Because not all images are square. Objects that aren't square (like a star for example), if you wanted to place that star on a page, and didn't have alpha transparency, you could use GIF, but the edges would look weird because it's either all on or all off transparency which creates jaggy edges. With alpha transparency you can make the object's edges more transparent as it gets closer to the edge which will let it look right no matter what type of background you have. Of course there are many other uses, like semi-transparent effects, shadows, etc that aren't easily replicated another way.
These type things allow a webpage to reuse assets over and over across many pages, or in many different sections of the same page, which reduces the amount of image manipulation you have to do, and also reduces the size of the page which is more and more important as devices with low bandwidth become more prominent (like phones/tablets).
it's only a matter of time before attackers start figuring out ways to use those statistical models
They already do. They use attacks that hit the largest number of targets. Using uncommon vulnerabilities would be wasteful when you could attack more common ones.
I can get UnitedHealthOne Saver 80 here in Illinois. Early 40's, non-smoker... $84.52/month. Not the best coverage, but it handles the catastrophic stuff which is all I really care about anyhow. Better insurance with lower deductibles are also available for $240.
I guess that depends on your goal. For example, the average US citizen pays $7,980 per person for healthcare. That's the highest of any country. The who report you gave says the the US is.838 efficient, while the average UK citizen pays $3,438 per person, but is 0.925 efficient.
Now who receives better care? Note, I said better, not better per dollar.
I don't think there are any hospitals with more than 1 CAT scanner. But for comparison, there are 34 CAT scanners per million people in the US, while there is 7 per million in the UK. I think they have it worse.
Why are you talking about customer side routers? What the hell does that have to do with anything?
Granted, I haven't had to deal with routers on the other side for nearly 12-15 years, I don't think they've changed all that much. They didn't allow per-port TX and RX QoS rules. You could use the port something arrives on to classify, but not a combination of port and protocol. Even then, switching from port 1 to port 2, it was port 2's queues things got dumped into, on a per-port basis you only really had 1 way QoS, and that was what you were about to receive.
There are so many things wrong with what you've described. First of all, let's start with the big one. I never said that you wait until the link is saturated before you implement a queuing policy. I said the FIFO buffers are empty except when the line is saturated. Now, sit down and think about what I said, and then what you said, and come back when you realize just how silly your whole message was.
Now think what happens when the tx queue & buffer are empty because you are transmitting slower than the line rate. At that point, *ALL* QoS queues are empty. Now, it makes absolutely no difference what QoS queue you toss the next packet into, it will immediately be removed and put into the tx queue, and then immediately pulled into the hardware tx buffers. High priority, low priority, low latency, high throughput, they all work exactly the same. The first one in gets transmitted immediately.
Now, on a microscale, you can consider the line saturated for until that packet has been completely transmitted, although I was referring to the tx queue being full, not necessarily a single packet.
As for your 50Mbps link over a 1gig line, the same thing applies, just that now you are working on a virtual 50Mbps link, and the same saturation issues apply, but now it's only a virtual issue rather than a hardware one. No difference at all, and the fact that you even tried to make the issue more complicated by tossing on yet another layer of uselessness typically means you don't know what you are talking about and just trying to complicate things so you don't have to explain you didn't understand the issue in the first place.
Unfortunately, there are alot of bad network engineers out there, and I just met another one.
Yes, but there isn't a router out there capable of allowing every port to have their own traffic prioritization rules and/or allow them to be configured by end users, so that is currently impossible to implement the way you want it to be. So long as the ISP prioritizes traffic in a fair and equal manner, it is a good thing. But you are talking about changing the priority based on the source rather than what type of content it is. That is completely different. What I've seen is that people using bit torrent get upset that their transfers may get slower during peak times clamoring about net neutrality, when it really has nothing to do about the content (or P2P vs non-P2P), but rather about time sensitive stuff (speed and latency) vs stuff that isn't. Of course everyone wants THEIR stuff to be first.
I'd much rather see bandwidth policies that are realistic, like having the maximum bandwidth on my line available during non-peak times, and the 5% of my bandwidth that is time sensitive to be undelayed, and the 95% of my bandwidth that isn't time sensitive to possibly be slower than its theoretical maximum during peak-times. I don't get frustrated when my steam downloads are a bit slow, but I do get frustrated when I can't understand my phone calls because the guy on the other side now sounds like he just turned into a mentally retarded robot from hell.
Well those FIFO buffers are empty except when the line is completely saturated. Granted, it may only be saturated for milliseconds, but what I said is accurate. You are just making the situation more complex and then describing what happens when the line is completely saturated, even if only for milliseconds. But yes, traffic prioritization should allow packets to jump to the front of the queue (or multiple queues, as many routes actually have multiple distinct buffers, one for each traffic priority).
Why feel bad for the programmers? I'd do if they paid me to do it. I've coded lots of stuff that were based on totally bad designs that I knew would flop almost instantly. I got paid the same, and they flop, and then they pay me to do something less retarded.
Only if they include content from the original webpage, which it most likely will not. It'll probably be implemented as a DNS redirect, but they might get fancy and just redirect based on URL, but the later requires significantly more hardware, so I'm guessing it's the former. They see you are trying to access www.moviepiratesgalore.com and redirect you to www.mpaa.com instead.
HTTP shouldn't (and wasn't) created with full end to end encryption or authentication because it's not supposed to be. That stuff is (and should be) all done at a lower level in the network stack.
I do not, nor did I ever have an issue with traffic prioritization. Mostly the people who don't know what that is, or how it works have issues with it, but that's just ignorance. Real traffic prioritization only kicks in when lines are completely full, and then it lets stuff through with higher priority (VOIP, gaming packets, web browsing, video on demand) first. Things that aren't time sensitive (FTP, HTTP, BitTorrent, NNTP, etc) are sent as soon as they can.
Of course, the alternative is that the internet connection becomes unusable during those times for anything time sensitive at all. I'd prefer the former. Of course, I'd prefer they upgrade the links, but we're talking reality not fantasy land like some people still believe in.
With the "hot" sections of reasonable size and modular, I imagine a dead one could be replaced, slab and all, in an overnight or over-weekend operation, scheduled for when the road is not too busy and lane closures or detours are available.
You obviously don't live in Illinois. Something like that would take at least a year and a half. Of course, you may only see people working on the road for a day or two during that time, but they'll close the road for a year+ easy.
The main areas of incompatibility between the NT5 (WinXP) and NT6 (Vista/7/8) kernels were that the sound and video drivers were moved from kernel space to user space to help improve stability. Most all other drivers were still compatible or only had minor changes required.
And I don't usually have to reboot to install new video drivers, and when the video driver does die out, I see the screen flicker, and a pop up notification tells me that the video driver crashed and was restarted, but everything else keeps on working.
MS propogated a culture of developers using Administrative Rights for nearly every application. It didn't help that many of their own APIs were broken so badly that you had to have those rights to do many things. However, they also warned developer for years that the change was coming, and developers had the opportunity to test on Vista before its release to make sure that wouldn't be an issue - yet most chose to ignore it. Thus the whole UAC debacle which is primarily a 3rd party issue.
I would say this is mostly right. I don't totally agree that MS propagated that culture, but it was prevalent, especially in bad programmers who really didn't know what they were doing. It forced some to actually realize they didn't need to do some of the crazy things they were doing, but they just didn't know better at the time.
In automotive terms, it's the equivalent of needing the key to start the car because you want to check the time on the radio. Of course, you could just hit the button on the radio which will turn the clock on temporarily, but some people will cry and moan that they need the key very loudly at the top of their lungs about it.
Slashdot Mirror
Because not all images are square. Objects that aren't square (like a star for example), if you wanted to place that star on a page, and didn't have alpha transparency, you could use GIF, but the edges would look weird because it's either all on or all off transparency which creates jaggy edges. With alpha transparency you can make the object's edges more transparent as it gets closer to the edge which will let it look right no matter what type of background you have. Of course there are many other uses, like semi-transparent effects, shadows, etc that aren't easily replicated another way.
These type things allow a webpage to reuse assets over and over across many pages, or in many different sections of the same page, which reduces the amount of image manipulation you have to do, and also reduces the size of the page which is more and more important as devices with low bandwidth become more prominent (like phones/tablets).
I know she does. Perhaps you should take away her video cam.
I did that. I only get messages that say "Drink more ovaltine".
The NSA would just hack his account, and keep posting ridiculous stupid stuff so we couldn't tell the difference.
it's only a matter of time before attackers start figuring out ways to use those statistical models
They already do. They use attacks that hit the largest number of targets. Using uncommon vulnerabilities would be wasteful when you could attack more common ones.
oops, I meant became Windows XP.
Only the Windows 3.1 line grew out of MS-DOS, and the last in that line was Windows ME. The NT line grew out of VMS, which became Windows 95.
OS/2 was a piece of garbage compared to Windows NT at the time.
I can get UnitedHealthOne Saver 80 here in Illinois. Early 40's, non-smoker... $84.52/month. Not the best coverage, but it handles the catastrophic stuff which is all I really care about anyhow. Better insurance with lower deductibles are also available for $240.
I guess that depends on your goal. For example, the average US citizen pays $7,980 per person for healthcare. That's the highest of any country. The who report you gave says the the US is .838 efficient, while the average UK citizen pays $3,438 per person, but is 0.925 efficient.
Now who receives better care? Note, I said better, not better per dollar.
I don't think there are any hospitals with more than 1 CAT scanner. But for comparison, there are 34 CAT scanners per million people in the US, while there is 7 per million in the UK. I think they have it worse.
Why are you talking about customer side routers? What the hell does that have to do with anything?
Granted, I haven't had to deal with routers on the other side for nearly 12-15 years, I don't think they've changed all that much. They didn't allow per-port TX and RX QoS rules. You could use the port something arrives on to classify, but not a combination of port and protocol. Even then, switching from port 1 to port 2, it was port 2's queues things got dumped into, on a per-port basis you only really had 1 way QoS, and that was what you were about to receive.
There are so many things wrong with what you've described. First of all, let's start with the big one. I never said that you wait until the link is saturated before you implement a queuing policy. I said the FIFO buffers are empty except when the line is saturated. Now, sit down and think about what I said, and then what you said, and come back when you realize just how silly your whole message was.
In the meantime, here is a nice image for you if you need help visualizing it: http://www.cisco.com/en/US/i/000001-100000/15001-20000/16501-17000/16756.jpg
Now think what happens when the tx queue & buffer are empty because you are transmitting slower than the line rate. At that point, *ALL* QoS queues are empty. Now, it makes absolutely no difference what QoS queue you toss the next packet into, it will immediately be removed and put into the tx queue, and then immediately pulled into the hardware tx buffers. High priority, low priority, low latency, high throughput, they all work exactly the same. The first one in gets transmitted immediately.
Now, on a microscale, you can consider the line saturated for until that packet has been completely transmitted, although I was referring to the tx queue being full, not necessarily a single packet.
As for your 50Mbps link over a 1gig line, the same thing applies, just that now you are working on a virtual 50Mbps link, and the same saturation issues apply, but now it's only a virtual issue rather than a hardware one. No difference at all, and the fact that you even tried to make the issue more complicated by tossing on yet another layer of uselessness typically means you don't know what you are talking about and just trying to complicate things so you don't have to explain you didn't understand the issue in the first place.
Unfortunately, there are alot of bad network engineers out there, and I just met another one.
Yes, but there isn't a router out there capable of allowing every port to have their own traffic prioritization rules and/or allow them to be configured by end users, so that is currently impossible to implement the way you want it to be. So long as the ISP prioritizes traffic in a fair and equal manner, it is a good thing. But you are talking about changing the priority based on the source rather than what type of content it is. That is completely different. What I've seen is that people using bit torrent get upset that their transfers may get slower during peak times clamoring about net neutrality, when it really has nothing to do about the content (or P2P vs non-P2P), but rather about time sensitive stuff (speed and latency) vs stuff that isn't. Of course everyone wants THEIR stuff to be first.
I'd much rather see bandwidth policies that are realistic, like having the maximum bandwidth on my line available during non-peak times, and the 5% of my bandwidth that is time sensitive to be undelayed, and the 95% of my bandwidth that isn't time sensitive to possibly be slower than its theoretical maximum during peak-times. I don't get frustrated when my steam downloads are a bit slow, but I do get frustrated when I can't understand my phone calls because the guy on the other side now sounds like he just turned into a mentally retarded robot from hell.
Well those FIFO buffers are empty except when the line is completely saturated. Granted, it may only be saturated for milliseconds, but what I said is accurate. You are just making the situation more complex and then describing what happens when the line is completely saturated, even if only for milliseconds. But yes, traffic prioritization should allow packets to jump to the front of the queue (or multiple queues, as many routes actually have multiple distinct buffers, one for each traffic priority).
Probably not. But if they are just doing redirection via DNS, then it depends on what you have your nameserver set as its upstream nameserver.
Just try and get an original cut of Disney's live action/animated hybrid "Song of the South"
Ok, here: http://classicmoviereel.com/SOTS.html
Why feel bad for the programmers? I'd do if they paid me to do it. I've coded lots of stuff that were based on totally bad designs that I knew would flop almost instantly. I got paid the same, and they flop, and then they pay me to do something less retarded.
Only if they include content from the original webpage, which it most likely will not. It'll probably be implemented as a DNS redirect, but they might get fancy and just redirect based on URL, but the later requires significantly more hardware, so I'm guessing it's the former. They see you are trying to access www.moviepiratesgalore.com and redirect you to www.mpaa.com instead.
HTTP shouldn't (and wasn't) created with full end to end encryption or authentication because it's not supposed to be. That stuff is (and should be) all done at a lower level in the network stack.
I do not, nor did I ever have an issue with traffic prioritization. Mostly the people who don't know what that is, or how it works have issues with it, but that's just ignorance. Real traffic prioritization only kicks in when lines are completely full, and then it lets stuff through with higher priority (VOIP, gaming packets, web browsing, video on demand) first. Things that aren't time sensitive (FTP, HTTP, BitTorrent, NNTP, etc) are sent as soon as they can.
Of course, the alternative is that the internet connection becomes unusable during those times for anything time sensitive at all. I'd prefer the former. Of course, I'd prefer they upgrade the links, but we're talking reality not fantasy land like some people still believe in.
With the "hot" sections of reasonable size and modular, I imagine a dead one could be replaced, slab and all, in an overnight or over-weekend operation, scheduled for when the road is not too busy and lane closures or detours are available.
You obviously don't live in Illinois. Something like that would take at least a year and a half. Of course, you may only see people working on the road for a day or two during that time, but they'll close the road for a year+ easy.
The main areas of incompatibility between the NT5 (WinXP) and NT6 (Vista/7/8) kernels were that the sound and video drivers were moved from kernel space to user space to help improve stability. Most all other drivers were still compatible or only had minor changes required.
And I don't usually have to reboot to install new video drivers, and when the video driver does die out, I see the screen flicker, and a pop up notification tells me that the video driver crashed and was restarted, but everything else keeps on working.
MS propogated a culture of developers using Administrative Rights for nearly every application. It didn't help that many of their own APIs were broken so badly that you had to have those rights to do many things. However, they also warned developer for years that the change was coming, and developers had the opportunity to test on Vista before its release to make sure that wouldn't be an issue - yet most chose to ignore it. Thus the whole UAC debacle which is primarily a 3rd party issue.
I would say this is mostly right. I don't totally agree that MS propagated that culture, but it was prevalent, especially in bad programmers who really didn't know what they were doing. It forced some to actually realize they didn't need to do some of the crazy things they were doing, but they just didn't know better at the time.
In automotive terms, it's the equivalent of needing the key to start the car because you want to check the time on the radio. Of course, you could just hit the button on the radio which will turn the clock on temporarily, but some people will cry and moan that they need the key very loudly at the top of their lungs about it.
Do they have streaming?
Wait until you start using brain scanners and continue to "type" after you've fallen asleep. I cringe to think of how the conversation changes.