Well, I remember, back in the 80s, fans of FORTH who claimed it was faster than compiled C.
Some of those who made those claims understood that the FORTH interpreter was a better run-time execution model than the model that many C compilers for microcomputers back then implemented. So, FORTH interpreted code could be faster than a bad compiler's compiled code.
But some didn't seem to understand the fundamental behavior of an interpreter. I heard one guy talk about, if your program was too slow, just make parts of it words of their own, and then those words would be fast. (Didn't seem to understand that a call to a routine still had to execute the body of the routine. Or maybe he was confusing size optimization with speed optimization. I've done similar things, said "It does cool thing A." when I meant "It does cool thing B".)
There's no way an interpreter can run without putting at least a branch/jump between every primitive executed.
Well, I know the JIT is supposed to solve that, but why not just unroll the invariant primitives? If the code itself is final, it ought to be possible to just pick up everything to the last branch and insert it in the stored instruction stream without the branch. Large primitives aren't going to be helped much, but small primitives would. And, at compile time, you could dredge the unrolled code for things like a move to the parameter stack followed directly by a move from the stack to an accumulator.
Anecdotes: My Mac Mini (1.2G PPC, 512M RAM, slow frontside cache) runs Java about half the speed of my Linux box (1.7G Sempron 2700+, 768M RAM, frontside about half CPU speed). The pause between double-click and application coming up on screen is a bit longer than with native code on both machines, which is not surprising considering the pre-flighting Java does for security and other purposes. You see the effects of the branch being in there, speed-wise, but, as you say, it's fast enough.
On my clamshell iBook, however, rendering of html is seriously slow. Makes the help screen I'm rendering that way more or less useless.
When apple owns the portable phone market, then, yes, this would become subject to anti-trust measures.
Why can't people tell the difference between a monopoly in the market for brand X of product B of product category Q and a monopoly in the market for the entire product category.
iPods are kind of close to the borderline, but not really approaching the degree to which MSWindows/MSOffice has become the market for general purpose computers except for mainframes. And, yes, Apple has used various things to keep control of their own products, but have done nothing approaching what Microsoft has done to establish and maintain their monopoly. Give it a break.
Big money at the end of the long ride for the lucky few is a good way to screw the whole race up. We should not all be running the same race.
Small prizes for small steps.
Better yet, just get rid of the stupid overtime psychology and let us have enough time after hours to pursue the hard problems for the reasons that are most likely to bring real success:
Solving a problem is a reward in and of itself.
And a solution that helps others gives something even better than warm fuzzies. (Linux, anyone? BSD?)
As I read the reports I've seen, your local government offered you the choice, but the supreme court just said that they didn't have the authority to do so.
A little entry point on the subject and problems of caste in Japan:
He just presented you with a very good representative version of the argument against this kind of ID.
How do you ID your partner?
To make it more obvious, how do you ID [insert role or person here]?
Frankly, assuming the existence of such ID cards, I'm not going to laugh at all when girls want to see a guy's ID before going out with him the first time. It's a logical extension of the concept. And girls should be less trusting of the guys they go out with up front, really.
But it makes much more sense for the girl to talk with other girls who know the guy or whatever.
Actual identity is out-of-band data in any system we can build, computerized or not.
Building systems to manage identity on a limited basis in a localized scope may make some sense.
Unifying the ID system increases the value of the ID card well beyond the cost to obtain one illegitimately.
And the people to whom it becomes most immediately valuable are the people who have some vested interested in keeping their party in office, which is the very people who are issuing the cards.
I am not familiar enough with the Japanese constitution to know what kind of legal issues Juki.net raises, but I have to be wondering why they were keeping it quiet so long.
And I realize now that I should have been making a fuss about a job I had four years ago.
Anyway, a bit off-topic, but that the US national government is using various loopholes (like federal, vs. national) to sidestep the Constitution on these sorts of things is evidence that the US government is out of control, and therefore that the paranoia is based in reality.
The problems occur when they decide to change their core business.
No, there is not much keeping the _people_ implementing the law from changing it, especially when you make it easy for them to retaliate against you when you challenge the changes they make to the laws.
Uhm, money is insurable. Is your identity insurable?
The solution is to quit using cards to "identify" people. Identity is necessarily out-of-band data for every real system.
If that system doesn't work, then it's okay to use id cards for single purposes. You only need id cards for a few purposes, and it is not unreasonable to carry one card for each:
Driver's license. It should be issued be the authority that licenses a person to drive.
Credit id. This is something of a new concept, but it should be issued by the authority behind the credit --the bank you consider your primary bank.
Why on earth don't credit cards have pictures on them?
Professional ids. Again, the issuing authority would be the certifying authority.
I'm thinking about voter ids and trying to remember why it would be wrong to have pictures on those, other than cost, and the temptation to use them as a national id.
We need to either get rid of taxes or bring them back to the local level. Then the city could issue a tax id if it found some worth in it. But taxes should not be so vital to a person's well-being that there would be any reason for someone to forge a tax id.
Three ids. Three, maybe four separate ids. When that is not good enough, there is something wrong.
Think about that. PKI requires competency to be used correctly.
And you are depending on the government's incompetency.
In this case, Juki.net _is_ scheduled to be used for everything, eventually, when the storm of protests blows over.
Oh. Did you know that Japan is busy implementing computerized voting machines? Already in operation in several places, and they are just now thinking about maybe implementing a paper trail.
Well, it's all good, until somebody breaches the government's private key.
I don't want my identity dependent on the sleepy guard watching the room where the government's private key is stored.
It's not a transparent system. Why is that so hard for people to understand? Are you also willing to trust the government to use PKI techniques to let you vote over the web?
I'd go for the guy ultimately responsible -- Ballmer.
(Gates, supposedly, wasn't directly involved any more, and it would be really hard to fire him now.
About the closest we could get to that would be to put a lien on all his shares, but what would the court do with shares of Microsoft?
Hmm. Maybe turn them over to the government to sell off to reduce the debt?
Or, maybe, fine him USD 30,000,000,000 so that he has to sell all his shares to pay the fine, and, no, that still doesn't approach the economic damage Gates has done through Microsoft. And the economic damage of so many shares of Microsoft hitting the block at once? Catch-22.)
If you don't get past Genesis, then, no, you're not going to get it. Not really.
The scriptures are not, and should not be sold as, a science textbook.
They're a collection of recollections and philosophizings of people who believe(d) in God.
I can't speak for everybody, but I can mention my experience reading some Isaiah quotes in the Book of Mormon. (And skeptics have fun with those, yes, but they're missing the point, still.)
There's a verse or two in there about men worshipping the creations of their own hands (For instance, Isaiah 2: 20.) and coming to a realization that the things which they have made with there own hands aren't going to save them in the end. I was reading that at about sixteen, while listening to Boston at close to 0 db. (My mom would come in later and told me it was her turn, and I would have to put on something classical, which I am now glad she did. She didn't mind volume if it was good music.) Something about throwing the idols they had made for themselves out of gold and silver to the moles and the bats.
I had built my own speaker enclosures (mid-fi Radio Snake speaker elements) and had replaced some dirty multi-pole switches with flip-flop driven 0-gain transistor switch circuits of my own design. (West Texas is dusty, and we still had refineries south of town burning off by-product gases and such, which definitely led to quick corrosion in the electro-mechanical parts of cheap "mid-fi" amps, etc.
I looked up at the works of my own hands, so to speak, on the shelf over my desk, and I thought about what might move me to throw the stereo system into a hole in the ground, and it hit me, something like it hit me when I figured out that 1+1=2 only holds because we restrict it to the cases where the units are of like objects and non-compressible. And the addition operator something like simple aggregation. And the usual assumptions about comparison.
I've had similar experiences, and those experiences are part of why I believe. The scriptures teach me a truth I need, and then there's nothing to do but accept the truth.
I have never had any such experience about the theory that the creation was seven 24 hour days. Quite the opposite. I noticed, rather, that the light and the dark don't seem to be divided during the first day and night. I also noticed that the word day is often used for other meanings than a time unit of 24 hours.
His reasons are wrong. The methods are wrong, too.
But we do not two internets, one for the clueless to use like a phone, and one for those willing to get a clue to be their own ISPs on.
The latter, of course, is where innovation will occur.
The former will _not_ be secure (and we geeks will have to raise a fuss every time we hear someone say it is). It will simply take a lot more effort than your ordinary user is willing to go to, for them to find their ways onto newsgroups where binaries are allowed and such.
There are lots of things we can do to reduce the windows, for instance, for e-mail address harvesting:
Mailing list servers should act as full mail providers for their members, providing address that are only valid for the mailing list members. The server would forward list mail to the registered members and provide black/white/grey/color-listing for the members. The filtering would be effective because the usual sources and destinations are from a limited set. The list contents could safely be published (even as Google archived newsgroups) because of the filtering.
ISPs should be providing similar filters, and they should not be charging extra, and the number of rules should not be limited to less than twenty. Google and Yahoo already do some filtering, but all ISPs should be filtering.
One problem with current filtering techniques is the lack of support for digging for false positives. Sequestered mail should be viewable, sorted by the reasons for the sequestering: Bad headers in one box, sorted by the types of problems. Suspicious subjects and content in another. Identifiably evil binary attachments in another and un-identifiable (by actual content, not be extension, of course) binaries in yet another. Sender names that include known probably bad patterns in another box, and unknown sender names in a box that's labeled "unknown sender" rather than "junk". Threading junk is also a useful technique.
(This kind of sorting and organization should be provided for archiving legitimate mail, as well, but that's another topic.)
If the service providers were actually providing services we want, unsolicited, illegal scatter pattern advertising mailers would mostly self-destruct. Not entirely, but mostly.
(The most prominent OS and applications vender is the biggest block to progress in this direction, of course.)
Web sites and web browsers are amenable to similar improvements:
Banks should not provide access to accounts through the general purpose browsers. Single-purpose browsers are not that hard to build, and don't need all sorts of bells and whistles. (If you want video feed on your screen while you are on-line to your bank, open up a separate application.)
Of course, the account browser could be even safer if it were in separate hardware, as someone has pointed out
But we do need two general classes of service on the internet.
For what it's worth, none of the Japanese I know whose jobs get anywhere close to rockets and/or nuclear devices are at all interested in revenge for Hiroshima. The few crazies who don't understand MAD and don't understand the economics of letting us handle their defense basically aren't allowed to work any job at all.
There are other things to worry about before worrying about whether Japan had nuclear missiles.
Well, yeah, not nearly enough if everyone were on the same private, of course.
We do assume that comcast doesn't just give up all its current set of IPs and try to put their entire customer base NATted under a single real address. They start NATting by block router or something.
But, then again, we also don't assume this is a smart thing to do.
Now, if I could have suggested a plan for beyond IPv4, I'd have suggested an extensible addressing scheme, some way of either just adding up to 12 extra octets to IPv4, or of concatenating up to three additional IPv4 addresses. Call it IPv4' or something, and if the extension scheme were carefully worked out, heel-draggers would simply be left unable to access parts of the net until they themselves upgraded.
I'm sure there's some technical reason it couldn't have worked.
I'm also sure a lot of people don't understand the difference between a truly extensible scheme, such as I might have suggested, and NAT. Maybe some don't understand the difference between ports and IP addresses, for instance.
I can get an AMD notebook next time I can afford a notebook, put Fedora Core on it.
There are two reasons I still use Mac OS. Claris/AppleWorks and Kotoeri. Anthy is close. I'm about two years away on figuring out a Java solution to the stuff I do in Claris/AppleWorks. (And, no, MSOffice/VB does not do the job.)
Slashdot Mirror
apple.laf.useScreenMenuBar
Can be set on the command line or programmatically. Or in the plist in the bundle, of course.
Well, I remember, back in the 80s, fans of FORTH who claimed it was faster than compiled C.
Some of those who made those claims understood that the FORTH interpreter was a better run-time execution model than the model that many C compilers for microcomputers back then implemented. So, FORTH interpreted code could be faster than a bad compiler's compiled code.
But some didn't seem to understand the fundamental behavior of an interpreter. I heard one guy talk about, if your program was too slow, just make parts of it words of their own, and then those words would be fast. (Didn't seem to understand that a call to a routine still had to execute the body of the routine. Or maybe he was confusing size optimization with speed optimization. I've done similar things, said "It does cool thing A." when I meant "It does cool thing B".)
There's no way an interpreter can run without putting at least a branch/jump between every primitive executed.
Well, I know the JIT is supposed to solve that, but why not just unroll the invariant primitives? If the code itself is final, it ought to be possible to just pick up everything to the last branch and insert it in the stored instruction stream without the branch. Large primitives aren't going to be helped much, but small primitives would. And, at compile time, you could dredge the unrolled code for things like a move to the parameter stack followed directly by a move from the stack to an accumulator.
Anecdotes: My Mac Mini (1.2G PPC, 512M RAM, slow frontside cache) runs Java about half the speed of my Linux box (1.7G Sempron 2700+, 768M RAM, frontside about half CPU speed). The pause between double-click and application coming up on screen is a bit longer than with native code on both machines, which is not surprising considering the pre-flighting Java does for security and other purposes. You see the effects of the branch being in there, speed-wise, but, as you say, it's fast enough.
On my clamshell iBook, however, rendering of html is seriously slow. Makes the help screen I'm rendering that way more or less useless.
When apple owns the portable phone market, then, yes, this would become subject to anti-trust measures.
Why can't people tell the difference between a monopoly in the market for brand X of product B of product category Q and a monopoly in the market for the entire product category.
iPods are kind of close to the borderline, but not really approaching the degree to which MSWindows/MSOffice has become the market for general purpose computers except for mainframes. And, yes, Apple has used various things to keep control of their own products, but have done nothing approaching what Microsoft has done to establish and maintain their monopoly. Give it a break.
Maybe we should just quit assuming that money isn't the only medium for communicating value and reward.
Money this. Money that. Money the other.
I'd just like people to let me do my job right.
Big money at the end of the long ride for the lucky few is a good way to screw the whole race up. We should not all be running the same race.
Small prizes for small steps.
Better yet, just get rid of the stupid overtime psychology and let us have enough time after hours to pursue the hard problems for the reasons that are most likely to bring real success:
Solving a problem is a reward in and of itself.
And a solution that helps others gives something even better than warm fuzzies. (Linux, anyone? BSD?)
As I read the reports I've seen, your local government offered you the choice, but the supreme court just said that they didn't have the authority to do so.
A little entry point on the subject and problems of caste in Japan:
http://en.wikipedia.org/wiki/Four_divisions_of_society
http://en.wikipedia.org/wiki/Burakumin
He just presented you with a very good representative version of the argument against this kind of ID.
How do you ID your partner?
To make it more obvious, how do you ID [insert role or person here]?
Frankly, assuming the existence of such ID cards, I'm not going to laugh at all when girls want to see a guy's ID before going out with him the first time. It's a logical extension of the concept. And girls should be less trusting of the guys they go out with up front, really.
But it makes much more sense for the girl to talk with other girls who know the guy or whatever.
Actual identity is out-of-band data in any system we can build, computerized or not.
Building systems to manage identity on a limited basis in a localized scope may make some sense.
Universal ID is an oxymoron.
Unifying the ID system increases the value of the ID card well beyond the cost to obtain one illegitimately.
And the people to whom it becomes most immediately valuable are the people who have some vested interested in keeping their party in office, which is the very people who are issuing the cards.
Something they did thirty or so years back (Working from very poor memory, it was not my record, but my manager's at radio snack.):
In case you can't tell, I don't think much of your plan for how your girlfriend/wife/partner is supposed to figure out that the imposter is not you.
I am not familiar enough with the Japanese constitution to know what kind of legal issues Juki.net raises, but I have to be wondering why they were keeping it quiet so long.
And I realize now that I should have been making a fuss about a job I had four years ago.
Anyway, a bit off-topic, but that the US national government is using various loopholes (like federal, vs. national) to sidestep the Constitution on these sorts of things is evidence that the US government is out of control, and therefore that the paranoia is based in reality.
This is the whole problem. The bigger the system the easier it is to crash it.
Absolute power doesn't just corrupt people.
Mass is real, whether it's data or gears.
The problems occur when they decide to change their core business.
No, there is not much keeping the _people_ implementing the law from changing it, especially when you make it easy for them to retaliate against you when you challenge the changes they make to the laws.
So, for years, the government has been surreptitiously watching me. Now they want to be be able to watch me with impunity.
I think you are the one who has been brain-washed. Brain washed right out of you, maybe?
(Sorry to be rude, but I'm hoping to stimulate to think, here.)
So, you're saying my identity should be all contained in a card?
I'm going to let someone mark me as a troll while you think about that.
Uhm, money is insurable. Is your identity insurable?
The solution is to quit using cards to "identify" people. Identity is necessarily out-of-band data for every real system.
If that system doesn't work, then it's okay to use id cards for single purposes. You only need id cards for a few purposes, and it is not unreasonable to carry one card for each:
Driver's license. It should be issued be the authority that licenses a person to drive.
Credit id. This is something of a new concept, but it should be issued by the authority behind the credit --the bank you consider your primary bank.
Why on earth don't credit cards have pictures on them?
Professional ids. Again, the issuing authority would be the certifying authority.
I'm thinking about voter ids and trying to remember why it would be wrong to have pictures on those, other than cost, and the temptation to use them as a national id.
We need to either get rid of taxes or bring them back to the local level. Then the city could issue a tax id if it found some worth in it. But taxes should not be so vital to a person's well-being that there would be any reason for someone to forge a tax id.
Three ids. Three, maybe four separate ids. When that is not good enough, there is something wrong.
Think about that. PKI requires competency to be used correctly.
And you are depending on the government's incompetency.
In this case, Juki.net _is_ scheduled to be used for everything, eventually, when the storm of protests blows over.
Oh. Did you know that Japan is busy implementing computerized voting machines? Already in operation in several places, and they are just now thinking about maybe implementing a paper trail.
Well, it's all good, until somebody breaches the government's private key.
I don't want my identity dependent on the sleepy guard watching the room where the government's private key is stored.
It's not a transparent system. Why is that so hard for people to understand? Are you also willing to trust the government to use PKI techniques to let you vote over the web?
I'd go for the guy ultimately responsible -- Ballmer.
(Gates, supposedly, wasn't directly involved any more, and it would be really hard to fire him now.
About the closest we could get to that would be to put a lien on all his shares, but what would the court do with shares of Microsoft?
Hmm. Maybe turn them over to the government to sell off to reduce the debt?
Or, maybe, fine him USD 30,000,000,000 so that he has to sell all his shares to pay the fine, and, no, that still doesn't approach the economic damage Gates has done through Microsoft. And the economic damage of so many shares of Microsoft hitting the block at once? Catch-22.)
If you don't get past Genesis, then, no, you're not going to get it. Not really.
The scriptures are not, and should not be sold as, a science textbook.
They're a collection of recollections and philosophizings of people who believe(d) in God.
I can't speak for everybody, but I can mention my experience reading some Isaiah quotes in the Book of Mormon. (And skeptics have fun with those, yes, but they're missing the point, still.)
There's a verse or two in there about men worshipping the creations of their own hands (For instance, Isaiah 2: 20.) and coming to a realization that the things which they have made with there own hands aren't going to save them in the end. I was reading that at about sixteen, while listening to Boston at close to 0 db. (My mom would come in later and told me it was her turn, and I would have to put on something classical, which I am now glad she did. She didn't mind volume if it was good music.) Something about throwing the idols they had made for themselves out of gold and silver to the moles and the bats.
I had built my own speaker enclosures (mid-fi Radio Snake speaker elements) and had replaced some dirty multi-pole switches with flip-flop driven 0-gain transistor switch circuits of my own design. (West Texas is dusty, and we still had refineries south of town burning off by-product gases and such, which definitely led to quick corrosion in the electro-mechanical parts of cheap "mid-fi" amps, etc.
I looked up at the works of my own hands, so to speak, on the shelf over my desk, and I thought about what might move me to throw the stereo system into a hole in the ground, and it hit me, something like it hit me when I figured out that 1+1=2 only holds because we restrict it to the cases where the units are of like objects and non-compressible. And the addition operator something like simple aggregation. And the usual assumptions about comparison.
I've had similar experiences, and those experiences are part of why I believe. The scriptures teach me a truth I need, and then there's nothing to do but accept the truth.
I have never had any such experience about the theory that the creation was seven 24 hour days. Quite the opposite. I noticed, rather, that the light and the dark don't seem to be divided during the first day and night. I also noticed that the word day is often used for other meanings than a time unit of 24 hours.
Actually, he's sort of right.
His reasons are wrong. The methods are wrong, too.
But we do not two internets, one for the clueless to use like a phone, and one for those willing to get a clue to be their own ISPs on.
The latter, of course, is where innovation will occur.
The former will _not_ be secure (and we geeks will have to raise a fuss every time we hear someone say it is). It will simply take a lot more effort than your ordinary user is willing to go to, for them to find their ways onto newsgroups where binaries are allowed and such.
There are lots of things we can do to reduce the windows, for instance, for e-mail address harvesting:
Mailing list servers should act as full mail providers for their members, providing address that are only valid for the mailing list members. The server would forward list mail to the registered members and provide black/white/grey/color-listing for the members. The filtering would be effective because the usual sources and destinations are from a limited set. The list contents could safely be published (even as Google archived newsgroups) because of the filtering.
ISPs should be providing similar filters, and they should not be charging extra, and the number of rules should not be limited to less than twenty. Google and Yahoo already do some filtering, but all ISPs should be filtering.
One problem with current filtering techniques is the lack of support for digging for false positives. Sequestered mail should be viewable, sorted by the reasons for the sequestering: Bad headers in one box, sorted by the types of problems. Suspicious subjects and content in another. Identifiably evil binary attachments in another and un-identifiable (by actual content, not be extension, of course) binaries in yet another. Sender names that include known probably bad patterns in another box, and unknown sender names in a box that's labeled "unknown sender" rather than "junk". Threading junk is also a useful technique.
(This kind of sorting and organization should be provided for archiving legitimate mail, as well, but that's another topic.)
If the service providers were actually providing services we want, unsolicited, illegal scatter pattern advertising mailers would mostly self-destruct. Not entirely, but mostly.
(The most prominent OS and applications vender is the biggest block to progress in this direction, of course.)
Web sites and web browsers are amenable to similar improvements:
Banks should not provide access to accounts through the general purpose browsers. Single-purpose browsers are not that hard to build, and don't need all sorts of bells and whistles. (If you want video feed on your screen while you are on-line to your bank, open up a separate application.)
Of course, the account browser could be even safer if it were in separate hardware, as someone has pointed out
But we do need two general classes of service on the internet.
So, what you're saying is that iNTEL's ability to flood the market is the only thing keeping them on top.
Right?
I find that large jumps in technology often leave the rabbit behind, but I'm not sure I think it's a good thing.
For what it's worth, none of the Japanese I know whose jobs get anywhere close to rockets and/or nuclear devices are at all interested in revenge for Hiroshima. The few crazies who don't understand MAD and don't understand the economics of letting us handle their defense basically aren't allowed to work any job at all.
There are other things to worry about before worrying about whether Japan had nuclear missiles.
Well, yeah, not nearly enough if everyone were on the same private, of course.
We do assume that comcast doesn't just give up all its current set of IPs and try to put their entire customer base NATted under a single real address. They start NATting by block router or something.
But, then again, we also don't assume this is a smart thing to do.
Now, if I could have suggested a plan for beyond IPv4, I'd have suggested an extensible addressing scheme, some way of either just adding up to 12 extra octets to IPv4, or of concatenating up to three additional IPv4 addresses. Call it IPv4' or something, and if the extension scheme were carefully worked out, heel-draggers would simply be left unable to access parts of the net until they themselves upgraded.
I'm sure there's some technical reason it couldn't have worked.
I'm also sure a lot of people don't understand the difference between a truly extensible scheme, such as I might have suggested, and NAT. Maybe some don't understand the difference between ports and IP addresses, for instance.
No. Not really.
I can get an AMD notebook next time I can afford a notebook, put Fedora Core on it.
There are two reasons I still use Mac OS. Claris/AppleWorks and Kotoeri. Anthy is close. I'm about two years away on figuring out a Java solution to the stuff I do in Claris/AppleWorks. (And, no, MSOffice/VB does not do the job.)
at it again