This is probably because in tennis, having the serve is a major advantage. Requiring a player to win by two means that they have to be ahead at a point when the serve advantage evens out.
It's not uncommon for two players who are relatively evenly matched to play for a very long time because they keep winning points (during a tiebreak game) and games (during a set that has no tiebreak game) alternately. (If a set is tied 6-6 in games, and it isn't a deciding set (i.e. it's possible for a player to win the set without winning the match), then it ends with a tiebreak game that has frequent changes of serves and ends, with the winner of the tiebreak game winning the set. This is probably because evenly matched players made sets last far too long otherwise. A deciding set is played out to a two-game margin no matter what, which can take hours.)
I'm not sure there's any method to guarantee that the source they linked is the same as the source they're running, but given that it's AGPL (and thus doing otherwise would be illegal), it seems highly likely that the FSF is in compliance; they seem like one of the least likely organizations to commit a GPL violation.
I wrote libuncursed specifically for NH4 (but intended to work for other roguelikes too), because curses solves the wrong problem nowadays (the problem of "how to talk to an obscure terminal from the 1980s that uses nonstandard terminal codes", rather than the problem of "how to talk to a modern-day terminal emulator that's incompatible with xterm but nonetheless claims to be compatible". I wrote more about the problems here.
Vanilla/mainline NetHack doesn't use libuncursed or curses, but rather a homerolled terminal codes library.
It wasn't me who submitted this story. I would have done if I thought there was a chance of it being accepted - more opinions are always good and Slashdot has lots of technically-inclined users who likely have relevant opinions - but it seemed a little offtopic.
I did write the blog post for the purpose of being linked to news aggregators so that people would have more than a bare post from the devteam to introduce the issues, though.
Mathematically speaking, all these games which are based around predicting what your opponent might do (and possibly a random factor, like in poker) have a perfect strategy, but that perfect strategy has random factors. For instance, the mathematically perfect strategy for rock/paper/scissors is to pick "rock", "paper", and "scissors" each with 1/3 probability. There is nothing an opponent can do to get more than a 50:50 chance of beating this strategy.
Rock/paper/scissors is unusual in that the game is symmetrical: a perfect strategy can't get any better than 50:50 against anyone. That's not true of poker, though; in such a case, a perfect strategy will have a better than 50% chance of beating anyone who plays imperfectly, and a 50% chance against a perfect strategy (due to symmetry).
I'm actually quite interested in the theory of this sort of game (where there are random factors and outguessing opponents involved), and even in simple cases, the calculations can be hard. I'm reasonably interested in whether this poker strategy is a probabilistic one (that can't be outpredicted as long as the random number generator used is sufficiently high-quality), or whether it just takes the best option without randomizing (which is much easier to implement, but which can be outplayed via knowledge of the algorithm like you suggest).
As a simple example, some games use a log of actions in order to store their save files; they replay the actions in order to reconstruct the state of the game. Even if those actions involve floating-point computations, the saves are still typically reproducible given the same executable (given that these games are normally written to avoid the use of uninitialized memory, and the like). If a processor starts handling floating-point differently, suddenly everyone's save files will be broken.
GPUs used to take mathematical shortcuts all the time. More recently, though, with the scientific community starting to use GPU clusters for computation, the main GPU manufacturers have been adding mathematically precise circuitry (and may well use it by default, on the basis that there's no point in having both an accurate and an inaccurate codepath).
You can set a karma modifier for foes and for friends; if you set your foes to have -6 karma, then they're going to be at -1 forever to your view and thus not show up. I know there's at least one Slashdot user who sets their friends to +1 karma, and their foes to +6 so as to not mod them up by mistake, which strikes me as a pretty backwards way of doing things.
One situation I was wondering about for that problem was the use of Japanese/Chinese/Korean marks/overrides, the same way that there are LTR and RTL overrides. Choice of language for a particular ideograph seems to be much the same as choice of direction for an inherently undirectional character (you're interpreting the character differently depending on context). This also has the advantage of being pretty much backwards compatible.
I have mod points right now, so I just checked the interface. There's a link "Moderate" on the bottom of each post. If you click on it, it takes you to the home page, for some reason. If you have enough restraint to hover it without clicking on it, it pops up a menu of the various moderation possibilities, and you can (presumably) click on one to moderate the post.
Posts still appear to have moderation values, and there's a link you can click on to choose a threshold (and thus filter by score), but it's very very small (to the right of the "All", "Informative", etc., line, and a few pixels large).
Conclusion: All the moderation functionality does appear to still exist, but the UI is terrible.
I have also observed some missing functionality (permalink to comment, comment without specifying a title, comment as Anonymous Coward without logging out). This comment was sent from Beta for research purposes, but I think I'm going to go back to Classic for actual browsing.
"==" is normally used for value or reference equality, and that's an "is-a" relationship. You'd want something like instanceof (that's Java, but many other languages have equivalents).
It may be better to realize that rewriting it in Python will prevent this whole class of problems and a bunch of others, and is the way to go.
It's well known among programmer circles that rewriting a program in a different language in order to work around bugs you don't understand just tends to make things worse. Such may be the same with patent laws.
Normally journals claim copyright on the published version of the paper, after it's been edited and typeset by the journal, and don't mind academics sharing the original "preprint" version that was edited and typeset by the original author. (They don't have any reasonable copyright claim on the preprints anyway.) Sending takedowns on preprints is unusual enough to make the news, which is why it's on Slashdot now.
The journal also doesn't pay the academics for their papers; journals work like distributors in the retail market, i.e. their purpose is to make the papers more widely available / discoverable / searchable, in addition to reviewing them to ensure appropriateness and quality (although it's arguable that this is actually a useful function of the journal, given that they don't pay the reviewers either).
Incidentally, my papers have been published in multiple conference proceedings, and I didn't sign a contract for any of them. I assume the contract exists, but the papers were all coauthored, and I think the journals only sought a contract with one of the authors. If this is indeed the case, it makes the situation even more complex.
I currently develop on a laptop with a little under 3GB main memory (and around as much swap). I haven't really noticed memory pressure. (Probably because while I'm programming, I tend not to have much open other than terminals and text editors; maybe a separate window for documentation.) I have more memory pressure when I'm web-browsing (like I am now), rather than developing. (The two activities are mutually exclusive for me; I can't concentrate on programming when I have Slashdot to distract me.)
I'm not convinced that it necessarily means the hardware will be particularly Linux-suited. I bought a Linux (Ubuntu) laptop from Dell a while back (this was in the era of Ubuntu Feisty, I think?). It came with a manual for getting started on Windows, and had a Windows key on the keyboard (which the installed version of Ubuntu happily interpreted as Super, as it usually does). It didn't particularly feel like anything other than a standard Windows Dell laptop that someone had installed Ubuntu on prior to shipping. (Also, the touchpad didn't work, except for a few minutes a few weeks after I obtained it; I suspect that might have been a hardware issue rather than drivers, though. I got used to doing without it.)
If you want a "this cannot happen" with core dump in C, just use abort(). Unless its behaviour is specifically overriden, it's specified to exit the program as unsuccessful termination by the C standards (e.g. 7.22.4.1p2 of C11), and to core dump by POSIX (about as portable as you can get where core dumps are concerned; in straight C, they might not necessarily exist). It also has the benefit of being pretty short, and not undefined behaviour at all.
Of course, that doesn't work in the kernel, but then neither would the other methods you suggested.
You can do that in Windows 8 too, right-click the very bottom-left corner of the screen. (The main problem with that is that it's basically impossible to discover until someone tells you about it.)
It's worse than that. In general, there's nothing stopping anyone sending an email from any address they like; the From: address is simply written onto the email by the sender, much the same way as there's nothing preventing someone sending physical mail writing any return address they like on the envelope. Of course, this makes it kind-of easy to spam, so various methods have sprung up over the years for people to validate the From: address on an email, but there's no universal method that will work for every email you might ever receive.
In general, you should never trust the From: address on an email for any purpose whatsoever other than determining who the sender wants you to think they are.
It's best known for working in some old, buggy FORTRAN situations. Nowadays, it's legal INTERCAL too, albeit mostly to be perverse (and in some compilers, you may have to alias the constant to a variable to be able to assign to it without the compiler complaining).
99.99% of programmers don't need to use single = in a conditional, so add a compiler switch to disallow it as a syntax error instead of just a warning.
In gcc, you can do -Werror=parentheses if you want to make this an error rather than a warning.
I like this trick, and am glad that you're publicising it. It was a pretty clever addition to many compilers; the compilers that don't understand it won't reject the code, the compilers that do will know it's intentional. And maintenance programmers reading the code will know it's intentional, too.
You have to do this no matter what privilege escalation method you use, because a rogue administrator might have left a random setuid binary around somewhere. Or has put a logic bomb in the script. Or something like that. Having only one door to guard is no use when the inside of the building carries a bunch of materials for building extra doors.
In the UK, it's impossible for the upper house to block a bill for more than two sessions; for bills purely about funding/taxation, it's one month. This seems to be a better solution to the solution used by the US, but I prefer Australia's solution even more.
Not every site does. For instance, I vaguely remembered that Microsoft EULAs have jurisdiction based on the country where you live, with the "you consent to jurisdiction in Washington" bit only applying to Americans. I checked the Terms of Service for Bing, and I was right. (For instance, for Europeans, it uses Luxembourg law for breaches of the ToS specifically, and the local jurisdiction for other claims.) Microsoft seems to have local companies set up for the purpose of sorting out contracts with people in countries other than the US.
Many other sites don't seem to consider jurisdictional issues in their TOS at all. I suspect that that might lead to complications if they ever have to sue someone, but it's nicer for their users.
Incidentally, local jurisdiction clauses in a ToS are actually one reason that causes me to avoid agreeing to them, unless they're set up in such a way that they only apply if I invoke them, they can't be invoked against me. (I end up avoiding a large number of major websites because of this.)
Slashdot Mirror
This is probably because in tennis, having the serve is a major advantage. Requiring a player to win by two means that they have to be ahead at a point when the serve advantage evens out.
It's not uncommon for two players who are relatively evenly matched to play for a very long time because they keep winning points (during a tiebreak game) and games (during a set that has no tiebreak game) alternately. (If a set is tied 6-6 in games, and it isn't a deciding set (i.e. it's possible for a player to win the set without winning the match), then it ends with a tiebreak game that has frequent changes of serves and ends, with the winner of the tiebreak game winning the set. This is probably because evenly matched players made sets last far too long otherwise. A deciding set is played out to a two-game margin no matter what, which can take hours.)
It's here, linked from the footer of the page.
I'm not sure there's any method to guarantee that the source they linked is the same as the source they're running, but given that it's AGPL (and thus doing otherwise would be illegal), it seems highly likely that the FSF is in compliance; they seem like one of the least likely organizations to commit a GPL violation.
I wrote libuncursed specifically for NH4 (but intended to work for other roguelikes too), because curses solves the wrong problem nowadays (the problem of "how to talk to an obscure terminal from the 1980s that uses nonstandard terminal codes", rather than the problem of "how to talk to a modern-day terminal emulator that's incompatible with xterm but nonetheless claims to be compatible". I wrote more about the problems here.
Vanilla/mainline NetHack doesn't use libuncursed or curses, but rather a homerolled terminal codes library.
It wasn't me who submitted this story. I would have done if I thought there was a chance of it being accepted - more opinions are always good and Slashdot has lots of technically-inclined users who likely have relevant opinions - but it seemed a little offtopic.
I did write the blog post for the purpose of being linked to news aggregators so that people would have more than a bare post from the devteam to introduce the issues, though.
Mathematically speaking, all these games which are based around predicting what your opponent might do (and possibly a random factor, like in poker) have a perfect strategy, but that perfect strategy has random factors. For instance, the mathematically perfect strategy for rock/paper/scissors is to pick "rock", "paper", and "scissors" each with 1/3 probability. There is nothing an opponent can do to get more than a 50:50 chance of beating this strategy.
Rock/paper/scissors is unusual in that the game is symmetrical: a perfect strategy can't get any better than 50:50 against anyone. That's not true of poker, though; in such a case, a perfect strategy will have a better than 50% chance of beating anyone who plays imperfectly, and a 50% chance against a perfect strategy (due to symmetry).
I'm actually quite interested in the theory of this sort of game (where there are random factors and outguessing opponents involved), and even in simple cases, the calculations can be hard. I'm reasonably interested in whether this poker strategy is a probabilistic one (that can't be outpredicted as long as the random number generator used is sufficiently high-quality), or whether it just takes the best option without randomizing (which is much easier to implement, but which can be outplayed via knowledge of the algorithm like you suggest).
As a simple example, some games use a log of actions in order to store their save files; they replay the actions in order to reconstruct the state of the game. Even if those actions involve floating-point computations, the saves are still typically reproducible given the same executable (given that these games are normally written to avoid the use of uninitialized memory, and the like). If a processor starts handling floating-point differently, suddenly everyone's save files will be broken.
GPUs used to take mathematical shortcuts all the time. More recently, though, with the scientific community starting to use GPU clusters for computation, the main GPU manufacturers have been adding mathematically precise circuitry (and may well use it by default, on the basis that there's no point in having both an accurate and an inaccurate codepath).
You can set a karma modifier for foes and for friends; if you set your foes to have -6 karma, then they're going to be at -1 forever to your view and thus not show up. I know there's at least one Slashdot user who sets their friends to +1 karma, and their foes to +6 so as to not mod them up by mistake, which strikes me as a pretty backwards way of doing things.
One situation I was wondering about for that problem was the use of Japanese/Chinese/Korean marks/overrides, the same way that there are LTR and RTL overrides. Choice of language for a particular ideograph seems to be much the same as choice of direction for an inherently undirectional character (you're interpreting the character differently depending on context). This also has the advantage of being pretty much backwards compatible.
I have mod points right now, so I just checked the interface. There's a link "Moderate" on the bottom of each post. If you click on it, it takes you to the home page, for some reason. If you have enough restraint to hover it without clicking on it, it pops up a menu of the various moderation possibilities, and you can (presumably) click on one to moderate the post.
Posts still appear to have moderation values, and there's a link you can click on to choose a threshold (and thus filter by score), but it's very very small (to the right of the "All", "Informative", etc., line, and a few pixels large).
Conclusion: All the moderation functionality does appear to still exist, but the UI is terrible.
I have also observed some missing functionality (permalink to comment, comment without specifying a title, comment as Anonymous Coward without logging out). This comment was sent from Beta for research purposes, but I think I'm going to go back to Classic for actual browsing.
"==" is normally used for value or reference equality, and that's an "is-a" relationship. You'd want something like instanceof (that's Java, but many other languages have equivalents).
It's well known among programmer circles that rewriting a program in a different language in order to work around bugs you don't understand just tends to make things worse. Such may be the same with patent laws.
Normally journals claim copyright on the published version of the paper, after it's been edited and typeset by the journal, and don't mind academics sharing the original "preprint" version that was edited and typeset by the original author. (They don't have any reasonable copyright claim on the preprints anyway.) Sending takedowns on preprints is unusual enough to make the news, which is why it's on Slashdot now.
The journal also doesn't pay the academics for their papers; journals work like distributors in the retail market, i.e. their purpose is to make the papers more widely available / discoverable / searchable, in addition to reviewing them to ensure appropriateness and quality (although it's arguable that this is actually a useful function of the journal, given that they don't pay the reviewers either).
Incidentally, my papers have been published in multiple conference proceedings, and I didn't sign a contract for any of them. I assume the contract exists, but the papers were all coauthored, and I think the journals only sought a contract with one of the authors. If this is indeed the case, it makes the situation even more complex.
I currently develop on a laptop with a little under 3GB main memory (and around as much swap). I haven't really noticed memory pressure. (Probably because while I'm programming, I tend not to have much open other than terminals and text editors; maybe a separate window for documentation.) I have more memory pressure when I'm web-browsing (like I am now), rather than developing. (The two activities are mutually exclusive for me; I can't concentrate on programming when I have Slashdot to distract me.)
I'm not convinced that it necessarily means the hardware will be particularly Linux-suited. I bought a Linux (Ubuntu) laptop from Dell a while back (this was in the era of Ubuntu Feisty, I think?). It came with a manual for getting started on Windows, and had a Windows key on the keyboard (which the installed version of Ubuntu happily interpreted as Super, as it usually does). It didn't particularly feel like anything other than a standard Windows Dell laptop that someone had installed Ubuntu on prior to shipping. (Also, the touchpad didn't work, except for a few minutes a few weeks after I obtained it; I suspect that might have been a hardware issue rather than drivers, though. I got used to doing without it.)
If you want a "this cannot happen" with core dump in C, just use abort(). Unless its behaviour is specifically overriden, it's specified to exit the program as unsuccessful termination by the C standards (e.g. 7.22.4.1p2 of C11), and to core dump by POSIX (about as portable as you can get where core dumps are concerned; in straight C, they might not necessarily exist). It also has the benefit of being pretty short, and not undefined behaviour at all.
Of course, that doesn't work in the kernel, but then neither would the other methods you suggested.
You can do that in Windows 8 too, right-click the very bottom-left corner of the screen. (The main problem with that is that it's basically impossible to discover until someone tells you about it.)
It's worse than that. In general, there's nothing stopping anyone sending an email from any address they like; the From: address is simply written onto the email by the sender, much the same way as there's nothing preventing someone sending physical mail writing any return address they like on the envelope. Of course, this makes it kind-of easy to spam, so various methods have sprung up over the years for people to validate the From: address on an email, but there's no universal method that will work for every email you might ever receive.
In general, you should never trust the From: address on an email for any purpose whatsoever other than determining who the sender wants you to think they are.
It's best known for working in some old, buggy FORTRAN situations. Nowadays, it's legal INTERCAL too, albeit mostly to be perverse (and in some compilers, you may have to alias the constant to a variable to be able to assign to it without the compiler complaining).
99.99% of programmers don't need to use single = in a conditional, so add a compiler switch to disallow it as a syntax error instead of just a warning.
In gcc, you can do -Werror=parentheses if you want to make this an error rather than a warning.
I like this trick, and am glad that you're publicising it. It was a pretty clever addition to many compilers; the compilers that don't understand it won't reject the code, the compilers that do will know it's intentional. And maintenance programmers reading the code will know it's intentional, too.
You have to do this no matter what privilege escalation method you use, because a rogue administrator might have left a random setuid binary around somewhere. Or has put a logic bomb in the script. Or something like that. Having only one door to guard is no use when the inside of the building carries a bunch of materials for building extra doors.
In the UK, it's impossible for the upper house to block a bill for more than two sessions; for bills purely about funding/taxation, it's one month. This seems to be a better solution to the solution used by the US, but I prefer Australia's solution even more.
Not every site does. For instance, I vaguely remembered that Microsoft EULAs have jurisdiction based on the country where you live, with the "you consent to jurisdiction in Washington" bit only applying to Americans. I checked the Terms of Service for Bing, and I was right. (For instance, for Europeans, it uses Luxembourg law for breaches of the ToS specifically, and the local jurisdiction for other claims.) Microsoft seems to have local companies set up for the purpose of sorting out contracts with people in countries other than the US. Many other sites don't seem to consider jurisdictional issues in their TOS at all. I suspect that that might lead to complications if they ever have to sue someone, but it's nicer for their users. Incidentally, local jurisdiction clauses in a ToS are actually one reason that causes me to avoid agreeing to them, unless they're set up in such a way that they only apply if I invoke them, they can't be invoked against me. (I end up avoiding a large number of major websites because of this.)
Alt-e opens a menu whose name starts with E. AltGr-e is (on Windows, at least) a standard way to type é. So you can't really conflate the two.