There was a direct correlation between accuracy of the exit poll in a given precinct with the balloting technique used. Where a paper trail exists, the exit polls were statistically more accurate than in precincts where electronic voting was used (no paper trail.)
In my opinion, that's a clear indicator of something being wrong with the vote counting in those precincts, not the exit polls. I have the raw data of exit poll numbers from several states, voting method in precincts, and the final results. There's a clear correlation. However, those data files aren't handy here - google it yourself, or follow the links from this Wikipedia article (I know -- not the most reliable source, but it's a starting point): http://en.wikipedia.org/wiki/2004_U.S._presidentia l_election_controversy,_exit_polls.
I was only commenting on the parent, who seemed to imply that there was a very large difference between 1080p and 1600x1200 -- I'm not gonna judge the cards available. He just annoyed me by ignoring basic math.;)
Err, you do realize that 1600x1200 is about 93% of the resolution for 1080p (in total pixels rendered), right? It's a very, very small difference. Not defending the PC gamers vs. the Console gamers (I actually use both, and tend to prefer the consoles for most games, but I'll be dead before I play Civ4 without a mouse and keyboard.)
Use a traffic shaper. You can do it in Linux, but it's easy as hell in OpenBSD. It'll fix that problem.
I configured a pair of them to allow 40 - 50 people, who were used to campus 100M ethernet, to live on the wrong side of an 802.11b bridge for a year without killing each other or their network support (me.) Their file servers, web servers, etc. were all on the other side of that bridge, along with the rest of the interweb. Thank $DEITY that they're back in a reasonable location and I don't have to deal with their refusal to move the microwave away from the AP...
I had to use a pair of shapers since you can't really shape *incoming* traffic, so both sides of the bridge were shaped. In this situation, you would get 90% of the benefit with only the outgoing traffic prioritized, in particular since it's an asymmetric link.
You're not noticing the obvious. Who defines "owner?" That definition works perfectly, if Microsoft and the media companies are the owner of the computer system.
There's one other aspect to season passes. Say I want it to record BSG. I also have something else set to record at the same time one week. If I override it, and don't let it record because I'm watching something else (or I tell it to record something else), it will pick up that episode of BSG when it repeats for the second time Tuesday, at 1AM. You queue things based on priority, but it's smart enough to not record something of a high priority that will repeat in three hours so it can pick up a low-priority show that doesn't repeat. I'm really curious - can the DirectTV box do this?
Oh, and BTW -- it can do a director wish list that will record everything directed by a certain people, or starring a certain actor, or where the title contains a word, etc.
Sun is horrible about power cables. The worst part is that there's basically no way to say "Don't bloody send them," and we can't use the ones they provide. They send a wall plug cable, we need C13 cables for the rack power management we use. Each power cable, no matter what you buy, is shipped in an oversized box, from a different warehouse (usually). Frackin' ridiculous.
I actually do that (X11 VMware console) from my G5 desktop. The trick is that you need to enable trusted X11 forwarding. Either have "ForwardX11Trusted yes" in your ssh configuration (~/.ssh/config), or use 'ssh -Y hostname' to connect to the remote host and have vmware work.
In the city where I live, the lots are 40' wide. Postage stamps. Assuming that the "65' effective radius" is correct, that means that it's probably designed to be intolerable inside that radius. That means it's still annoying outside. Now, from his back yard, he's made two yards on each side intolerable, probably at least one yard more likely three on the other side of the alley, and it's probably annoying for a good distance past that.
Representative Hogan is out of touch with the people of Minnesota. Despite the well-known danger, he voted against a measure that would protect our children from sexual predators in schools and libraries. Many children have been abducted or assaulted by predators already, and he doesn't want to do anything about it. You can do something about it. Protect your children this November.
I remember quite a few ads that were at least this bad in 2004, and I'm expecting it to be worse in Minnesota this year. Note that I never mention the word "vote" or any other candidate. That makes this an "issue" ad, rather than a direct campaign ad, and it's now under looser finance restrictions. The campaigns do a lot of research. If it didn't work on a large, vulnerable (ignorant) part of the population, they wouldn't waste money on it.
Sun typically worries more about redundancy than noise. The 10 small fans are hot-swappable and run at ridiculous speeds (and yes, sound like a A320 revving up for takeoff), but I bet the thermal budget allows four of them to be dead at any given time.
Beyond that, if you're developing the model, it would make it easier to see places where the model was incorrect (not realistically proportioned) with a skin that looked like a human body.
Some people are way too worried about nudity. Often, the same people who immediately called their congresscritter after the wardrobe malfunction would have no trouble with little Johnny or Jane seeing 24 or CSI or Law and Order: SVU. Personally, I think either of those are more likely to do damage to a kid than a 500ms flash of nipple.
Besides, why does everyone focus on the nudity from the Superbowl incident rather than the violence? (Wouldn't you consider ripping clothing off of someone else to be an act of violence against that person?)
we're talking about really fundamental stuff like IP subnetting, which you need to understand as soon as you get even a moderately sized network.
Larger companies, like the ones likely surveyed, will have CCNAs or equivalent to do that for you. In most cases, one person isn't responsible for both a network and servers. People with servers will request a jack be connected, just like a desktop tech, and they might get to specify what vlan it's on (for firewall/security zone reasons). They'll be assigned an IP address and a jack will be connected for them -- no knowledge of networking is really required, at least not at the level of being able to figure out that a/27 has a netmask of 255.255.255.248, or what the broadcast address for that subnet would be given the mask and an IP from the network. That information is provided by the network administrator.
I don't disagree that understanding should be required to admin servers, but that's not the reality.
When you're talking about uptime of a server or service at the enterprise level, it's not just time between reboots. If it's a DNS server, and it's not responding to DNS queries, it's down. If it's a mail server that's sending out 500 errors because the spool disk is full, that's downtime.
Now that the pedantic definitions are out of the way, yes, you don't need to reboot to upgrade a package on a debian box. However, dpkg might kill the running service while preparing to upgrade, install the new files, update config files if necessary, and finally restart the service. If you're updating multiple packages, those steps are done by preparing all packages, upgrading all packages, and running the postinstall for all packages (restarting the service). That can take some time, which using a sane definition qualifies as downtime for the server.
From experience, may I strongly recommend a backup before pruning your home and data directories? I don't know how many times I've had the "oh shit" effect doing cleanup of that type...
Re:Sun - Corporate mismanagement at its finest
on
Sun to Cut 5000 Jobs
·
· Score: 1
Sun basically refused to *admit* that they had a NUMA system (E3K - E10K were the earliest non-uniform systems). They did, but damned if they'd say so after spending years bashing the high-end SGI boxes *for being NUMA.* The OS became fully aware of NUMA as of Solaris 7.
To answer the other question, yes, *part* of that is hardware. On Sun hardware, Solaris can tell you that it's DIMM 17 that has the failure, and on more expensive Sun hardware, you can hotswap the CPU or memory. Hell, on some of the hardware you can *upgrade* the CPU to a faster model without shutting down the OS, or add memory to a board with 0 downtime.
However, it can shut down a CPU and stop using it, or stop using a bank of memory on any system. I have no knowledge of any Linux distrubution, or *BSD, that can keep a system with a bad CPU or memory from crashing on the spot.
Re:Sun - Corporate mismanagement at its finest
on
Sun to Cut 5000 Jobs
·
· Score: 1
To clarify, where I work we run a mix of Solaris, Linux, and OS X, with a sprinkling of BSD (ok, and a couple of Windows machines and the old Netware cluster, but that's not my problem). I don't dislike Linux. Most of our new deployments run Linux. That doesn't mean that Solaris sucks, in any way, especially once you've installed the [plug] Blastwave community-provided software, which makes SunFreeware look rather inconsistent and incomplete.
Re:Sun - Corporate mismanagement at its finest
on
Sun to Cut 5000 Jobs
·
· Score: 4, Insightful
Jebus, you people (linux zealots) are nuts. Maybe now, with the current state of the kernel, could it start replacing Solaris in some of the places it really shines. Maybe. Probably not.
Here's the thing. It's really hard to make Solaris crash. I can throw a system load of 80 at a two-processor box and still get a response (enough of one to fix the problem causing a load of 80). It can run on a 216-processor single-system-image NUMA box efficiently, including some "self-healing" properties. Bank of memory throwing correctable ECC errors? Map it out. Processor that has ECC errors in it's cache? Map it out. Hotswap the board containing the processor or memory without a reboot. Users don't notice. On lower-end hardware, like the new AMD-based boxes, it will just map out and stop using the offending hardware until you have a chance to fix it. Isn't it better to have a machine drop from 8G of memory to 4G of memory until you can schedule downtime rather than just crash?
There's another, even larger factor. The government (one of Sun's biggest customers) likes Solaris. A lot. And they especially like Trusted Solaris, for which there's basically no *certified* comparable Linux distro. There's a lot of stuff painted Army green or Navy gray that has Solaris machines inside.
Did Sun mismanage things? Hell, yes. Was the major problem that they didn't throw out 20 years of engineering work to switch to Linux? Hell, no.
Right. Why don't you actually try this before saying that it only makes a show of security, or at least think it through?
The setuid bit would not be copied by a straight cp, and if it was, you still would not be able to set the ownership of/tmp/.us to root. At most, you would create a binary that could be used to su to yourself, and it would not change any of the logging from it, other than making it obvious that you were trying something asinine.
So.... you mean that you want to run Ubuntu (or straight Debian, or Fedora, or Mandrake^H^H^H^Hiva, or...)?
I'm not trying to be insulting, but if you don't want to deal with rebuilding, why do you run Gentoo? I personally don't have the kind of free time that Gentoo seems to requires, so most of the boxes I deal with are Solaris or RHEL.
Slashdot Mirror
There was a direct correlation between accuracy of the exit poll in a given precinct with the balloting technique used. Where a paper trail exists, the exit polls were statistically more accurate than in precincts where electronic voting was used (no paper trail.)
In my opinion, that's a clear indicator of something being wrong with the vote counting in those precincts, not the exit polls. I have the raw data of exit poll numbers from several states, voting method in precincts, and the final results. There's a clear correlation. However, those data files aren't handy here - google it yourself, or follow the links from this Wikipedia article (I know -- not the most reliable source, but it's a starting point): http://en.wikipedia.org/wiki/2004_U.S._presidentia l_election_controversy,_exit_polls.
I was only commenting on the parent, who seemed to imply that there was a very large difference between 1080p and 1600x1200 -- I'm not gonna judge the cards available. He just annoyed me by ignoring basic math. ;)
Err, you do realize that 1600x1200 is about 93% of the resolution for 1080p (in total pixels rendered), right? It's a very, very small difference. Not defending the PC gamers vs. the Console gamers (I actually use both, and tend to prefer the consoles for most games, but I'll be dead before I play Civ4 without a mouse and keyboard.)
Excellent troll, good sir!/>
That would indeed be very interesting.
September has 30 days.
Use a traffic shaper. You can do it in Linux, but it's easy as hell in OpenBSD. It'll fix that problem. I configured a pair of them to allow 40 - 50 people, who were used to campus 100M ethernet, to live on the wrong side of an 802.11b bridge for a year without killing each other or their network support (me.) Their file servers, web servers, etc. were all on the other side of that bridge, along with the rest of the interweb. Thank $DEITY that they're back in a reasonable location and I don't have to deal with their refusal to move the microwave away from the AP... I had to use a pair of shapers since you can't really shape *incoming* traffic, so both sides of the bridge were shaped. In this situation, you would get 90% of the benefit with only the outgoing traffic prioritized, in particular since it's an asymmetric link.
You're not noticing the obvious. Who defines "owner?" That definition works perfectly, if Microsoft and the media companies are the owner of the computer system.
Think about it.
Oh, and BTW -- it can do a director wish list that will record everything directed by a certain people, or starring a certain actor, or where the title contains a word, etc.
Sun is horrible about power cables. The worst part is that there's basically no way to say "Don't bloody send them," and we can't use the ones they provide. They send a wall plug cable, we need C13 cables for the rack power management we use. Each power cable, no matter what you buy, is shipped in an oversized box, from a different warehouse (usually). Frackin' ridiculous.
I actually do that (X11 VMware console) from my G5 desktop. The trick is that you need to enable trusted X11 forwarding. Either have "ForwardX11Trusted yes" in your ssh configuration (~/.ssh/config), or use 'ssh -Y hostname' to connect to the remote host and have vmware work.
In the city where I live, the lots are 40' wide. Postage stamps. Assuming that the "65' effective radius" is correct, that means that it's probably designed to be intolerable inside that radius. That means it's still annoying outside. Now, from his back yard, he's made two yards on each side intolerable, probably at least one yard more likely three on the other side of the alley, and it's probably annoying for a good distance past that.
"Is a grammatically lacking?" Pot, meet kettle. You two will get along well.
I remember quite a few ads that were at least this bad in 2004, and I'm expecting it to be worse in Minnesota this year. Note that I never mention the word "vote" or any other candidate. That makes this an "issue" ad, rather than a direct campaign ad, and it's now under looser finance restrictions. The campaigns do a lot of research. If it didn't work on a large, vulnerable (ignorant) part of the population, they wouldn't waste money on it.
Sun typically worries more about redundancy than noise. The 10 small fans are hot-swappable and run at ridiculous speeds (and yes, sound like a A320 revving up for takeoff), but I bet the thermal budget allows four of them to be dead at any given time.
Beyond that, if you're developing the model, it would make it easier to see places where the model was incorrect (not realistically proportioned) with a skin that looked like a human body.
Some people are way too worried about nudity. Often, the same people who immediately called their congresscritter after the wardrobe malfunction would have no trouble with little Johnny or Jane seeing 24 or CSI or Law and Order: SVU. Personally, I think either of those are more likely to do damage to a kid than a 500ms flash of nipple.
Besides, why does everyone focus on the nudity from the Superbowl incident rather than the violence? (Wouldn't you consider ripping clothing off of someone else to be an act of violence against that person?)
Not my experience with an older version, actually. If that's the case, excellent.
I don't disagree that understanding should be required to admin servers, but that's not the reality.
Now that the pedantic definitions are out of the way, yes, you don't need to reboot to upgrade a package on a debian box. However, dpkg might kill the running service while preparing to upgrade, install the new files, update config files if necessary, and finally restart the service. If you're updating multiple packages, those steps are done by preparing all packages, upgrading all packages, and running the postinstall for all packages (restarting the service). That can take some time, which using a sane definition qualifies as downtime for the server.
From experience, may I strongly recommend a backup before pruning your home and data directories? I don't know how many times I've had the "oh shit" effect doing cleanup of that type...
To answer the other question, yes, *part* of that is hardware. On Sun hardware, Solaris can tell you that it's DIMM 17 that has the failure, and on more expensive Sun hardware, you can hotswap the CPU or memory. Hell, on some of the hardware you can *upgrade* the CPU to a faster model without shutting down the OS, or add memory to a board with 0 downtime.
However, it can shut down a CPU and stop using it, or stop using a bank of memory on any system. I have no knowledge of any Linux distrubution, or *BSD, that can keep a system with a bad CPU or memory from crashing on the spot.
To clarify, where I work we run a mix of Solaris, Linux, and OS X, with a sprinkling of BSD (ok, and a couple of Windows machines and the old Netware cluster, but that's not my problem). I don't dislike Linux. Most of our new deployments run Linux. That doesn't mean that Solaris sucks, in any way, especially once you've installed the [plug] Blastwave community-provided software, which makes SunFreeware look rather inconsistent and incomplete.
Here's the thing. It's really hard to make Solaris crash. I can throw a system load of 80 at a two-processor box and still get a response (enough of one to fix the problem causing a load of 80). It can run on a 216-processor single-system-image NUMA box efficiently, including some "self-healing" properties. Bank of memory throwing correctable ECC errors? Map it out. Processor that has ECC errors in it's cache? Map it out. Hotswap the board containing the processor or memory without a reboot. Users don't notice. On lower-end hardware, like the new AMD-based boxes, it will just map out and stop using the offending hardware until you have a chance to fix it. Isn't it better to have a machine drop from 8G of memory to 4G of memory until you can schedule downtime rather than just crash?
There's another, even larger factor. The government (one of Sun's biggest customers) likes Solaris. A lot. And they especially like Trusted Solaris, for which there's basically no *certified* comparable Linux distro. There's a lot of stuff painted Army green or Navy gray that has Solaris machines inside.
Did Sun mismanage things? Hell, yes. Was the major problem that they didn't throw out 20 years of engineering work to switch to Linux? Hell, no.
The setuid bit would not be copied by a straight cp, and if it was, you still would not be able to set the ownership of /tmp/.us to root. At most, you would create a binary that could be used to su to yourself, and it would not change any of the logging from it, other than making it obvious that you were trying something asinine.
Now go away.
Sorry. After work, where people do seem to be that .. unobservant, it's entirely believable. It does make it hard to spot a joke.
I'm not trying to be insulting, but if you don't want to deal with rebuilding, why do you run Gentoo? I personally don't have the kind of free time that Gentoo seems to requires, so most of the boxes I deal with are Solaris or RHEL.