I believe you only need the clockevents portion of the new feature for that (except that you mean usleep and 5 microseconds; it's windows where the unitless sleep function is milliseconds). The first part was to have accurate code inside the kernel for "give an interrupts at this list of arbitrary times"; the second part was to make the scheduler use this mechanism for routine task switching. The big advantage of the second part is that, if nothing is going to be able to run in the next ten minutes (without external input), the CPU can suspend for ten minutes instead of waking up every millisecond or ten. And, if you've got a big virtual server, running a hundred idle kernels, those idle kernels take 0 CPU.
The third part, coming in a few kernel versions, is having the scheduler schedule different tasks for different durations according to policy and optimization. If one task is supposed to get 70% of the processor time and another is supposed to get 30%, the kernel can alternate between then, running the first for 7ms and the second for 3ms, instead of using a complicated pattern of 1ms slices.
It's a neat system, but you don't really want to use one. It's a laptop designed for kids, which means that the keyboard is uncomfortably small for adult hands, and the screen is too small to look at from an adult's typing distance and torso length. It's hard to tell from the photos, because it's got the normal laptop proportions, but it's actually scaled down to kid size.
I've actually had the opposite experience with Gentoo. I've done all sorts of wrong things with it, and had all sorts of things wrong, and the system has continued working perfectly despite that. (The main exception being qmail, which never fails to be running, but doesn't necessarily, you know, deliver mail.)
It's often difficult to get things built and configured the way you want (what, I need the xml USE flag for php for this? Time to rebuild it... And I need to hand-edit an apache config file to have a https-only web app), but until things are set up, they're clearly not working, and once they're working, they stay working.
My coworker's accidentally upgraded his Debian stable machine this week, and various things were actually not running for a couple of days, and he had to pay actual attention to working on the system for non-trivial amounts of that time, despite the fact that it had been working a few days before.
The closest I've had to that level of breakage is when I accidentally lost a CUPS configuration in an upgrade because I didn't realize it was locally modified (by another co-worker). Next closest is having qmail-local not restart (actually, not stop so it could be restarted) when I switched from "qmail" to "netqmail", an operation which the system automatically put off until I told it to really do it, because I had an evening to spend getting it working again.
On occasion, I've had new versions of things just not work (mostly drivers for my laptop). In these cases, the solution has been to mask the version that it just installed, and emerge again.
I agree that Gentoo exposes you to exactly what's going on with the system a lot more than other distributions, and I don't know if that means I'm just not breaking things due to knowing what not to mess with, but it seems to me to just not break very much. And when it's going to break, it tells you it's going to break, and you can tell it not to (e.g., upgrading from mysql-4.0).
That law can't be right, because it would mean that the correct kerning wouldn't depend on the letters at all. Give each letter a nominal bounding box, and use this box for spacing. The "before" area is the area between the letter and the left edge; the "after" area is on the right. The area between two letters is the total of the first one's "after" area and the second one's "before" area. If you want the area to be a particular constant, adjust all the left edges so that all of the letters have half that constant as their "before" areas, and adjust the right edges so all the letters also have half that constant as their "after" areas. Obviously, then, all of the spacing is perfect by the law, regardless of the letters involved. (Okay, so I'm ignoring different heights of adjacent space; there's more height in the area in "ll" than "mm", but that translates to constant space removals between adjacent right and left ascenders or descenders.)
But correct kerning depends on the letter pairs, with some examples mattering a lot, and some being less important. So this law, as stated, is clearly wrong. Correct kerning takes into account the fact that "AV" has room to overlap vertically, while "VV" and "AA" don't, so the particular shapes actually matter. Any possible good rule has to be non-linear to account for this, and certainly trickier than the rule given. I suspect that there isn't a good rule known (other than "get sombody with a good eye to adjust it"), or fonts would just have a "crowdedness" parameter, and the rendering engine would use the rule and the parameter to generate ideal kernings for all pairs of characters on the fly.
I didn't think you wanted no gun laws. I was talking about the people who write the gun laws that actually get passed, who seem to exclusively write pointless and unconstitutional ones. I think that this is because the debate largely gets framed between people making knee-jerk reactions to rare events and people opposing those knee-jerk reactions, and little attention goes into proposing actual solutions, even when everybody who've looked into the statistics and thought about the issues knows what a reasonable solution would be. You let yourself end up in a debate between something implicit you don't agree with and something explicit you firmly oppose, and I think that does our position a disservice.
What I don't understand is why nobody takes the "well-regulated militia" part seriously. Just arming arbitrary people won't help, and could hurt, depending on how jumpy people are. (E.g., I bet, if they repealled the law today, somebody would shoot somebody for gesturing too wildly next week, because people will be paranoid.) An elaborate gun safety and situation awareness program which is non-trivial but accessible to students, as a requirement for being permitted to carry a (concealed) weapon on a campus, with the whole thing widely publicized, would increase the likelihood of somebody taking down an attacker like this.
Having a solution that's all {someone} is worse than a solution that's part {someone} and part {someone else}, regardless of who they are, so long as they're separate, because it's a lot easier to replace a component if there's a documented interfaces between components that is mutually agreed upon by different organizations. That is, if Google goes bad, it'll be a lot easier to write something that's close enough to Google Calendar for Thunderbird than it was to get away from Microsoft in the first place.
If Grandma is running Linux, she's probably aware that her grandson takes care of all that sort of stuff. If it's a Mac, she knows that Apple takes care of everything.
Except, of course, that GTAIII had aircraft. Actually taking off and flying the Dodo was somewhat impractical, but possible. While you could actually fly it into a building, it didn't damage the building (since nothing damaged buildings in GTAIII), and if it blew up, the explosion wasn't any more significant than a car blowing up.
Presumably, he didn't actually make a semiconductor device whose stated purpose was to emit light, but just considered it a side effect of certain configurations. Vacuum tubes emit light as a side effect when overloaded, but this is undesirable most of the time. He probably didn't realize that the light emission from semiconductors could be useful.
Since they don't define "public computer network", I can't see any reason not to consider the PSTN a "public computer network" (T1 and ISDN internet connections are actually blocks of routed phone numbers dedicated to last-mile internet routes, and these were well-established in '95). If that is true, every PSTN phone call is prior art.
Now, they obviously didn't mean to include POTS as a computer network, but if they're too vague to exclude it by some distinguishing feature, then it's impossible to determine what their patent should cover.
In my office, there are two people with dual monitors and the rest have single monitors. The people with dual monitors have things they want to be able to refer to without other stuff going out of sight. Personally, I have only one monitor because the other space is reserved for a scope, which I use for debugging a lot more than I'd use another screen.
Of course, I have small windows for my code (80 columns wide, in a little font), and so there's room on the screen for a datasheet in a window behind it, such that, with the margins next to tables and figures, the code don't overlap those at all, and only a bit of the text is hidden. And I generally want to finish reading the section of the datasheet, and think about it while I flick the mouse (focus follow mouse) to the editor window, so that's not a big deal.
And they don't seem to have provided any evidence that they did try to serve PJ with a subpoena. They provide ample evidence that they claim to want to serve her with a subpoena, and that she knows they claim to want to serve her with a subpoena, and she and try agree that she has not actually been served with a subpoena, but there's no evidence that they actually issued a subpoena on January 30, nor that such a subpoena specified February 21, nor that she heard over any of the listed channels about the necessary information to receive a subpoena or show up for the deposition.
And, of course, Pacer evidently hasn't seen anything in SCO v. Novell, where the subpoena would actually be from. All of the claims in this filing about PJ avoiding service are actually completely irrelevant to the motion, because this court isn't even asked to care about whether she shows up.
It's actually the city of Boston that thinks everything is a bomb. State government and neighboring cities are a lot more sane.
And all of those documents are available in PDF, which is what they always said they'd use for distributing documents to the public (i.e., people who don't need to edit the documents).
The energy benefit is supposed to be that people have their lights on at night for less time, because they go to bed at a time which is closer to the time the sun sets. In the morning, they don't need to turn on their lights at all because it's light already, because it's summer. Of course, this requires that it actually be light when people are getting up, so there's limited gains for making DST last longer.
Viacom is being sued for copyright infringement, too (by people whose copyrighted works were the subject of misguided DMCA removal demands to YouTube). I like the idea that these plaintiffs could get, not only all of Viacom's computers, but also all of Viacom's employees' computers.
Unfortunately for that, you need to convince the judge that there's a good reason to think that you'll turn up evidence for the crime you already have a good reason to think has been committed.
Really, Firefox should drop Preferences (which doesn't have any of the really useful stuff, and is hard to find things in anyway), and just make about:config nicer. Preferences has a bunch of stuff sorted into non-orthogonal categories that are often vague ("main", "content", "advanced") or inappropriate (the choice of font is clearly a matter of presentation, but it's under "content").
The configuration engine should be extended to keep track of what the options are for things with options (and, in general, the effective types of settings, like "font" instead of "string"), what the description is, and keywords you might be searching for. Then ditch the "status" column (the same info is given by boldness) and the type (you only care when changing it), show the descriptive text instead of the name, and handle the configuration stuff that's not in about:config similarly (e.g., handlers for various types), and have type-specific widgets. And have the window blank if the search field is empty and "Show All" isn't clicked.
The "with prejudice" statement was for after the plaintiff's technicians examined the defendant's hard drive. They didn't get that far. At this point, the defendants could totally be bluffing, for all they know, so dismissing with prejudice is inappropriate. And it would be hard for the defendants, in a countersuit, to demonstrate that this wasn't an honest and reasonable mistake on the RIAA's part, since they quit at the "sorry, wrong number" stage. They can join the class action if they want.
Ballot initiatives don't have much of a direct effect (although the actual news story I found says that they're still deprioritizing non-felony possession), but one of the commissioners who voted to change the initiative needs to run for re-election in 2008. If anyone plausible wants her job, it probably wouldn't be hard to defeat her on a platform of not second-guessing the electorate and the pot declaration that voters already went for.
The people specifically mentioned in the article are approving AdWords buys. I think Google was probably not benefitting much from these particular people working on side projects (like approving initial cell phone customers?) 20% of the time. I'd like to see an article about people doing data entry for Google, who are happy about the opportunity to spend part of their time entering the data they want to enter, but I suspect that would only appear in the Onion.
Slashdot Mirror
I believe you only need the clockevents portion of the new feature for that (except that you mean usleep and 5 microseconds; it's windows where the unitless sleep function is milliseconds). The first part was to have accurate code inside the kernel for "give an interrupts at this list of arbitrary times"; the second part was to make the scheduler use this mechanism for routine task switching. The big advantage of the second part is that, if nothing is going to be able to run in the next ten minutes (without external input), the CPU can suspend for ten minutes instead of waking up every millisecond or ten. And, if you've got a big virtual server, running a hundred idle kernels, those idle kernels take 0 CPU.
The third part, coming in a few kernel versions, is having the scheduler schedule different tasks for different durations according to policy and optimization. If one task is supposed to get 70% of the processor time and another is supposed to get 30%, the kernel can alternate between then, running the first for 7ms and the second for 3ms, instead of using a complicated pattern of 1ms slices.
It's a neat system, but you don't really want to use one. It's a laptop designed for kids, which means that the keyboard is uncomfortably small for adult hands, and the screen is too small to look at from an adult's typing distance and torso length. It's hard to tell from the photos, because it's got the normal laptop proportions, but it's actually scaled down to kid size.
To the north, you can see what looks like an aging high-school gymnasium.
I've actually had the opposite experience with Gentoo. I've done all sorts of wrong things with it, and had all sorts of things wrong, and the system has continued working perfectly despite that. (The main exception being qmail, which never fails to be running, but doesn't necessarily, you know, deliver mail.)
It's often difficult to get things built and configured the way you want (what, I need the xml USE flag for php for this? Time to rebuild it... And I need to hand-edit an apache config file to have a https-only web app), but until things are set up, they're clearly not working, and once they're working, they stay working.
My coworker's accidentally upgraded his Debian stable machine this week, and various things were actually not running for a couple of days, and he had to pay actual attention to working on the system for non-trivial amounts of that time, despite the fact that it had been working a few days before.
The closest I've had to that level of breakage is when I accidentally lost a CUPS configuration in an upgrade because I didn't realize it was locally modified (by another co-worker). Next closest is having qmail-local not restart (actually, not stop so it could be restarted) when I switched from "qmail" to "netqmail", an operation which the system automatically put off until I told it to really do it, because I had an evening to spend getting it working again.
On occasion, I've had new versions of things just not work (mostly drivers for my laptop). In these cases, the solution has been to mask the version that it just installed, and emerge again.
I agree that Gentoo exposes you to exactly what's going on with the system a lot more than other distributions, and I don't know if that means I'm just not breaking things due to knowing what not to mess with, but it seems to me to just not break very much. And when it's going to break, it tells you it's going to break, and you can tell it not to (e.g., upgrading from mysql-4.0).
That law can't be right, because it would mean that the correct kerning wouldn't depend on the letters at all. Give each letter a nominal bounding box, and use this box for spacing. The "before" area is the area between the letter and the left edge; the "after" area is on the right. The area between two letters is the total of the first one's "after" area and the second one's "before" area. If you want the area to be a particular constant, adjust all the left edges so that all of the letters have half that constant as their "before" areas, and adjust the right edges so all the letters also have half that constant as their "after" areas. Obviously, then, all of the spacing is perfect by the law, regardless of the letters involved. (Okay, so I'm ignoring different heights of adjacent space; there's more height in the area in "ll" than "mm", but that translates to constant space removals between adjacent right and left ascenders or descenders.)
But correct kerning depends on the letter pairs, with some examples mattering a lot, and some being less important. So this law, as stated, is clearly wrong. Correct kerning takes into account the fact that "AV" has room to overlap vertically, while "VV" and "AA" don't, so the particular shapes actually matter. Any possible good rule has to be non-linear to account for this, and certainly trickier than the rule given. I suspect that there isn't a good rule known (other than "get sombody with a good eye to adjust it"), or fonts would just have a "crowdedness" parameter, and the rendering engine would use the rule and the parameter to generate ideal kernings for all pairs of characters on the fly.
I didn't think you wanted no gun laws. I was talking about the people who write the gun laws that actually get passed, who seem to exclusively write pointless and unconstitutional ones. I think that this is because the debate largely gets framed between people making knee-jerk reactions to rare events and people opposing those knee-jerk reactions, and little attention goes into proposing actual solutions, even when everybody who've looked into the statistics and thought about the issues knows what a reasonable solution would be. You let yourself end up in a debate between something implicit you don't agree with and something explicit you firmly oppose, and I think that does our position a disservice.
What I don't understand is why nobody takes the "well-regulated militia" part seriously. Just arming arbitrary people won't help, and could hurt, depending on how jumpy people are. (E.g., I bet, if they repealled the law today, somebody would shoot somebody for gesturing too wildly next week, because people will be paranoid.) An elaborate gun safety and situation awareness program which is non-trivial but accessible to students, as a requirement for being permitted to carry a (concealed) weapon on a campus, with the whole thing widely publicized, would increase the likelihood of somebody taking down an attacker like this.
Having a solution that's all {someone} is worse than a solution that's part {someone} and part {someone else}, regardless of who they are, so long as they're separate, because it's a lot easier to replace a component if there's a documented interfaces between components that is mutually agreed upon by different organizations. That is, if Google goes bad, it'll be a lot easier to write something that's close enough to Google Calendar for Thunderbird than it was to get away from Microsoft in the first place.
If Grandma is running Linux, she's probably aware that her grandson takes care of all that sort of stuff. If it's a Mac, she knows that Apple takes care of everything.
Except, of course, that GTAIII had aircraft. Actually taking off and flying the Dodo was somewhat impractical, but possible. While you could actually fly it into a building, it didn't damage the building (since nothing damaged buildings in GTAIII), and if it blew up, the explosion wasn't any more significant than a car blowing up.
Presumably, he didn't actually make a semiconductor device whose stated purpose was to emit light, but just considered it a side effect of certain configurations. Vacuum tubes emit light as a side effect when overloaded, but this is undesirable most of the time. He probably didn't realize that the light emission from semiconductors could be useful.
Since they don't define "public computer network", I can't see any reason not to consider the PSTN a "public computer network" (T1 and ISDN internet connections are actually blocks of routed phone numbers dedicated to last-mile internet routes, and these were well-established in '95). If that is true, every PSTN phone call is prior art.
Now, they obviously didn't mean to include POTS as a computer network, but if they're too vague to exclude it by some distinguishing feature, then it's impossible to determine what their patent should cover.
In my office, there are two people with dual monitors and the rest have single monitors. The people with dual monitors have things they want to be able to refer to without other stuff going out of sight. Personally, I have only one monitor because the other space is reserved for a scope, which I use for debugging a lot more than I'd use another screen.
Of course, I have small windows for my code (80 columns wide, in a little font), and so there's room on the screen for a datasheet in a window behind it, such that, with the margins next to tables and figures, the code don't overlap those at all, and only a bit of the text is hidden. And I generally want to finish reading the section of the datasheet, and think about it while I flick the mouse (focus follow mouse) to the editor window, so that's not a big deal.
1. Install CVS
2. Check out my "dotfiles" modules
3. Run a script in it to make symlinks
4. Log out and log back in
If I'm building a new box from scratch, rather than setting up my account on an existing box, add:
5. Install things on the hard drive (while doing other stuff)
6. Set up my dotfiles again on the hard drive.
7. Reboot without the live CD
And they don't seem to have provided any evidence that they did try to serve PJ with a subpoena. They provide ample evidence that they claim to want to serve her with a subpoena, and that she knows they claim to want to serve her with a subpoena, and she and try agree that she has not actually been served with a subpoena, but there's no evidence that they actually issued a subpoena on January 30, nor that such a subpoena specified February 21, nor that she heard over any of the listed channels about the necessary information to receive a subpoena or show up for the deposition.
And, of course, Pacer evidently hasn't seen anything in SCO v. Novell, where the subpoena would actually be from. All of the claims in this filing about PJ avoiding service are actually completely irrelevant to the motion, because this court isn't even asked to care about whether she shows up.
It's actually the city of Boston that thinks everything is a bomb. State government and neighboring cities are a lot more sane.
And all of those documents are available in PDF, which is what they always said they'd use for distributing documents to the public (i.e., people who don't need to edit the documents).
The energy benefit is supposed to be that people have their lights on at night for less time, because they go to bed at a time which is closer to the time the sun sets. In the morning, they don't need to turn on their lights at all because it's light already, because it's summer. Of course, this requires that it actually be light when people are getting up, so there's limited gains for making DST last longer.
Viacom is being sued for copyright infringement, too (by people whose copyrighted works were the subject of misguided DMCA removal demands to YouTube). I like the idea that these plaintiffs could get, not only all of Viacom's computers, but also all of Viacom's employees' computers.
Unfortunately for that, you need to convince the judge that there's a good reason to think that you'll turn up evidence for the crime you already have a good reason to think has been committed.
Fortunately for them, they were rescued by a quick-thinking, emacs-using paramedic, who restored them all unharmed with a control-y.
The internet's going to be shut down for spring cleaning for most of that time anyway, so it's just as well.
Really, Firefox should drop Preferences (which doesn't have any of the really useful stuff, and is hard to find things in anyway), and just make about:config nicer. Preferences has a bunch of stuff sorted into non-orthogonal categories that are often vague ("main", "content", "advanced") or inappropriate (the choice of font is clearly a matter of presentation, but it's under "content").
The configuration engine should be extended to keep track of what the options are for things with options (and, in general, the effective types of settings, like "font" instead of "string"), what the description is, and keywords you might be searching for. Then ditch the "status" column (the same info is given by boldness) and the type (you only care when changing it), show the descriptive text instead of the name, and handle the configuration stuff that's not in about:config similarly (e.g., handlers for various types), and have type-specific widgets. And have the window blank if the search field is empty and "Show All" isn't clicked.
The "with prejudice" statement was for after the plaintiff's technicians examined the defendant's hard drive. They didn't get that far. At this point, the defendants could totally be bluffing, for all they know, so dismissing with prejudice is inappropriate. And it would be hard for the defendants, in a countersuit, to demonstrate that this wasn't an honest and reasonable mistake on the RIAA's part, since they quit at the "sorry, wrong number" stage. They can join the class action if they want.
Ballot initiatives don't have much of a direct effect (although the actual news story I found says that they're still deprioritizing non-felony possession), but one of the commissioners who voted to change the initiative needs to run for re-election in 2008. If anyone plausible wants her job, it probably wouldn't be hard to defeat her on a platform of not second-guessing the electorate and the pot declaration that voters already went for.
The people specifically mentioned in the article are approving AdWords buys. I think Google was probably not benefitting much from these particular people working on side projects (like approving initial cell phone customers?) 20% of the time. I'd like to see an article about people doing data entry for Google, who are happy about the opportunity to spend part of their time entering the data they want to enter, but I suspect that would only appear in the Onion.
You're using "good" and "windows" in the same noun phrase? You must be new here.