Look, I get what you are saying. But freedom is useless if crime and terror hit a certain level. Nobody is going to feel free to engage in business and leisure if they fear that guys will be blowing shit up on a regular basis. There are all kinds of tyranny.
Seems to make sense. I haven't known too many old people, but the ones who got dementia as they aged were also the ones with diabetes and kidney problems. There could also be a direct blood sugar to brain cell connection. Perhaps higher blood sugar makes the brain form connections differently, or somehow contributes to the plaque formations that sometimes come with dementia.
There are two kinds of addictions: physical and psychological. Psychological addiction actually does remove the choice. If you don't get your next hit, you go crazy.
The compression algorithm slices the document into blocks, and stores each block once. If another part of the document uses a block of data that is identical to another one already in memory, it just has to store a pointer to the first one, not the whole block of data. If you set the function that compares blocks from "identical" to "good enough", that's how you get these errors.
Simple example: fonts. You have a 8.5 x 11 sheet of paper @ 300 dpi. 90,000 dots per square inch, and 8,415,000 dots per page. That's about one megabyte of data. However, that sheet of paper isn't random data. It's just a set of blocks, each containing one of 52 letters and 10 numbers. 10 characters per inch wide, 6 characters per inch long. Or, 5610 blocks that can contain one of 62 different characters, or 6 bits of information per block. 5610 blocks of 6 bits per block is only 43 kB.
Now, that assumes the best case: both sides of the communication know what the font looks like and we don't have to transmit pictures of each font. Even if we do, we only have to transmit each picture of each block once, and then just refer back to the pictures. So we have 43kB of data and 11 kB of font information. Completely lossless, if done right, and saves a ton of data.
This algorithm does the same thing, but on a completely ad-hoc basis. Scan the document and figure out what block size has the most repetition, and then carve it into that block size. Transmit the "font" or "palette" of blocks that are contained in the document, and then transmit a map of which type of block goes where. If the machine doesn't have to perfectly match the blocks, it can reduce the size of the "font" its using.
It is done to make the engineering easier. The machine has three uses: printing documents, copying documents and scanning to a file. (Basically). So instead of building those three separate functional modes, you just build two and simulate the third. The scanner half scans to a file, and the printer half prints a document. The "copier" function is just those two functions piped into one another. Adding in compression makes the two halves of the machine perform faster with less memory, and *shouldn't* affect the copier function, except to make it easier to run at full speed.
Also, don't underestimate the size of raw image data. A black and white, 300 dpi letter sized page needs something close to 1mb of ram to store. No problem. Double that to 600 dpi and the amount of ram quadruples. Start adding in multi bitdepth pixels, or color, across multiple pages, and the amount of data starts getting silly. Not even just the RAM needed to store it, but the speed of the RAM and the data paths needed to push the data around and still print at the rated X pages per minute of the printer. Not the world's most difficult problems to solve, but they are made a lot easier if you throw in some nice compression. Even easier if the compression can be lossy.
The end result is that you can build a machine that is cheaper and performs up to the expectations of the user. The user really doesn't want to hear that their 1200 x 1200 dpi printer can only do that resolution at 4 pages per minute when the thing is advertised as being 47 pages per minute.
IE 10 really is pretty darn good. I especially like the F12 window that opens, where you can tell it to emulate any version of IE back to 7. The only thing that keeps me from using it full time instead of Firefox is the hassle of moving my bookmarks over. I'm just not in the mood for it.
If people want it, it should be made into a plugin. That's the firefox motto, isn't it? All they need to do is make the fucking thing work properly. Silly things like non-shit font rendering, rendering speed and maybe something truly groundbreaking, like an engine that doesn't lock up everything when one window/tab has some shit content on it. Or an engine that uses more than one core.
The Surface RT would be fine for just tablet use, but the problem is that there aren't enough Metro apps that work well. If there was a better ecosystem of apps, I'm sure it would be doing better.
Agree. Law enforcement has a lot of latitude in their official activities. It is often written into the laws. "X is illegal unless being done as part of a bona fide official investigation." So computer hacking specifically might be untested, the broader concept is on solid ground.
I agree that it functionally doesn't end that way, but on the other hand, a libertarian would say that the big guys would never have gotten so big if the state hadn't manipulated the market to somehow favor them.
Look at medical lawsuits- those have fundamentally changed how a lot of medicine is practiced. Same with consumer protection lawsuits. One could argue that no amount of regulation would have had the same affect on safety for consumers of either industry.
You'd think the Android people would have considered this when developing their framework. Or maybe they did and the PBS developers just don't know how to do it.
Adding in a time component to the calculation doesn't make it any less true. If you aren't paying for the service, you aren't the customer. It really is that simple. Almost all business have to buy inventory before they can sell their product. It's really the same business model as newspapers. The price you pay really just covers the cost of printing and distribution. Their customers are the advertisers, and the content and consumers are just grist for the advertising mill.
The idea is that the only regulation a market needs is a functional system to arbitrate disagreements. The courts, for example. If someone is being greedy and rips another market participant off, they should sue them and recover their losses. The threat of a lawsuit should be all the encouragement a bad guy needs to straighten up or exit the business. The big conglomerations of business (and labor unions) exist and existed almost completely because of the big boys' ability to buy regulatory capture. Business wouldn't have gotten nearly as powerful if they didn't press the government to give them all kinds of freebies, and neither would unions be as powerful if the government didn't give them special treatment.
That's the problem. People think the free market is based on greed, or works via greed. Look up greed sometime- it's not a positive trait. Especially not in a marketplace. One might even say a really well done free market works against greed. Free markets work on the principle of mutual benefit. All participants to a deal want something that will benefit them somehow, and trade something that they don't need as much. [Libertarians will tell you that] markets only favor the greedy when the state puts up (or allows to be put up) barriers and restraints to participation in the marketplace.
You ain't just whistling Dixie. I just got a tablet with a dual core atom, and I was expecting it to be capable at best. But upon using it, it has shown to be pretty goddamned fast for 80% of what anyone does. When I looked at the block diagram of the new Atom chips, I can see why. Dedicated hardware blobs for everything.
Reminds me of my first computer DVD player. It was some kind of Creative thing, where there was an adapter card with a hardware decoder on it. It played DVDs perfectly and beautifully on a Pentium 75. Only until a few years ago have I seen similar quality coming from general purpose hardware and software decoding.
When the software can take advantage of multiple cores, then fine. But not very much does. You just get multiple programs running at one time, and in my personal use case, very rarely are more than 2-4 executables trying to run at once. When Qualcomm sees phone software that needs more cores, they will put them in.
Slashdot Mirror
Who says who won? Maybe the neandertal blood in us is what kicked-started humanity's ascent.
It didn't look too free in Northern Ireland.
Look, I get what you are saying. But freedom is useless if crime and terror hit a certain level. Nobody is going to feel free to engage in business and leisure if they fear that guys will be blowing shit up on a regular basis. There are all kinds of tyranny.
Which doesn't say much.
Seems to make sense. I haven't known too many old people, but the ones who got dementia as they aged were also the ones with diabetes and kidney problems. There could also be a direct blood sugar to brain cell connection. Perhaps higher blood sugar makes the brain form connections differently, or somehow contributes to the plaque formations that sometimes come with dementia.
Except by the FDA and nutritionists. Their advice is still low fat, lots of grains.
The addiction is what makes you want to smoke. You know something is bad and you do it anyway = addiction at some level.
There are two kinds of addictions: physical and psychological. Psychological addiction actually does remove the choice. If you don't get your next hit, you go crazy.
The compression algorithm slices the document into blocks, and stores each block once. If another part of the document uses a block of data that is identical to another one already in memory, it just has to store a pointer to the first one, not the whole block of data. If you set the function that compares blocks from "identical" to "good enough", that's how you get these errors.
Simple example: fonts. You have a 8.5 x 11 sheet of paper @ 300 dpi. 90,000 dots per square inch, and 8,415,000 dots per page. That's about one megabyte of data. However, that sheet of paper isn't random data. It's just a set of blocks, each containing one of 52 letters and 10 numbers. 10 characters per inch wide, 6 characters per inch long. Or, 5610 blocks that can contain one of 62 different characters, or 6 bits of information per block. 5610 blocks of 6 bits per block is only 43 kB.
Now, that assumes the best case: both sides of the communication know what the font looks like and we don't have to transmit pictures of each font. Even if we do, we only have to transmit each picture of each block once, and then just refer back to the pictures. So we have 43kB of data and 11 kB of font information. Completely lossless, if done right, and saves a ton of data.
This algorithm does the same thing, but on a completely ad-hoc basis. Scan the document and figure out what block size has the most repetition, and then carve it into that block size. Transmit the "font" or "palette" of blocks that are contained in the document, and then transmit a map of which type of block goes where. If the machine doesn't have to perfectly match the blocks, it can reduce the size of the "font" its using.
It is done to make the engineering easier. The machine has three uses: printing documents, copying documents and scanning to a file. (Basically). So instead of building those three separate functional modes, you just build two and simulate the third. The scanner half scans to a file, and the printer half prints a document. The "copier" function is just those two functions piped into one another. Adding in compression makes the two halves of the machine perform faster with less memory, and *shouldn't* affect the copier function, except to make it easier to run at full speed.
Also, don't underestimate the size of raw image data. A black and white, 300 dpi letter sized page needs something close to 1mb of ram to store. No problem. Double that to 600 dpi and the amount of ram quadruples. Start adding in multi bitdepth pixels, or color, across multiple pages, and the amount of data starts getting silly. Not even just the RAM needed to store it, but the speed of the RAM and the data paths needed to push the data around and still print at the rated X pages per minute of the printer. Not the world's most difficult problems to solve, but they are made a lot easier if you throw in some nice compression. Even easier if the compression can be lossy.
The end result is that you can build a machine that is cheaper and performs up to the expectations of the user. The user really doesn't want to hear that their 1200 x 1200 dpi printer can only do that resolution at 4 pages per minute when the thing is advertised as being 47 pages per minute.
Because programming is an art, maaaaaan!
One would wonder why camels are ok to eat, but not pigs? They are closely related.
IE 10 really is pretty darn good. I especially like the F12 window that opens, where you can tell it to emulate any version of IE back to 7. The only thing that keeps me from using it full time instead of Firefox is the hassle of moving my bookmarks over. I'm just not in the mood for it.
"The Facebook button you installed on my computer last night doesn't work!"
If people want it, it should be made into a plugin. That's the firefox motto, isn't it? All they need to do is make the fucking thing work properly. Silly things like non-shit font rendering, rendering speed and maybe something truly groundbreaking, like an engine that doesn't lock up everything when one window/tab has some shit content on it. Or an engine that uses more than one core.
Yeah, this was a pretty shit article. Congratulations to the author, he just invented qos the hard way.
The Surface RT would be fine for just tablet use, but the problem is that there aren't enough Metro apps that work well. If there was a better ecosystem of apps, I'm sure it would be doing better.
Agree. Law enforcement has a lot of latitude in their official activities. It is often written into the laws. "X is illegal unless being done as part of a bona fide official investigation." So computer hacking specifically might be untested, the broader concept is on solid ground.
I agree that it functionally doesn't end that way, but on the other hand, a libertarian would say that the big guys would never have gotten so big if the state hadn't manipulated the market to somehow favor them.
Look at medical lawsuits- those have fundamentally changed how a lot of medicine is practiced. Same with consumer protection lawsuits. One could argue that no amount of regulation would have had the same affect on safety for consumers of either industry.
You'd think the Android people would have considered this when developing their framework. Or maybe they did and the PBS developers just don't know how to do it.
Adding in a time component to the calculation doesn't make it any less true. If you aren't paying for the service, you aren't the customer. It really is that simple. Almost all business have to buy inventory before they can sell their product. It's really the same business model as newspapers. The price you pay really just covers the cost of printing and distribution. Their customers are the advertisers, and the content and consumers are just grist for the advertising mill.
The idea is that the only regulation a market needs is a functional system to arbitrate disagreements. The courts, for example. If someone is being greedy and rips another market participant off, they should sue them and recover their losses. The threat of a lawsuit should be all the encouragement a bad guy needs to straighten up or exit the business. The big conglomerations of business (and labor unions) exist and existed almost completely because of the big boys' ability to buy regulatory capture. Business wouldn't have gotten nearly as powerful if they didn't press the government to give them all kinds of freebies, and neither would unions be as powerful if the government didn't give them special treatment.
That's the problem. People think the free market is based on greed, or works via greed. Look up greed sometime- it's not a positive trait. Especially not in a marketplace. One might even say a really well done free market works against greed. Free markets work on the principle of mutual benefit. All participants to a deal want something that will benefit them somehow, and trade something that they don't need as much. [Libertarians will tell you that] markets only favor the greedy when the state puts up (or allows to be put up) barriers and restraints to participation in the marketplace.
You ain't just whistling Dixie. I just got a tablet with a dual core atom, and I was expecting it to be capable at best. But upon using it, it has shown to be pretty goddamned fast for 80% of what anyone does. When I looked at the block diagram of the new Atom chips, I can see why. Dedicated hardware blobs for everything.
Reminds me of my first computer DVD player. It was some kind of Creative thing, where there was an adapter card with a hardware decoder on it. It played DVDs perfectly and beautifully on a Pentium 75. Only until a few years ago have I seen similar quality coming from general purpose hardware and software decoding.
When the software can take advantage of multiple cores, then fine. But not very much does. You just get multiple programs running at one time, and in my personal use case, very rarely are more than 2-4 executables trying to run at once. When Qualcomm sees phone software that needs more cores, they will put them in.