Well, a Computer Science degree would cover this solidly in a course with a title like "Analysis of Algorithms". I'll summerize though:
Both zip files and XML files are data streams that must be read sequentially, like a typical tape. You can't skip around. There is no significant indexing ability. You can't skip to page 123 of a presentation, cell $12345$67890 of a spreadsheet, or page 1234 of a book. You have to read the whole damn thing into memory, doing a mildly expensive decompression operation as you go.
With something like *.doc, skipping ahead is done via a tree structure or hash. This is fast.
This isn't just a minor little speed difference. This is a difference that grows with the file size. Access time in zipped XML grows linearly with the file size. Access time in a decent file format grows with the log of the file size at most.
Note that the quality of the app can't improve this, though a bad app can make things worse. These performance issues are fundamental to the data layout.
You can do that with Microsoft formats too. For ODF you used unzip and mindom. For the Microsoft stuff, you just use Microsoft's interfaces. You could write an ActiveX object, then have Office Basic load it up and call it. The other way, if you'd really like this to be an external tool, is to invoke the various COM/D-COM/etc. interfaces -- effectively using Microsoft Office as a library to read the files.
No, you can't reliably write your own parser, but you didn't do that for ODF either.
As for ODF being a turd... yes, but not as smelly as the OO-XML turd. All this zipped XML stuff is very low performance.
With two genes, you get sickle cell anemia and are quite likely to die young. With just one, you're much more likely to survive malaria.
For homosexuality:
With two genes, you like dudes and are quite unlikely to get one pregnant. With just one... maybe you can better resist the urge to screw women with obvious signs of an STD. Maybe you don't get yourself killed in a fight over a woman.
Maybe you exist to be an unthreatening way to get women to spend time with your family... so that your brothers (sharing much DNA with you) can grab them.
First of all, "like the original" is subjective too, for all the reasons you gave. So that's no good.
Second of all, if we test with enough people, "best" is what typical people think is best. That's what matters when reporting results to the general population.
Of course, nothing beats deciding "best" for your own personal use, but that isn't worth reporting to the world.
Take 50 random people of each interesting race and skin color. (Chinese, Ashkenazi Jew, Irish, central African) Breed them such that each grandchild is fully mixed.
Chances are, skin color won't match up with intelligence. It could happen if a skin color gene actually had a second purpose, or if skin color was on the same chromosome as something brain-related, but it's rather unlikely.
We aren't all mixed-race though. Most of us get our skin color from the same part of the world as everything else.
You need an unbiased observer, essentially a robot. This could be a human following strict instructions for doing measurements, or it could be a computer program that performs a radar-like scan. Of course, the inclusion of internal structure (via CAT scan, MRI scan, or PET scan) would be better.
You need an unbiased diagnosis too. This could be done via rather normal statistics. Better would be a score coming out of a Bayesian or neural network. One could even modify those via genetic algorithm, etc.
In the government, the good person can't be paid well. That would violate all sorts of stupid government rules. The only way to pay the person well is if they are not considered a person, but a business.
In the government, nobody can fire the differently abled buddhist afro-american transsexual. It can slack off all it wants, generally fuck things up, etc. Competent people aren't all willing to stomach having to deal with that in a coworker, boss, or underling.
Lots of competent people don't wish to pee in a cup. A few may be light drug users that would still manage to do good work. The rest are just insulted that anybody would suggest that drugs might be in use.
The government isn't providing stock options, fancy food, etc. It's not like working at Google.
So contracters, big and small, are how shit gets done. It's economics finding a way past some of the stupidity. The cost is of course some other types of stupidity.
We want an experiment designed to identify the codec that will make our music sound the best.
That's it. "best" Not "like the original", which is a poor substitute for "best".
If the original was tainted by phono record pops and clicks, and the codec made that go away, great! We really don't care about "like original". We care about the best.
Your experiment measures something that, while mildly interesting, isn't what we need to determine.
SCO shipped Xenix compatibility crap with OpenServer, as required by the contract that transferred Xenix to SCO. SCO didn't actually want to ship this crap, as least for the price being paid. BTW, Microsoft also had shares of SCO from that deal.
SCO got their freedom in a lawsuit a few years back. As I recall, there was a settlement. Microsoft sold/lost the shares and SCO stopped shipping the Xenix crap.
Let's send him a hot tip about stock in a make-penis-fast venture. We can FedEx it on carved stone tablets. That ought to get his attention.
Then... let's offer him a great deal on loans for 6ener1c v1agra. We can have UPS deliver it on sheepskin scrolls.
Finally, we can offer him huge savings on gambling in Nigeria. We can have DHL bring it, tatooed on the side of a water buffalo. (DHL delivers anything!)
People vary in size, both by individual and by racial group. Were the parents smallish by your standards? You're not talking about a European family you know.
It is comparable to removing labia. (which would not leave her with NO feeling down there; the clitoris and vagina would remain)
Physilogically, it's like removing the clittoral hood.
Sex may be "great", but you have no knowledge of how much better it might be with a foreskin. The foreskin is the only part of the human body with estrogen sensors. ("sensor" meaning that you can feel it)
It takes a brave man to admit that his privates have been damaged.
We think we're civilized. We think Europe is civilized. When a bunch of Europeans go slaughtering a group of people as if killing cockroaches, it freaks us out. It makes us worry that we ourselves could be that evil.
Africa is where we expect people in loincloths to throw spears at each other. It's one small step above cavemen. If they kill each other, we don't blink an eye. That always happens in uncivilized places. It's not as if we could tell a Hutu from a Tootsie anyway; they all look the same.
Known: 8-bit channels. Near-total ignorance of gamma. Pretty much needs a virtual desktop to itself. Must look up Pantone crap by hand (try a web page) because of intellectual property restrictions.
Bullshit: Non-RGB working space (CMYK is device-specific). Menus not word-for-word identical to Photoshop, and window titlebar doesn't say "Adobe Photoshop".
(actually not a bad idea to be first in the "market" for this, if you can pay the bills long enough for desktop Linux to really take off -- or just do a whole negative-cost distribution for OEMs)
Run-away growth can lead to starvation, so the fungus might normally hold back a bit. If weird damage is occuring though, the situation changes. Doing nothing leads to quick death; reproducing fills the voids left by those who've died.
You need a well-defined color space. You need more than 8 bits per channel. For sanity, you need 3 additive channels.
CMYK is not in general well-defined. It's device-specific. CMYK on my printer will not match CMYK on your printer. That's crap for reliable editing.
CMYK belongs in your printer driver or in the device itself. There you convert from something well-defined and standard (sRGB, AdobeRGB, CIE-XYZ, etc.) to whatever your printer wants today. To get picky, CYMK even varies with the brand of paper you bought.
Slashdot Mirror
Well, a Computer Science degree would cover this solidly in a course with a title like "Analysis of Algorithms". I'll summerize though:
Both zip files and XML files are data streams that must be read sequentially, like a typical tape. You can't skip around. There is no significant indexing ability. You can't skip to page 123 of a presentation, cell $12345$67890 of a spreadsheet, or page 1234 of a book. You have to read the whole damn thing into memory, doing a mildly expensive decompression operation as you go.
With something like *.doc, skipping ahead is done via a tree structure or hash. This is fast.
This isn't just a minor little speed difference. This is a difference that grows with the file size. Access time in zipped XML grows linearly with the file size. Access time in a decent file format grows with the log of the file size at most.
Note that the quality of the app can't improve this, though a bad app can make things worse. These performance issues are fundamental to the data layout.
You can do that with Microsoft formats too. For ODF you used unzip and mindom. For the Microsoft stuff, you just use Microsoft's interfaces. You could write an ActiveX object, then have Office Basic load it up and call it. The other way, if you'd really like this to be an external tool, is to invoke the various COM/D-COM/etc. interfaces -- effectively using Microsoft Office as a library to read the files.
No, you can't reliably write your own parser, but you didn't do that for ODF either.
As for ODF being a turd... yes, but not as smelly as the OO-XML turd. All this zipped XML stuff is very low performance.
Pretty much anything written with LaTeX has that ugly LaTeX look.
I'm sure you could write lots of TeX code to avoid this, but then you've given the "fairly quickly" advantage to Word.
It's a sandbox.
They can set a cookie all they like. It doesn't stay set for long.
With two genes, you get sickle cell anemia and are quite likely to die young. With just one, you're much more likely to survive malaria.
For homosexuality:
With two genes, you like dudes and are quite unlikely to get one pregnant. With just one... maybe you can better resist the urge to screw women with obvious signs of an STD. Maybe you don't get yourself killed in a fight over a woman.
Maybe you exist to be an unthreatening way to get women to spend time with your family... so that your brothers (sharing much DNA with you) can grab them.
Oh no.
First of all, "like the original" is subjective too, for all the reasons you gave. So that's no good.
Second of all, if we test with enough people, "best" is what typical people think is best. That's what matters when reporting results to the general population.
Of course, nothing beats deciding "best" for your own personal use, but that isn't worth reporting to the world.
Take 50 random people of each interesting race and skin color. (Chinese, Ashkenazi Jew, Irish, central African) Breed them such that each grandchild is fully mixed.
Chances are, skin color won't match up with intelligence. It could happen if a skin color gene actually had a second purpose, or if skin color was on the same chromosome as something brain-related, but it's rather unlikely.
We aren't all mixed-race though. Most of us get our skin color from the same part of the world as everything else.
You need an unbiased observer, essentially a robot. This could be a human following strict instructions for doing measurements, or it could be a computer program that performs a radar-like scan. Of course, the inclusion of internal structure (via CAT scan, MRI scan, or PET scan) would be better.
You need an unbiased diagnosis too. This could be done via rather normal statistics. Better would be a score coming out of a Bayesian or neural network. One could even modify those via genetic algorithm, etc.
There you go, legitimate phrenology.
I don't think Franz Joseph Gall or the Nazis bothered with statistics, but supposing one did that... problem?
The numbers are as the numbers are, even if you don't like them or can't properly explain them.
We could probably train a neural network to recognize intelligent people via exterior features.
In the government, the good person can't be paid well. That would violate all sorts of stupid government rules. The only way to pay the person well is if they are not considered a person, but a business.
In the government, nobody can fire the differently abled buddhist afro-american transsexual. It can slack off all it wants, generally fuck things up, etc. Competent people aren't all willing to stomach having to deal with that in a coworker, boss, or underling.
Lots of competent people don't wish to pee in a cup. A few may be light drug users that would still manage to do good work. The rest are just insulted that anybody would suggest that drugs might be in use.
The government isn't providing stock options, fancy food, etc. It's not like working at Google.
So contracters, big and small, are how shit gets done. It's economics finding a way past some of the stupidity. The cost is of course some other types of stupidity.
Well, how long would have them save your state? A minute? An hour? While slashdotted?
Doesn't your browser support throwing away all cookies when you close it? If so, why isn't that good enough?
I could ask for throwing them away when the site passes from your history in all tabs and windows, but geez... browser shutdown is pretty good.
We want an experiment designed to identify the codec that will make our music sound the best.
That's it. "best" Not "like the original", which is a poor substitute for "best".
If the original was tainted by phono record pops and clicks, and the codec made that go away, great! We really don't care about "like original". We care about the best.
Your experiment measures something that, while mildly interesting, isn't what we need to determine.
No sir, I didn't chop down THAT cherry tree.
SCO shipped Xenix compatibility crap with OpenServer, as required by the contract that transferred Xenix to SCO. SCO didn't actually want to ship this crap, as least for the price being paid. BTW, Microsoft also had shares of SCO from that deal.
SCO got their freedom in a lawsuit a few years back. As I recall, there was a settlement. Microsoft sold/lost the shares and SCO stopped shipping the Xenix crap.
Really, he's not going to get upset about the video.
Anyway it's public record and even a news story.
Let's send him a hot tip about stock in a make-penis-fast venture. We can FedEx it on carved stone tablets. That ought to get his attention.
Then... let's offer him a great deal on loans for 6ener1c v1agra. We can have UPS deliver it on sheepskin scrolls.
Finally, we can offer him huge savings on gambling in Nigeria. We can have DHL bring it, tatooed on the side of a water buffalo. (DHL delivers anything!)
I bet everybody there, at least in the engineering department, has a DRM-free Zune.
Not to say the kid isn't malnourished, but...
People vary in size, both by individual and by racial group. Were the parents smallish by your standards? You're not talking about a European family you know.
It is comparable to removing labia. (which would not leave her with NO feeling down there; the clitoris and vagina would remain)
Physilogically, it's like removing the clittoral hood.
Sex may be "great", but you have no knowledge of how much better it might be with a foreskin. The foreskin is the only part of the human body with estrogen sensors. ("sensor" meaning that you can feel it)
It takes a brave man to admit that his privates have been damaged.
We think we're civilized. We think Europe is civilized. When a bunch of Europeans go slaughtering a group of people as if killing cockroaches, it freaks us out. It makes us worry that we ourselves could be that evil.
Africa is where we expect people in loincloths to throw spears at each other. It's one small step above cavemen. If they kill each other, we don't blink an eye. That always happens in uncivilized places. It's not as if we could tell a Hutu from a Tootsie anyway; they all look the same.
Known: 8-bit channels. Near-total ignorance of gamma. Pretty much needs a virtual desktop to itself. Must look up Pantone crap by hand (try a web page) because of intellectual property restrictions.
Bullshit: Non-RGB working space (CMYK is device-specific). Menus not word-for-word identical to Photoshop, and window titlebar doesn't say "Adobe Photoshop".
Anything else?
Write crapware for Linux. Pay Dell to install it.
Quick, let's all start Linux crapware companies!
(actually not a bad idea to be first in the "market" for this, if you can pay the bills long enough for desktop Linux to really take off -- or just do a whole negative-cost distribution for OEMs)
Run-away growth can lead to starvation, so the fungus might normally hold back a bit. If weird damage is occuring though, the situation changes. Doing nothing leads to quick death; reproducing fills the voids left by those who've died.
Use 1 bit per subpixel, for a total of 8 colors. Crank up the refresh rate. Do temporal (time based) dithering.
1-bit is going to be more reliable and accurate than 6-bit.
You need a well-defined color space. You need more than 8 bits per channel. For sanity, you need 3 additive channels.
CMYK is not in general well-defined. It's device-specific. CMYK on my printer will not match CMYK on your printer. That's crap for reliable editing.
CMYK belongs in your printer driver or in the device itself. There you convert from something well-defined and standard (sRGB, AdobeRGB, CIE-XYZ, etc.) to whatever your printer wants today. To get picky, CYMK even varies with the brand of paper you bought.