorithm itself is well researched, the specific implementation needs extensive testing for which good spare hardware is unavailable.
If only we had a good simulator to test political ideas on...
1. An ordinary user has very limited access to your server's internal logic, so it is often hard to know, without actually attempting a harmless exploit, whether some suspicious-looking code is actually a bug and is actually exploitable. Your suggestion is impractical unless the sysadmin has the time to deal with people pointing out false positives rather than real bugs.
2. After a sysadmin finds a security hole in his system, he can avoid cleaning up everything if the logs can be trusted and they say the hole has not been exploited. Now, if the trusted log instead says that the only exploit to your SQL injection bug is that someone did a "CREATE TABLE this_is_a_security_hole...", this does not take much more effort to deal with.
3. s/your house/the government building containing everyone's personal data. A security hole may affect other people more than it affects you.
Back then Ruby's C interface appeared much easier to use than Python's reference-counting mess (I don't know about recent developments). Apart from that, both are good as far as scripting languages go.
If you don't want your article to be messed up by well-intentioned but mostly unprofessional edits, you can put it in your user subpage and link to it from the article's discussion page.
...if the student can understand the stuff on Wikipedia but the primary source is too difficult for him. It's not good to cite an encyclopedia, but it is at least better than citing something you don't properly understand.
I'd impose a ban on all MAFIAA material on campus, pirated or not, just like how schools around here ban pornography. I'd also put up banners like "When you listen to RIAA's music or watch MPAA's movies, you are putting our school in danger. STOP!".
If the candidates are equally good (or bad) to someone, it doesn't matter whom he votes. However, if a significant portion of people gives up their right to vote, one cannot reasonably expect the behavior of all the candidates to remain the same.
To be fair Mathtype has plenty of keyboard shortcuts as well. However, LaTeX's macros are just so much more convenient when there are a lot of font changes and notations I have not decided on. Besides, the looks of papers written with Word/Mathtype just doesn't look quite "right" to me, even though I guess I have a pretty good grasp of Word's advanced features. Maybe this has something to do with Word's inability to insert line breaks inside inline equations and the strange line spacing. Finally, Word seems to get unstable when I have more than a hundred or so equations in a single document, which happens pretty often.
The ability to regenerate the key (and that the majority of the users actually does it) is important. The initial key might have gotten leaked by a bad manufacturer, might identify me in a way I don't like, and the previous owner of the computer might have gotten the key signed somewhere and subsequently earned a bad reputation with it. In any case I have little control over it. After generating a key, I can get it signed by whoever I wanted to be trusted by, e.g. an online game server, a distributed computing project, my employer, or even myself if malware-prevention is the goal. In some cases the trusting party (e.g. my employer or myself) may want a copy of the private key, so there should be the option that the private key be copied somewhere at key generation time (and never again, so the trusting party knows that no one else can hack the TPM). Since I may want to assume multiple identities, to be trusted by multiple different parties, all with the same computer (e.g. I may not want anyone to know that two user accounts in two different online games are both operated by me with the same computer, even if the same company runs the servers for the two games), the key had better be pluggable, particularly when there is no commonly trusted third party.
In other words, the TPM I might want is not much different from a smart card, except that it has a tamper-proof way of hashing the hardware and software. This hashing is meant to be used for things like cheat-prevention, distributed computing and enterprise system management, and not for unsuitable purposes such as forcing the user of a website to use a certain operating system or browser. Of course, an ordinary website should not need to have my key signed, anyway.
...and I want to protect trade secrets from malwares, harddrive thieves and malicious employees. Then it is very reasonable for me to want the TPM to be hackable by myself but not by any of my employees, so that I don't need to trust any third party to properly manage my endorsement keys, and I can be sure that my encrypted documents are still accessible even if something breaks horribly.
My English teacher has probably signed some sort of agreement that prohibits him from talking about some political or religious topics in classes. Maybe the agreement also asks him not to badmouth the government even after he goes back.
Anyway, in the excitement of visiting a whole new country for the first time, censorship issues may well appear unimportant to most of these teachers.
In China, we sometimes use "little sister" to refer to the people hired by the authorities to check posts on Internet forums for political correctness. Of course this is sort of a parody to "Big Brother", but indeed most such people are just young, politically unmotivated university students, frequently female, that are looking for some pocket money.
I live in China and such take-down notices by phone are well known, though I have never got any myself as I'm not a webmaster. Censorship issues aside, I wonder how the "security officials" can authenticate themselves. It will be incredibly insecure if a webmaster just listens to a random guy over the phone.
I wish patents work more like copyrights in this aspect. If there are likely a huge number of ways to accomplish something, someone can patent one method he has found and others will not be allowed to copy it without permission. If there is only one correct (efficient) way to do something, the corresponding parts of the claims get filtered out.
As a relatively experienced user, I don't mind text configuration files per se, but I do hope the configuration process is made more intuitive with a good amount of log messages (and possibly a working debuginfo package so that I can start gdb when REALLY necessary). Right now I often change an option and wonder why it does not have the desired effect. Did I understand its meaning incorrectly? Does the problem lie elsewhere (e.g. missing file or bad permission bits)? Maybe the new configuration hasn't taken effect yet because I forgot to restart something? Or maybe I have just encountered a software bug or limitation?
Very often, for example in the case of fontconfig, the necessary debugging messages are available, but only if some poorly documented environment variable or command is used.
Your second requirement might encourage patent trolls. They would need to do more than coming up with some idea, but a half-baked proof-of-concept is almost trivial to make compared to a finished product, especially since profitability is not a concern. For example, for things like the one-click patent, one can simply make some web pages and server scripts and put them into a computer, which may count as the "working model". This should not cost much compared to the other costs in obtaining a patent, and once obtained, such patents can still be used for trolling.
I'm opposed to patents for the most part, but if we have to grant a patent to someone, it would be better if the patented idea actually gets fully implemented and turns into a useful, profitable product, than having it sitting around like a landmine, serving no purpose except for denying others' use of the idea.
Of course, that's unless your first requirement intends to make patents as limited in scope as copyrights, that is, they only cover blatant copying of e.g. blueprints and PCB designs and do not cover clean-room reimplementations of the same idea. Maybe that's not such a bad idea, after all...
Then you can keep the invention secret, and disclose it to some MegaCorp for manufacture only after they accept your conditions. If you can afford enough lawyers to enforce the patent, you can enforce this contract as well. After all, something between you and the MegaCorp should not affect everybody else.
Such contracts should be fairly common. For example, companies such as Nvidia can ask a fab to manufacture chips for it, and the fab does not automagically gain rights to use the design arbitrarily.
When the invention is finally implemented and ready to hit the market, you may worry that its internal workings cannot be kept secret anymore (e.g. for drugs). At such times a patent sounds more reasonable.
A few days ago I was running a numerical program that has a working set size of about 50MB. There is no way all these data can fit in the L2 cache, and memory bandwidth is crucial to its performance.
Slashdot Mirror
The same reason system administrators are reluctant to reboot their production servers in order to apply a kernel patch that fixes serious performance issues. Government downtime is costly. And in this case we have a fairly invasive patch involving a new algorithm, and although the algorithm itself is well researched, the specific implementation needs extensive testing for which good spare hardware is unavailable.
If only we had a good simulator to test political ideas on...
1. An ordinary user has very limited access to your server's internal logic, so it is often hard to know, without actually attempting a harmless exploit, whether some suspicious-looking code is actually a bug and is actually exploitable. Your suggestion is impractical unless the sysadmin has the time to deal with people pointing out false positives rather than real bugs.
2. After a sysadmin finds a security hole in his system, he can avoid cleaning up everything if the logs can be trusted and they say the hole has not been exploited. Now, if the trusted log instead says that the only exploit to your SQL injection bug is that someone did a "CREATE TABLE this_is_a_security_hole ...", this does not take much more effort to deal with.
3. s/your house/the government building containing everyone's personal data. A security hole may affect other people more than it affects you.
1KG of U235 = 1.557*10^6KG of Gasoline
The patent law isn't being changed. If the applicants feel that their patents are illegally rejected, of course they can challenge the decision.
Back then Ruby's C interface appeared much easier to use than Python's reference-counting mess (I don't know about recent developments). Apart from that, both are good as far as scripting languages go.
If you don't want your article to be messed up by well-intentioned but mostly unprofessional edits, you can put it in your user subpage and link to it from the article's discussion page.
...if the student can understand the stuff on Wikipedia but the primary source is too difficult for him. It's not good to cite an encyclopedia, but it is at least better than citing something you don't properly understand.
I'd impose a ban on all MAFIAA material on campus, pirated or not, just like how schools around here ban pornography. I'd also put up banners like "When you listen to RIAA's music or watch MPAA's movies, you are putting our school in danger. STOP!".
If the candidates are equally good (or bad) to someone, it doesn't matter whom he votes. However, if a significant portion of people gives up their right to vote, one cannot reasonably expect the behavior of all the candidates to remain the same.
even though it does not create any new information.
CM fonts look pretty good in xpdf IMHO, although they appear a little too light when antialiased at my current gamma settings.
To be fair Mathtype has plenty of keyboard shortcuts as well. However, LaTeX's macros are just so much more convenient when there are a lot of font changes and notations I have not decided on. Besides, the looks of papers written with Word/Mathtype just doesn't look quite "right" to me, even though I guess I have a pretty good grasp of Word's advanced features. Maybe this has something to do with Word's inability to insert line breaks inside inline equations and the strange line spacing. Finally, Word seems to get unstable when I have more than a hundred or so equations in a single document, which happens pretty often.
The ability to regenerate the key (and that the majority of the users actually does it) is important. The initial key might have gotten leaked by a bad manufacturer, might identify me in a way I don't like, and the previous owner of the computer might have gotten the key signed somewhere and subsequently earned a bad reputation with it. In any case I have little control over it. After generating a key, I can get it signed by whoever I wanted to be trusted by, e.g. an online game server, a distributed computing project, my employer, or even myself if malware-prevention is the goal. In some cases the trusting party (e.g. my employer or myself) may want a copy of the private key, so there should be the option that the private key be copied somewhere at key generation time (and never again, so the trusting party knows that no one else can hack the TPM). Since I may want to assume multiple identities, to be trusted by multiple different parties, all with the same computer (e.g. I may not want anyone to know that two user accounts in two different online games are both operated by me with the same computer, even if the same company runs the servers for the two games), the key had better be pluggable, particularly when there is no commonly trusted third party. In other words, the TPM I might want is not much different from a smart card, except that it has a tamper-proof way of hashing the hardware and software. This hashing is meant to be used for things like cheat-prevention, distributed computing and enterprise system management, and not for unsuitable purposes such as forcing the user of a website to use a certain operating system or browser. Of course, an ordinary website should not need to have my key signed, anyway.
...and I want to protect trade secrets from malwares, harddrive thieves and malicious employees. Then it is very reasonable for me to want the TPM to be hackable by myself but not by any of my employees, so that I don't need to trust any third party to properly manage my endorsement keys, and I can be sure that my encrypted documents are still accessible even if something breaks horribly.
Anyway, in the excitement of visiting a whole new country for the first time, censorship issues may well appear unimportant to most of these teachers.
In China, we sometimes use "little sister" to refer to the people hired by the authorities to check posts on Internet forums for political correctness. Of course this is sort of a parody to "Big Brother", but indeed most such people are just young, politically unmotivated university students, frequently female, that are looking for some pocket money.
I live in China and such take-down notices by phone are well known, though I have never got any myself as I'm not a webmaster. Censorship issues aside, I wonder how the "security officials" can authenticate themselves. It will be incredibly insecure if a webmaster just listens to a random guy over the phone.
I wish patents work more like copyrights in this aspect. If there are likely a huge number of ways to accomplish something, someone can patent one method he has found and others will not be allowed to copy it without permission. If there is only one correct (efficient) way to do something, the corresponding parts of the claims get filtered out.
As a relatively experienced user, I don't mind text configuration files per se, but I do hope the configuration process is made more intuitive with a good amount of log messages (and possibly a working debuginfo package so that I can start gdb when REALLY necessary). Right now I often change an option and wonder why it does not have the desired effect. Did I understand its meaning incorrectly? Does the problem lie elsewhere (e.g. missing file or bad permission bits)? Maybe the new configuration hasn't taken effect yet because I forgot to restart something? Or maybe I have just encountered a software bug or limitation?
Very often, for example in the case of fontconfig, the necessary debugging messages are available, but only if some poorly documented environment variable or command is used.
Your second requirement might encourage patent trolls. They would need to do more than coming up with some idea, but a half-baked proof-of-concept is almost trivial to make compared to a finished product, especially since profitability is not a concern. For example, for things like the one-click patent, one can simply make some web pages and server scripts and put them into a computer, which may count as the "working model". This should not cost much compared to the other costs in obtaining a patent, and once obtained, such patents can still be used for trolling.
I'm opposed to patents for the most part, but if we have to grant a patent to someone, it would be better if the patented idea actually gets fully implemented and turns into a useful, profitable product, than having it sitting around like a landmine, serving no purpose except for denying others' use of the idea.
Of course, that's unless your first requirement intends to make patents as limited in scope as copyrights, that is, they only cover blatant copying of e.g. blueprints and PCB designs and do not cover clean-room reimplementations of the same idea. Maybe that's not such a bad idea, after all...
Does it help if you adjust the tile cache size?
Then you can keep the invention secret, and disclose it to some MegaCorp for manufacture only after they accept your conditions. If you can afford enough lawyers to enforce the patent, you can enforce this contract as well. After all, something between you and the MegaCorp should not affect everybody else. Such contracts should be fairly common. For example, companies such as Nvidia can ask a fab to manufacture chips for it, and the fab does not automagically gain rights to use the design arbitrarily. When the invention is finally implemented and ready to hit the market, you may worry that its internal workings cannot be kept secret anymore (e.g. for drugs). At such times a patent sounds more reasonable.
He disagrees with the law (including the Berne Convention), period. There is no inconsistency in this.
A few days ago I was running a numerical program that has a working set size of about 50MB. There is no way all these data can fit in the L2 cache, and memory bandwidth is crucial to its performance.
It will be great for sites that have a LOT of mirrors. Particularly sourceforge, of which no single mirror is reliably "fast enough" for me.