Software gets delayed when it's not done when they thought it would be, but it's not nearly 2005 yet. Putting off the next release for more than 2 years is just begging to fall behind. Just about anyone else would at least plan to have some sort of release before then, even if it doesn't have every feature in the roadmap. Of course, some projects might not actually manage the scheduled release in 2 years, but that would be due to not estimating correctly in advance.
Where I live, the guy who calls for Patrolmen's donations sounds on the phone like you'd expect a local cop to sound (deep voice, local accent and idiom; the sort of voice that makes you expect a friendly but over-eager clap on the back). I wouldn't be too surprised if the caller wasn't a real police officer, but I think that friends or family are more likely than professional telemarketters.
Ticketting probably only applies to fund-raising speed traps in your town, since they actually care about people who are causing traffic hazards, and non-local cops don't care about your donation anyway. In my town, at least, traps are largely aimed at people commuting through the town, anyway, so having a local parking sticker is probably just as good as a donation sticker.
Re:Something to keep in mind.
on
MIT Everyware
·
· Score: 1
Certainly a motivated student can do better with a set of good coursework and no school than with a lousy course (or, more likely, a program that doesn't have a course that applies).
I just wanted to point out that courses often have a substantial component that isn't purely information, which would need to be provided locally for the course to really work at all. I actually feel that MIT's coursework isn't necessarily better than other schools; what makes MIT work is that the students have access to really smart and experienced people, as course staff, as other students in the class, and as students who have taken the class before. This means that MIT's coursework can be somewhat more advanced and still be possible.
My somewhat cynical guess is that the courses with information online are mostly courses which rely somewhat heavily on things other than the coursework.
It would be interesting to try to replicate MIT courses online with a group of other motivated students at your local school, and, potentially, some sort of group communication with students in other places doing the course at the same time. I think that adding local or online resources to the MIT coursework would be much more effective than trying to do the MIT coursework individually.
Re:Something to keep in mind.
on
MIT Everyware
·
· Score: 1
6.170 is a team course; a substantial part of the point is that it is more work than one person can do in a term, so you will have to work successfully with other students. So if you want to replicate the experience outside of MIT, you'll want to find 2-3 like-minded friends when you try it. Of course, you could do it when you had 3-4 times as much time to spend on it as a student taking a full courseload, but then you'd be missing half of the point: good design means that other people will be able to make productive use of your code.
There are two factors that suggest that it'll be stable at about the time it is debugged:
A lot of the things which differ between 2.4 and 2.6 with respect to drivers are the APIs of some important things they depend on. The new APIs are less error-prone. This suggests that a driver will work correctly if it worked before and has been updated, and it won't compile otherwise.
Drivers don't get updated until the interfaces are stable and the kernel otherwise works well enough for the people who maintain those drivers to pay attention to the new version. This means that the core kernel will be pretty stable before drivers get updated, and when drivers get updated, the core kernel works.
Of course, new drivers that compile won't necessarily work, and new optional features that compile may not yet be stable (many of the power management bits, for instance). But I think that everything from 2.4 will be working when all of the modules compile for 2.6, due to these factors.
Of course, scheduler tuning is somewhat unrelated, and might delay the release.
Games are frequently targetted at males, but the games that last well are frequently not. I suppose that the scarcity of games that aren't annoying to women would tend to lead to women playing those games longer, meaning that they'd do better. The demand seems to be 41% of the market, and the supply is clearly much lower, even ignoring the fact that 41% is only the portion of the potential market which finds something to play.
Regardless, it seems to me that games that aren't targetted at males (or anyone else in particular) generally last better. Generally they don't just involve flashy graphics, but have some sort of interesting mechanic, room for skill, lots of amusing and obscure things to find, and so forth.
Tomb Raider is the example you give of a game with a female sex object, but it turns out to be reasonably popular with women, and actually has clever puzzles. Of the dozens of others you refer to, the ones without a substantial female following are largely not worth mentioning by name. The only vastly successful game I can think of that has managed to not have much of a female following (to my knowledge) is GTA, which probably only avoids having a female following by being somewhat gratuitously offensive and not having a female protagonist.
The real issue with the color output from LEDs is not that the color is wrong (since you can change that by changing the portions of different color output), but that they produce thin lines of spectrum, rather than the black body curves that incandescents produce. While your eyes can't tell the difference directly (since you only have three different colors of perception), surfaces respond differently to different wavelengths in such a way that light that looks the same to you makes surfaces look different. This means that LED light looks artificial in a way that incandescent light does not.
The only way of getting a wide spectrum of light is to have an object glow with heat, where the energy released per photon varies chaotically, rather than using a process that outputs individual photons which will only produce light at the wavelengths that correspond to energy gaps. Glowing with heat is lower efficiency than emitting individual photons.
I suspect that LEDs will become more popular in step with paint formulas that look good (and look right) under LED light, and also with people coming to expect LED light more.
Earl Grey doesn't need lemon, because it's got bergamot. If you need to put more citrus in your Earl Grey, you should get stronger Earl Grey. If you're going to put lemon in it, get a nice Orange Pekoe instead.
With a reasonable installer, installing new Linux software is a matter of determining the name of a piece of software you want. "apt-get install " should suffice, and shouldn't need anything other than your root password to complete. Good software packaging isn't exclusive to Windows, these days.
TFM generally doesn't have any information on installing stuff, although there's generally a lot on getting things configured the way you want, which is the stuff that the InstallShield wizard wouldn't support.
The self-documenting aspect of the community actually makes "RTFM" useful, because there is documentation written by people who have had your problem for the community to point you at. It's more helpful than people trying to write a new explanation of the solution every time someone has a question. Asking your question and getting told "RTFM" is sometimes more effective than trying to determine for yourself whether TFM will answer your question, or even finding the applicable documentation.
It also starts the dialogue about TFM, which may be out of date or have gotten lost somewhere, as was reported on linux-kernel recently about some documentation.
Ideally, ALSA would support mixing regardless of whether the hardware did it (that is, if the hardware or driver doesn't support it, it would be done in software). Probably the right way to handle it would be for ALSA to do it in userspace if the hardware doesn't support it.
Having only some machines need an external mixer (with a different API) is just a bad idea.
For example, Fox released "Bend It Like Beckham", which is actually a really good movie. But it's not the sort of movie that makes good advertizing (the jokes are only funny in context, there aren't any explosions, and the plot doesn't make a good single line), so it's only made $30 million in the US, less than the usual openning weekend, total since March.
Fox (like the other studios) does make good movies, but good movies that don't make good ads don't make much money. People tend to decide first to go to the movies, and then what to see. Unfortunately, people then decide to see bad movies with good ads, rather than good movies that are obscure, which means that they have to compete with the other studios who pump bad movies.
If you read the article (rather than just the blurb), nowhere do the movie people actually say that this is a bad thing, that they don't like this turn of events, or that they want to do anything to change it.
It could well be a good thing overall, such that they can release good movies with staying power rather than going for glitzy special effects that make good ads. The movie business, unlike the music business, actually likes to produce good stuff, but they haven't been able to do so successfully very often, because it was so much more effective to focus on advertizing than on good movies.
The old way was a case of a degenerate strategy which sucks for everyone but is successful; using a more pleasent strategy just isn't cost effective. If people ignore ads and hear whether movies are any good from their friends, there is a much better chance of good movies not flopping in the box office like they have before.
The main SCO complaint is against IBM, but this has little to do with that claim. This goes to the lawsuits against SCO for claiming in the press to own things they do not and that things which are properly licensed are not. The licensing dispute between two software companies started this mess, but it is no longer the real issue; SCO may not survive to come to the trial on the original claims if the claims on which their stock manipulation is based are found to be fraudulent.
The issue is that the game is not designed to limit the advantage that people take of their edge. In the long run, the house will do better normally, because they have better odds overall. But the odds fluctuate, and the players set the stakes. This means that the advantage is on the players' side if there are skilled players, because the stakes will be negligable while the house has the advantage, and huge while the players do.
If the players are allowed to use skill, the house has to be allowed to use skill as well, or they'll go broke. The only choice the house makes in blackjack is whether to play at all, which means that the only way the house can use skill to compete with player skill is by refusing to play with people who are too good or well-informed, and kick out the people who count cards.
It's not terribly fair to play smart at a game where your opponent is prohibited by the rules from playing smart. Casinos aren't designed for skill but for luck; if you want to use intelligence, you should be playing poker with a group of players on equal footing, or you need to use some of the skill on escaping notice, which is the game of skill that you and the house can both play.
If the cables coming into the box are in as bad shape and as out-of-date as they sound, there's not much you can do on your own. Chances are that the utility will need to be involved in reliably shutting off power to some of this mess, at which point you really need someone licensed if you expect to have it turned back on eventually.
For that matter, the FCC rules don't say that you can't make a profit on this upgrade, just that the fees be "fair and reasonable". This means that, if you only need a minor upgrade to deal with it, you can charge a "fair and reasonable" fee which goes directly to profit. This, of course, rewards the companies that were planning ahead for this sort of thing at the expense of companies like (evidentally) Sprint, which is good, because it was presumably the companies who could handle portability easily that got the FCC to think this was possible, and thus caused it all to happen. Assuming we like cell number portability, the companies that pushed it should get rewarded at the expense of the companies that had to resist it.
It was the biggest blackout, in terms of size (which is what "big" means, after all). Of course, it's not the size that matters...
For a blackout to be bad, it has to come at a time when electricity is really important, or go on for a long time, or people have to be really unprepared, or a lot of equipment has to be damaged as the system goes down, or something like that. Having everyone lose power at the same time is probably actually better; if it's not at the same time, you'll be waiting on people in other places. If everyone loses power together, everyone (who isn't working on it) just takes a long weekend.
One criterion for a good solution to a problem is that the proof allow you to establish the result more quickly than you could establish the result without the proof; it is easier to verify the proof that it is to generate it.
This work merely demonstrates the result; it does not do so in a way that's faster to verify than it was to generate. In general, the proofs of open questions tend to involve work which extends related theories; proving Fermat's Last Theorem, for example, involved developing and relating a number of fields of mathematics which are turning out to be interesting for cryptography, even though there aren't really any interesting consequences to FLT.
On the other hand, this was only for a single size, which had just been bothering people. There's nothing really theoretically significant to it. For odd sizes and larger multiples of 4, it has already been solved; the other even sizes will require another proper proof.
It's not an issue with calculating it, it's the fact that, while the odds are in your favor, the worst case is really bad. Would you rather buy something for $5, or have a 1 in 10,000 chance of paying $10,000 for it? Sure, the first is more expensive on average, but for the second, you have to be able to afford $10,000 in order to be safe. The insurance company is going to play 100,000 times, so they can take advantage of the fact that it will average out; you play only once, so it doesn't average out-- it's either good or really bad.
The other thing is that insurance companies get bulk discounts on a lot of the things that payments go towards. It costs them less to get your car repaired, because they have a lot of car repair business, than it would cost you for the same repair.
The difference between deserts and plains is not the amount of water that lands then, it's the way the water is retained by the ground. With dirt, water that falls on it gets absorbed in the dirt and is available to plants in the area for a while; with dust, water that falls on it is not absorbed much (there's a lot less water in wet sand than in mud), so you get flooding and then the water seeps down to the bedrock. The other thing about dust is that it doesn't clump together, but blows away.
Where does dirt come from? The right mix of plants and other organisms in the area. It's incredibly complicated stuff, and needs maintenance (by the organisms) or it loses its properties and turns to dust. Some people have guessed that we'll be able to prolong our lives indefinitely, travel regularly through the solar system, build space elevators, etc., before we can fabricate dirt in some other way than taking a bit of it at a time from a place with lots of stuff growing.
Africa gets plenty of rain. It causes landslides and floods, but not much else (in the desert regions, that is; there are also jungles which have plenty of moisture). As a side note, the dust from the Sahara is nasty stuff, and blows all around northern Africa, and actually causes respiratory problems in South America in bad years.
I think that additional input methods will add to, rather than replacing, existing ones, when they become common. It would be relatively effective to use eye tracking to move the keyboard focus; more predictable than Tab or arrows, and faster than the mouse. The window you're typing into is the window you're looking at. It would also be good for 3D games, where you could control where you're looking with your eyes.
Voice recognition will never replace the keyboard entirely, because it will never get sufficiently accurate for commands. People can't understand what you're saying well enough to follow directions without interaction; there's no reason computers would be able to, especially because you just don't enunciate enough that the commands are unambiguous. On the other hand, voice recognition will soon be good enough for bulk text entry; you'll dictate your slashdot posts, and edit and submit them with the keyboard.
Slashdot Mirror
Paper?
Software gets delayed when it's not done when they thought it would be, but it's not nearly 2005 yet. Putting off the next release for more than 2 years is just begging to fall behind. Just about anyone else would at least plan to have some sort of release before then, even if it doesn't have every feature in the roadmap. Of course, some projects might not actually manage the scheduled release in 2 years, but that would be due to not estimating correctly in advance.
Where I live, the guy who calls for Patrolmen's donations sounds on the phone like you'd expect a local cop to sound (deep voice, local accent and idiom; the sort of voice that makes you expect a friendly but over-eager clap on the back). I wouldn't be too surprised if the caller wasn't a real police officer, but I think that friends or family are more likely than professional telemarketters.
Ticketting probably only applies to fund-raising speed traps in your town, since they actually care about people who are causing traffic hazards, and non-local cops don't care about your donation anyway. In my town, at least, traps are largely aimed at people commuting through the town, anyway, so having a local parking sticker is probably just as good as a donation sticker.
Certainly a motivated student can do better with a set of good coursework and no school than with a lousy course (or, more likely, a program that doesn't have a course that applies).
I just wanted to point out that courses often have a substantial component that isn't purely information, which would need to be provided locally for the course to really work at all. I actually feel that MIT's coursework isn't necessarily better than other schools; what makes MIT work is that the students have access to really smart and experienced people, as course staff, as other students in the class, and as students who have taken the class before. This means that MIT's coursework can be somewhat more advanced and still be possible.
My somewhat cynical guess is that the courses with information online are mostly courses which rely somewhat heavily on things other than the coursework.
It would be interesting to try to replicate MIT courses online with a group of other motivated students at your local school, and, potentially, some sort of group communication with students in other places doing the course at the same time. I think that adding local or online resources to the MIT coursework would be much more effective than trying to do the MIT coursework individually.
6.170 is a team course; a substantial part of the point is that it is more work than one person can do in a term, so you will have to work successfully with other students. So if you want to replicate the experience outside of MIT, you'll want to find 2-3 like-minded friends when you try it. Of course, you could do it when you had 3-4 times as much time to spend on it as a student taking a full courseload, but then you'd be missing half of the point: good design means that other people will be able to make productive use of your code.
I suspect that the previous poster meant to address people who aren't testing, not people who aren't using it.
There are two factors that suggest that it'll be stable at about the time it is debugged:
A lot of the things which differ between 2.4 and 2.6 with respect to drivers are the APIs of some important things they depend on. The new APIs are less error-prone. This suggests that a driver will work correctly if it worked before and has been updated, and it won't compile otherwise.
Drivers don't get updated until the interfaces are stable and the kernel otherwise works well enough for the people who maintain those drivers to pay attention to the new version. This means that the core kernel will be pretty stable before drivers get updated, and when drivers get updated, the core kernel works.
Of course, new drivers that compile won't necessarily work, and new optional features that compile may not yet be stable (many of the power management bits, for instance). But I think that everything from 2.4 will be working when all of the modules compile for 2.6, due to these factors.
Of course, scheduler tuning is somewhat unrelated, and might delay the release.
Games are frequently targetted at males, but the games that last well are frequently not. I suppose that the scarcity of games that aren't annoying to women would tend to lead to women playing those games longer, meaning that they'd do better. The demand seems to be 41% of the market, and the supply is clearly much lower, even ignoring the fact that 41% is only the portion of the potential market which finds something to play.
Regardless, it seems to me that games that aren't targetted at males (or anyone else in particular) generally last better. Generally they don't just involve flashy graphics, but have some sort of interesting mechanic, room for skill, lots of amusing and obscure things to find, and so forth.
Tomb Raider is the example you give of a game with a female sex object, but it turns out to be reasonably popular with women, and actually has clever puzzles. Of the dozens of others you refer to, the ones without a substantial female following are largely not worth mentioning by name. The only vastly successful game I can think of that has managed to not have much of a female following (to my knowledge) is GTA, which probably only avoids having a female following by being somewhat gratuitously offensive and not having a female protagonist.
The real issue with the color output from LEDs is not that the color is wrong (since you can change that by changing the portions of different color output), but that they produce thin lines of spectrum, rather than the black body curves that incandescents produce. While your eyes can't tell the difference directly (since you only have three different colors of perception), surfaces respond differently to different wavelengths in such a way that light that looks the same to you makes surfaces look different. This means that LED light looks artificial in a way that incandescent light does not.
The only way of getting a wide spectrum of light is to have an object glow with heat, where the energy released per photon varies chaotically, rather than using a process that outputs individual photons which will only produce light at the wavelengths that correspond to energy gaps. Glowing with heat is lower efficiency than emitting individual photons.
I suspect that LEDs will become more popular in step with paint formulas that look good (and look right) under LED light, and also with people coming to expect LED light more.
Earl Grey doesn't need lemon, because it's got bergamot. If you need to put more citrus in your Earl Grey, you should get stronger Earl Grey. If you're going to put lemon in it, get a nice Orange Pekoe instead.
With a reasonable installer, installing new Linux software is a matter of determining the name of a piece of software you want. "apt-get install " should suffice, and shouldn't need anything other than your root password to complete. Good software packaging isn't exclusive to Windows, these days.
TFM generally doesn't have any information on installing stuff, although there's generally a lot on getting things configured the way you want, which is the stuff that the InstallShield wizard wouldn't support.
The self-documenting aspect of the community actually makes "RTFM" useful, because there is documentation written by people who have had your problem for the community to point you at. It's more helpful than people trying to write a new explanation of the solution every time someone has a question. Asking your question and getting told "RTFM" is sometimes more effective than trying to determine for yourself whether TFM will answer your question, or even finding the applicable documentation.
It also starts the dialogue about TFM, which may be out of date or have gotten lost somewhere, as was reported on linux-kernel recently about some documentation.
Ideally, ALSA would support mixing regardless of whether the hardware did it (that is, if the hardware or driver doesn't support it, it would be done in software). Probably the right way to handle it would be for ALSA to do it in userspace if the hardware doesn't support it.
Having only some machines need an external mixer (with a different API) is just a bad idea.
For example, Fox released "Bend It Like Beckham", which is actually a really good movie. But it's not the sort of movie that makes good advertizing (the jokes are only funny in context, there aren't any explosions, and the plot doesn't make a good single line), so it's only made $30 million in the US, less than the usual openning weekend, total since March.
Fox (like the other studios) does make good movies, but good movies that don't make good ads don't make much money. People tend to decide first to go to the movies, and then what to see. Unfortunately, people then decide to see bad movies with good ads, rather than good movies that are obscure, which means that they have to compete with the other studios who pump bad movies.
If you read the article (rather than just the blurb), nowhere do the movie people actually say that this is a bad thing, that they don't like this turn of events, or that they want to do anything to change it.
It could well be a good thing overall, such that they can release good movies with staying power rather than going for glitzy special effects that make good ads. The movie business, unlike the music business, actually likes to produce good stuff, but they haven't been able to do so successfully very often, because it was so much more effective to focus on advertizing than on good movies.
The old way was a case of a degenerate strategy which sucks for everyone but is successful; using a more pleasent strategy just isn't cost effective. If people ignore ads and hear whether movies are any good from their friends, there is a much better chance of good movies not flopping in the box office like they have before.
The main SCO complaint is against IBM, but this has little to do with that claim. This goes to the lawsuits against SCO for claiming in the press to own things they do not and that things which are properly licensed are not. The licensing dispute between two software companies started this mess, but it is no longer the real issue; SCO may not survive to come to the trial on the original claims if the claims on which their stock manipulation is based are found to be fraudulent.
The issue is that the game is not designed to limit the advantage that people take of their edge. In the long run, the house will do better normally, because they have better odds overall. But the odds fluctuate, and the players set the stakes. This means that the advantage is on the players' side if there are skilled players, because the stakes will be negligable while the house has the advantage, and huge while the players do.
If the players are allowed to use skill, the house has to be allowed to use skill as well, or they'll go broke. The only choice the house makes in blackjack is whether to play at all, which means that the only way the house can use skill to compete with player skill is by refusing to play with people who are too good or well-informed, and kick out the people who count cards.
It's not terribly fair to play smart at a game where your opponent is prohibited by the rules from playing smart. Casinos aren't designed for skill but for luck; if you want to use intelligence, you should be playing poker with a group of players on equal footing, or you need to use some of the skill on escaping notice, which is the game of skill that you and the house can both play.
If the cables coming into the box are in as bad shape and as out-of-date as they sound, there's not much you can do on your own. Chances are that the utility will need to be involved in reliably shutting off power to some of this mess, at which point you really need someone licensed if you expect to have it turned back on eventually.
For that matter, the FCC rules don't say that you can't make a profit on this upgrade, just that the fees be "fair and reasonable". This means that, if you only need a minor upgrade to deal with it, you can charge a "fair and reasonable" fee which goes directly to profit. This, of course, rewards the companies that were planning ahead for this sort of thing at the expense of companies like (evidentally) Sprint, which is good, because it was presumably the companies who could handle portability easily that got the FCC to think this was possible, and thus caused it all to happen. Assuming we like cell number portability, the companies that pushed it should get rewarded at the expense of the companies that had to resist it.
It was the biggest blackout, in terms of size (which is what "big" means, after all). Of course, it's not the size that matters...
For a blackout to be bad, it has to come at a time when electricity is really important, or go on for a long time, or people have to be really unprepared, or a lot of equipment has to be damaged as the system goes down, or something like that. Having everyone lose power at the same time is probably actually better; if it's not at the same time, you'll be waiting on people in other places. If everyone loses power together, everyone (who isn't working on it) just takes a long weekend.
One criterion for a good solution to a problem is that the proof allow you to establish the result more quickly than you could establish the result without the proof; it is easier to verify the proof that it is to generate it.
This work merely demonstrates the result; it does not do so in a way that's faster to verify than it was to generate. In general, the proofs of open questions tend to involve work which extends related theories; proving Fermat's Last Theorem, for example, involved developing and relating a number of fields of mathematics which are turning out to be interesting for cryptography, even though there aren't really any interesting consequences to FLT.
On the other hand, this was only for a single size, which had just been bothering people. There's nothing really theoretically significant to it. For odd sizes and larger multiples of 4, it has already been solved; the other even sizes will require another proper proof.
It's not an issue with calculating it, it's the fact that, while the odds are in your favor, the worst case is really bad. Would you rather buy something for $5, or have a 1 in 10,000 chance of paying $10,000 for it? Sure, the first is more expensive on average, but for the second, you have to be able to afford $10,000 in order to be safe. The insurance company is going to play 100,000 times, so they can take advantage of the fact that it will average out; you play only once, so it doesn't average out-- it's either good or really bad.
The other thing is that insurance companies get bulk discounts on a lot of the things that payments go towards. It costs them less to get your car repaired, because they have a lot of car repair business, than it would cost you for the same repair.
So the resulting cells are human cells, except that tracing the maternal line using the standard techniques would... "Your mother was a hamster!"
The difference between deserts and plains is not the amount of water that lands then, it's the way the water is retained by the ground. With dirt, water that falls on it gets absorbed in the dirt and is available to plants in the area for a while; with dust, water that falls on it is not absorbed much (there's a lot less water in wet sand than in mud), so you get flooding and then the water seeps down to the bedrock. The other thing about dust is that it doesn't clump together, but blows away.
Where does dirt come from? The right mix of plants and other organisms in the area. It's incredibly complicated stuff, and needs maintenance (by the organisms) or it loses its properties and turns to dust. Some people have guessed that we'll be able to prolong our lives indefinitely, travel regularly through the solar system, build space elevators, etc., before we can fabricate dirt in some other way than taking a bit of it at a time from a place with lots of stuff growing.
Africa gets plenty of rain. It causes landslides and floods, but not much else (in the desert regions, that is; there are also jungles which have plenty of moisture). As a side note, the dust from the Sahara is nasty stuff, and blows all around northern Africa, and actually causes respiratory problems in South America in bad years.
I think that additional input methods will add to, rather than replacing, existing ones, when they become common. It would be relatively effective to use eye tracking to move the keyboard focus; more predictable than Tab or arrows, and faster than the mouse. The window you're typing into is the window you're looking at. It would also be good for 3D games, where you could control where you're looking with your eyes.
Voice recognition will never replace the keyboard entirely, because it will never get sufficiently accurate for commands. People can't understand what you're saying well enough to follow directions without interaction; there's no reason computers would be able to, especially because you just don't enunciate enough that the commands are unambiguous. On the other hand, voice recognition will soon be good enough for bulk text entry; you'll dictate your slashdot posts, and edit and submit them with the keyboard.