If you used three or four different hash functions, to produce a long combined message digest (concatenated from the several message digests), it would seem to be much more difficult to produce files that collide. An attack resulting in a collision on one algorithm, is not likely to work for another algorithm. Finding two files that simultaneously fool two or more algorithms seems a much more difficult task than just fooling a single algorithm like MD5.
This is of utmost importance. We can't have the RIAA infiltrating garbage files ("what the fiddlesticks do you think you're doing?") into our p2p networks.
No. I was halfway kidding, and halfway serious.
If I have plenty of bandwidth and cpu, then I would like the option of making FireFox pretend like it downloaded and "displayed" the ad, and any pop up windows.
Just because FireFox could be made to retrieve pop up windows, "click" on ads, and download the ad pages, does not mean that I have to see any of this on my screen.
The advertisers would suddenly see FireFox users as having the largest number of clickthroughs of ads.
Linus' comments in the kernel's CodingStyle document are relevant too -- try to keep your functions so that you can read most of them on one screen. Python allows you to do this more often (I find) without resorting to strange brace positions.
its called evolution, if all the companies compete to make the best product, it brings down prices and new technology is developed.
its a win-win situation for the consumer, i welcome it.
I do not welcome our new disk format overlords.
Look what happened to quadraphonic in the 70's. By the time all the smoke of standards wars cleared, all that was left was plain stereo.
There must be other examples of great ideas that died due to multiple competing standards?
I'm no fan of SCO, but what exactly did SCO buy? And who owns the rights to any additions that SCO made?
SCO bought a business of selling and maintaining the licenses. Novell gets 95 % of all of the royalty payments on these licenses. (Does that sound like SCO owns unix when SCO only keeps 5 % of royalty payments?)
But SCO paid all that money!
Look how much money Novell paid for Unix. Nearly a Beeelion US dollars. Now how much did SCO pay for the right to sell and maintain the licenses -- a whole lot less. Judge Kimball already said in (8) of the order that the APA clearly did not transfer copyrights.
Ammendment 2 by itself is not a copyright transfer instrument. That ammendment was to ensure that SCO had something saying that they had the copyrights necessary to exercise their rights in the unix licensing business. That way, by making copies of unix, SCO is not infringing on a copyright that they do not own.
SCO also bought rights to make derrivative works, which they would own.
IBM bought their unix from AT&T in 1985, long before the Novell-oldSantaCruseOp deal. So any additions that SCO made are not in IBM's code.
>>"Think not that I am come to send peace on earth: I came not to send peace, but a sword." -- Mathew 10:34
>what the heck is that supposed to mean?
If you look at the context, versus 34 through, say, about 37, you can see exactly what it means.
Jesus didn't come to bring unity, but to bring division. He goes on to detail that division, between husband and wife, division between families, etc. Then in v.37 it is clear that the meaning is that He calls you to serve Him. That is the division. Some will, and some won't. There will be division between those who choose to believe and those who will never believe. Even within families.
On my reading of it, that appears to be the plain simple intended meaning.
How do you and your friends manage to keep your songs original and not accidentally copied?
May I suggest a solution to this problem? Purely in the interest of protecting copyrights, of course.
If you think that you might ever write any music, then first, download every possible mp3 you can get from the internet. Listen to them regularly. This way, you can be sure that when or if you write your own music, you aren't infringing anyone else's music copyright.
But please, download all these mp3s only in the interest of protecting the RIAA members' copyrights. Think of the poor record labels.
The main reason they do it is for bragging rights.
Some people might do it for money. In this case, they probably would also be keeping their mouth shut.
What you say? Yes, money. I increasingly believe that the sophistication of some of the new attacks (have you actually read the explanations of how some of these attacks work?) could only be the result of a funded development effort. Who would pay to develop such things? (1) spammers, (2) evil communist terrorist open....er... um.. nevermind.:-)
Now wouldn't it be nice if a standard were made and users could be assured that, for the most part, regardless of what distribution they're using:
apt is available,
A consistent filesystem hierarchy is followed from distribution to distribution, and
A large number of packages are available (and, more importantly, compatible) due to point 2.
Yeah, wouldn't it be nice. And if anyone creates (or has created, i.e. Knoppix, etc.) such a system, then it would be referred to as....
A Debian-based distribution.
So then we still wouldn't have solved the problem you mention of a general solution for Linux, vs. a distribution-specific solution.
I think that any attempt at a general solution automatically becomes a distribution specific solution. Either for an existing distribution, or a new one that you would be trying to promote.
Indexing your PC is one thing. The topic. If I save, say, OpenOffice.org documents on the Internet, then protocols will need to evolve to make the rich metadata and search capabilities available, so that I can just as transparently mount a "Samba" share (say in Samba version 12), and still give queries such as "Show me all the documents that mention any of our employees who happen to live in New York, are over 9 feet tall, and have green hair." From my point of view, a part of the filesystem should always offer the same rich capabilities. I don't care whether it's remotely mounted on the Internet via. Samba, NFS, SFTP, WebDAV, or whatever. When I put my end user hat on, I just want my artificially intelligent, speech recognition enabled computer to fetch the correct group of documents I asked for. It's about usability.
I'm really unclear on what OS Ring protection has to do with it? I'm assuming we're thinking about the same very low-level thing here?
I don't see what fundamental I/O bottlenecks there are. Most of the time on my desktop, it is doing very little I/O. When I pick the File --> Save command, a flurry of disk activity ensues. So what if three times as much activity happend? What if hard disks get faster and cheaper with each passing day?
Since I well remember the days of full blown accounting systems that had to run in 20 MB disk and less than 64 K of RAM, I find today's hardware quite generous. The hardware capability always gets soaked up by newer software that has better usability features. Look at today's software vs. 20 years ago (1984) software. Despite the predictions of the end of Moore's Law, I don't see any reason to expect that computers won't continue to get bigger, faster, and cheaper. See the recent story about dual-cores on a single socket? I suspect eventually every ordinary desktop PC will have first two, and later four, and later eight processors. The latest dick measuring contents between testosterone posioned adolescents (see Slashdot) will go like... "Yeah, but MYYYYY system has 64 processors and 8 drives!". I'm not sure what fundamental I/O problem there will be.
Basically, I see an amazing future. Because 20 years ago, I would have called you a fool if you had told me that by the 21st century everyone would have 2.0 GHz machines with 512 MB of RAM, and 80 GB disks, and that this would be commonplace! Therefore, I'm all in favor of database filesystems, rich metadata, standardized libraries to support it, and changes to all the old tools (such as the "cp" command and "ls" command) to support the new capabilities. When I "ls", I should be able to see not only the name, bytes, date, attributes, etc., I should be able to see the mimetype, the search keywords, or whatever other arbitrary metadata that was attached to the file. When I "cp" the file, all these attributes should get transferred, just as the date and permissions get transferred.
If you're going to open the computer's case, it is easier to remove the HD and use a jury-rigged "external usb enclosure" canabalized guts to connect it to your laptop, then steal the data onto your laptop.
I'm not making an argument that efficiency doesn't matter. I'm not addressing efficiency at all. I'm all for efficient indexing algorithms. Efficient graphics algorithms. Efficient everything.
What I'm saying is that just because an extremely useful feature requires three times as much disk space as the raw data doesn't matter to most people when that disk space is sufficiently cheap, and backup of it is sufficiently cheap. (For any reasonable definition of "sufficiently".)
Despite all the complaints back in 1984 that the Macintosh is so inefficient (in terms of how much cpu horsepower and memory it takes to do the same job as MS-DOS); the implementation of the Mac 1984 GUI is downright efficient compared to much of what we see today that everyone just takes for granted. This despite the complaints of how the machine needed a minimum of 128 K (not meg) of memory, or even more to do much serious work.
When terrabytes of memory and petabytes of disk space can be purchesed for $50 in a blister pack at Micro Center, nobody is going to care that all file systems index their data multiple times over in various ways. Seriously, though, we won't have to wait that long. If the actual overhead is as expensive as two or three times the raw data, even this will seem cheap at the price for the wonderful features it will bring to the users.
Compare how much memory Microsoft Word takes compared to how much memory Edlin (for DOS) takes. Edlin is perfectly suitable for editing text files, memos, business letters, etc. And after all, memory is expensive. Everyone should worry about this inefficiency and switch back to Edlin.
Want to keep your data secure? How about banning some really dangerous external devices. Like the Keyboard and Mouse. (Except for the essential Ctrl, Alt and Del keys, of course.)
The other problem is that it's not uncommon in the database world to spend far more disk indexing complex data for access than it actually takes to store the raw information itself.
So what?
Do you really want the possibility that your inseperable all-in-one file system is using more space for the equivalent of directory entries than for data itself?
I don't really care.
Let me explain.
Back in Feb 1984, when the Macintosh first began shipping, there were lots of arguments very similar to some of the ones you are making (in the parent post).
Example argument: But think how inefficient it is just to draw one character onto the display! The processor has to read the glyph for the font, and then individually set the right pixels in the display bitmap. Lots of other complex graphics processing as well. The font's character may be partially clipped by a window stacked on top of it. The pixels to set for the font have to be loacted within the display frame buffer bitmap -- offset by the window location, and possibly other nested panels within the window where the text is drawn. On an IBM PC, you just change one byte in the textmode frame buffer and the video card's character generator displays the character on your green-screen monitor.
Now fast forward to the 21st century. Does anybody care? Even modern Linux now runs pure "textmode" console screens in graphics mode and the kernel manipulates the individual pixels in a graphics framebuffer (and displays a nice penguin graphic in the same framebuffer). Nobody even uses hardware text mode anymore.
My Point...
Stop caring so much about machine efficiency and start caring about usability.
I don't really care if the search indexing system takes ten times, or one hundred times, or one thousand times the disk space as the raw data. As long as today's hardware can handle it, I don't really care. If today's hardware can't handle it, then tomorrow's hardware will, and we'll see these innovations appear as soon as common everyday hardware can handle it. All I really care about is that I can ask the computer to: "Show me all OpenOffice.org Writer documents, where I mention the name of any of our employees who live in New York, are over 9 feet tall, and have green hair.".
Remember this isn't about special cases like a user too lazy to sort their home directory or documents folder
Like lazy Mac (or Windows) users too lazy to learn convenient DOS commands.
Make the machine fit the man, not the man fit the machine.
In fifty years, there will be people complaining because our artificially intelligent speech and vision enabled applications, that can instantly access all the world's data and intelligently filter it, and even think about problems for us, require "bloated" computers using 500 Petabytes of disk storage, and 256 Terabytes of RAM. That is just too bloated. I prefer the old applications that only required 800 Terabytes of disk, and merely 16 Terabytes of RAM. Ah, those were the days of machine efficiency, back when a computer and its applications fit into a nice "slim" configuration of only 16 Terabytes of RAM. Ahhhhh.
Why? Because servicemen at war might be more likely to vote against a current administration that sent them there. Therefore, it needs to be easier to locate the "bad voters", or easier to "correct" their votes.
There is a reason for this. Servicemen might be more likely than the general population to vote against the current administration. We cannot have this. Therefore, it needs to be more difficult for servicemen to vote, or alternately, it needs to be easier to find out how the "bad voters" are, or at least "intercept and change" their votes.
Slashdot Mirror
I was thinking the same thing.
See the long list of cryptographic hash functions on Wikipedia. (Bottom of page.)
If you used three or four different hash functions, to produce a long combined message digest (concatenated from the several message digests), it would seem to be much more difficult to produce files that collide. An attack resulting in a collision on one algorithm, is not likely to work for another algorithm. Finding two files that simultaneously fool two or more algorithms seems a much more difficult task than just fooling a single algorithm like MD5.
This is of utmost importance. We can't have the RIAA infiltrating garbage files ("what the fiddlesticks do you think you're doing?") into our p2p networks.
No. I was halfway kidding, and halfway serious. If I have plenty of bandwidth and cpu, then I would like the option of making FireFox pretend like it downloaded and "displayed" the ad, and any pop up windows. Just because FireFox could be made to retrieve pop up windows, "click" on ads, and download the ad pages, does not mean that I have to see any of this on my screen. The advertisers would suddenly see FireFox users as having the largest number of clickthroughs of ads.
Why can't FireFox be made to auto-clickthrough, and even go through the motions (from the server's pov) of downloading and "displaying" the ad?
Then advertisers would love FireFox as much as users do.
Linus' comments in the kernel's CodingStyle document are relevant too -- try to keep your functions so that you can read most of them on one screen. Python allows you to do this more often (I find) without resorting to strange brace positions.
My functions do each fit on one screen.
For any sufficiently large definition of screen.
if all else fails blame Microsoft
That is just not right. People should not blame Microsoft when all else fails. It is not how things should be done.
People should blame Microsoft first, and not bother trying all else.
its called evolution, if all the companies compete to make the best product, it brings down prices and new technology is developed. its a win-win situation for the consumer, i welcome it.
I do not welcome our new disk format overlords.
Look what happened to quadraphonic in the 70's. By the time all the smoke of standards wars cleared, all that was left was plain stereo.
There must be other examples of great ideas that died due to multiple competing standards?
You forgot...
7. Profit
I'm no fan of SCO, but what exactly did SCO buy? And who owns the rights to any additions that SCO made?
SCO bought a business of selling and maintaining the licenses. Novell gets 95 % of all of the royalty payments on these licenses. (Does that sound like SCO owns unix when SCO only keeps 5 % of royalty payments?)
But SCO paid all that money!
Look how much money Novell paid for Unix. Nearly a Beeelion US dollars. Now how much did SCO pay for the right to sell and maintain the licenses -- a whole lot less. Judge Kimball already said in (8) of the order that the APA clearly did not transfer copyrights.
Ammendment 2 by itself is not a copyright transfer instrument. That ammendment was to ensure that SCO had something saying that they had the copyrights necessary to exercise their rights in the unix licensing business. That way, by making copies of unix, SCO is not infringing on a copyright that they do not own.
SCO also bought rights to make derrivative works, which they would own.
IBM bought their unix from AT&T in 1985, long before the Novell-oldSantaCruseOp deal. So any additions that SCO made are not in IBM's code.
when SCO bleeds enough, Novell can reacquire all Unix rights cheaply
Re-acquire?
Novell already owns the Unix copyrights.
Let me get this straight.
Gradually migrate desktops to Linux. Make them do sign on and authentication to a Windows server.
End result: Linux on the desktops, Windows as the server.
That way, each platform is being used for what it is best at.
>>"Think not that I am come to send peace on earth: I came not to send peace, but a sword." -- Mathew 10:34
>what the heck is that supposed to mean?
If you look at the context, versus 34 through, say, about 37, you can see exactly what it means.
Jesus didn't come to bring unity, but to bring division. He goes on to detail that division, between husband and wife, division between families, etc. Then in v.37 it is clear that the meaning is that He calls you to serve Him. That is the division. Some will, and some won't. There will be division between those who choose to believe and those who will never believe. Even within families.
On my reading of it, that appears to be the plain simple intended meaning.
How do you and your friends manage to keep your songs original and not accidentally copied?
May I suggest a solution to this problem? Purely in the interest of protecting copyrights, of course.
If you think that you might ever write any music, then first, download every possible mp3 you can get from the internet. Listen to them regularly. This way, you can be sure that when or if you write your own music, you aren't infringing anyone else's music copyright.
But please, download all these mp3s only in the interest of protecting the RIAA members' copyrights. Think of the poor record labels.
I want to sing the general tone of a song I heard on the radio in a microphone and have google direct me to that album on froogle.
THAT would be awesome!
I want to sing the general tone of a song that I heard on the radio in a microphone and have Kazaa direct me to that album on their download network.
That would be awesome!
Why do you find it so unbelievable?
As an American, I find it unbelievable that Americans are so puritanical about all things related to.... um... sex.
Straight American males are generally obsessed about any perception of them being slightly gay or interested in anything homoerotic.
Look at the scandal over a president's affair.
Now given the righteous condemnation of anything sexual, and probably lack of any experimentation, should you really be surprised?
The main reason they do it is for bragging rights.
:-)
Some people might do it for money. In this case, they probably would also be keeping their mouth shut.
What you say? Yes, money. I increasingly believe that the sophistication of some of the new attacks (have you actually read the explanations of how some of these attacks work?) could only be the result of a funded development effort. Who would pay to develop such things? (1) spammers, (2) evil communist terrorist open....er... um.. nevermind.
Yeah, wouldn't it be nice. And if anyone creates (or has created, i.e. Knoppix, etc.) such a system, then it would be referred to as....
A Debian-based distribution.
So then we still wouldn't have solved the problem you mention of a general solution for Linux, vs. a distribution-specific solution.
I think that any attempt at a general solution automatically becomes a distribution specific solution. Either for an existing distribution, or a new one that you would be trying to promote.
Yes, you can access ntfs from Knoppix. See icon on the desktop. Double click.
Indexing your PC is one thing. The topic. If I save, say, OpenOffice.org documents on the Internet, then protocols will need to evolve to make the rich metadata and search capabilities available, so that I can just as transparently mount a "Samba" share (say in Samba version 12), and still give queries such as "Show me all the documents that mention any of our employees who happen to live in New York, are over 9 feet tall, and have green hair." From my point of view, a part of the filesystem should always offer the same rich capabilities. I don't care whether it's remotely mounted on the Internet via. Samba, NFS, SFTP, WebDAV, or whatever. When I put my end user hat on, I just want my artificially intelligent, speech recognition enabled computer to fetch the correct group of documents I asked for. It's about usability.
I'm really unclear on what OS Ring protection has to do with it? I'm assuming we're thinking about the same very low-level thing here?
I don't see what fundamental I/O bottlenecks there are. Most of the time on my desktop, it is doing very little I/O. When I pick the File --> Save command, a flurry of disk activity ensues. So what if three times as much activity happend? What if hard disks get faster and cheaper with each passing day?
Since I well remember the days of full blown accounting systems that had to run in 20 MB disk and less than 64 K of RAM, I find today's hardware quite generous. The hardware capability always gets soaked up by newer software that has better usability features. Look at today's software vs. 20 years ago (1984) software. Despite the predictions of the end of Moore's Law, I don't see any reason to expect that computers won't continue to get bigger, faster, and cheaper. See the recent story about dual-cores on a single socket? I suspect eventually every ordinary desktop PC will have first two, and later four, and later eight processors. The latest dick measuring contents between testosterone posioned adolescents (see Slashdot) will go like... "Yeah, but MYYYYY system has 64 processors and 8 drives!". I'm not sure what fundamental I/O problem there will be.
Basically, I see an amazing future. Because 20 years ago, I would have called you a fool if you had told me that by the 21st century everyone would have 2.0 GHz machines with 512 MB of RAM, and 80 GB disks, and that this would be commonplace! Therefore, I'm all in favor of database filesystems, rich metadata, standardized libraries to support it, and changes to all the old tools (such as the "cp" command and "ls" command) to support the new capabilities. When I "ls", I should be able to see not only the name, bytes, date, attributes, etc., I should be able to see the mimetype, the search keywords, or whatever other arbitrary metadata that was attached to the file. When I "cp" the file, all these attributes should get transferred, just as the date and permissions get transferred.
If you're going to open the computer's case, it is easier to remove the HD and use a jury-rigged "external usb enclosure" canabalized guts to connect it to your laptop, then steal the data onto your laptop.
I'm not making an argument that efficiency doesn't matter. I'm not addressing efficiency at all. I'm all for efficient indexing algorithms. Efficient graphics algorithms. Efficient everything.
What I'm saying is that just because an extremely useful feature requires three times as much disk space as the raw data doesn't matter to most people when that disk space is sufficiently cheap, and backup of it is sufficiently cheap. (For any reasonable definition of "sufficiently".)
Despite all the complaints back in 1984 that the Macintosh is so inefficient (in terms of how much cpu horsepower and memory it takes to do the same job as MS-DOS); the implementation of the Mac 1984 GUI is downright efficient compared to much of what we see today that everyone just takes for granted. This despite the complaints of how the machine needed a minimum of 128 K (not meg) of memory, or even more to do much serious work.
When terrabytes of memory and petabytes of disk space can be purchesed for $50 in a blister pack at Micro Center, nobody is going to care that all file systems index their data multiple times over in various ways. Seriously, though, we won't have to wait that long. If the actual overhead is as expensive as two or three times the raw data, even this will seem cheap at the price for the wonderful features it will bring to the users.
Compare how much memory Microsoft Word takes compared to how much memory Edlin (for DOS) takes. Edlin is perfectly suitable for editing text files, memos, business letters, etc. And after all, memory is expensive. Everyone should worry about this inefficiency and switch back to Edlin.
Priorities wrong indeed.
Want to keep your data secure? How about banning some really dangerous external devices. Like the Keyboard and Mouse. (Except for the essential Ctrl, Alt and Del keys, of course.)
Can Windows also prevent me from booting a Knoppix CD to copy files to my USB device?
The other problem is that it's not uncommon in the database world to spend far more disk indexing complex data for access than it actually takes to store the raw information itself.
So what?
Do you really want the possibility that your inseperable all-in-one file system is using more space for the equivalent of directory entries than for data itself?
I don't really care.
Let me explain.
Back in Feb 1984, when the Macintosh first began shipping, there were lots of arguments very similar to some of the ones you are making (in the parent post).
Example argument: But think how inefficient it is just to draw one character onto the display! The processor has to read the glyph for the font, and then individually set the right pixels in the display bitmap. Lots of other complex graphics processing as well. The font's character may be partially clipped by a window stacked on top of it. The pixels to set for the font have to be loacted within the display frame buffer bitmap -- offset by the window location, and possibly other nested panels within the window where the text is drawn. On an IBM PC, you just change one byte in the textmode frame buffer and the video card's character generator displays the character on your green-screen monitor.
Now fast forward to the 21st century. Does anybody care? Even modern Linux now runs pure "textmode" console screens in graphics mode and the kernel manipulates the individual pixels in a graphics framebuffer (and displays a nice penguin graphic in the same framebuffer). Nobody even uses hardware text mode anymore.
My Point...
Stop caring so much about machine efficiency and start caring about usability.
I don't really care if the search indexing system takes ten times, or one hundred times, or one thousand times the disk space as the raw data. As long as today's hardware can handle it, I don't really care. If today's hardware can't handle it, then tomorrow's hardware will, and we'll see these innovations appear as soon as common everyday hardware can handle it. All I really care about is that I can ask the computer to: "Show me all OpenOffice.org Writer documents, where I mention the name of any of our employees who live in New York, are over 9 feet tall, and have green hair.".
Remember this isn't about special cases like a user too lazy to sort their home directory or documents folder
Like lazy Mac (or Windows) users too lazy to learn convenient DOS commands.
Make the machine fit the man, not the man fit the machine.
In fifty years, there will be people complaining because our artificially intelligent speech and vision enabled applications, that can instantly access all the world's data and intelligently filter it, and even think about problems for us, require "bloated" computers using 500 Petabytes of disk storage, and 256 Terabytes of RAM. That is just too bloated. I prefer the old applications that only required 800 Terabytes of disk, and merely 16 Terabytes of RAM. Ah, those were the days of machine efficiency, back when a computer and its applications fit into a nice "slim" configuration of only 16 Terabytes of RAM. Ahhhhh.
Why? Because servicemen at war might be more likely to vote against a current administration that sent them there. Therefore, it needs to be easier to locate the "bad voters", or easier to "correct" their votes.
There is a reason for this. Servicemen might be more likely than the general population to vote against the current administration. We cannot have this. Therefore, it needs to be more difficult for servicemen to vote, or alternately, it needs to be easier to find out how the "bad voters" are, or at least "intercept and change" their votes.