It's much easier to find one copy out of ten linke than it is to find one copy out of nothing at all.
Users have to find ONE copy. Then they stop looking and have fun with the software.
MPAA wants to find ALL copies. They want to name everyone with deCSS on their homepage in the lawsuit.
However, the intention is that the judge will laugh them out of court if their list contains 50% or even 10% "legit" deCSS copies. Moreover, I'd recommend everyone to switch to publishing the other deCSS the moment his name shows up on an MPAA list....
the result is that the average slashdotter works between 6.6 and 7.7 hours a day, depending on how you estimate "less than 8 hours". Put it at 4, and the average slashdotter works 7.2 hours a day....
I tried installing a new machine with the GUI (I swear I didn't know I was going to get a GUI. I just popped in the first CD like I always do...).
It omits a few questions that I find essential. I fogot which. One thing I do remember is that it tells me that the root password that I've been using for way too long now has illegal characters. Jee, and I've been using those illegal characters all those years, and NOW it tells me!
They have been trying to get things to walk, just by letting "trial and error" evolve. This is an advancement.
Now they are a bit further by having a 14-motor ape swing from rung to rung on a ladder. Some people may be afraid for AIs, but then they should just protest. I mean, this is just another step on the big ladder.
The first time a chunk of code is run, there's a performance hit as it's translated to the native instruction set for the particular chip on which it's running.
Nope. the first few times it is executed it takes a performance hit as if it is interpreted. As that's exactly what the transmeta CPU is doing.
Only after a few times the code is going into the compiler. And after a few hundred (my guess) times the optimizer is kicked into action....
If you think of a cache like a window on a portion of the code, and then you can imagine that only a small portion of the code (less than 1K) would be kept in the cache. The TM3120 has 108KB of cache, but the TM5400 has 400KB cache, which sounds like a good amount.
Crusoe uses about seven to fifteen Megabytes of code-cache. The 1M codemorpher is copied from ROM to sdram, and then the system uses 8 to 16Mb of system ram which is not accessible to the x86 code.
The word cache refers to several different things. Just like a screw can be in the dashboard of your car, on the rear wheel, or in the motor.
So, you would still need a crusoe to crusoe translator to resolve dependancies and do OOO.
No. You need a compiler. Some things are hard to do. For example:
*p = a * *b; *q = a * *b + 1;
here a normal compiler will have to assume that a and *b can change by the assignment to *p. It therefore will have to reevaluate the a * *b expression for the assignment to *q. However, on a TM CPU, you can ask the CPU to raise an exception on the case that that *p happens to hit on a or *b!
There are several different areas where a compiler would benefit just as much as code-morpher does...
But the power of the Transmeta CPUs is that they have an efficient bytecode that they pretend to interpret which in fact they compile, optimize and cache. That bytecode happens to be industry standard x86 code.... They can change the internal architecture without having to bother with a large installed base of programs. There is just one 64k instruction program that really matters: The code morpher.
These instructions on the Crusoe, are in almost all certainty RISC rather than CISC, since it's easier (drawing on intuition) to decompose CISC instructions into sequences of RISC instructions.
Some of the boundaries are slippery. For example: Is an intel 960 a RISC CPU?
Anyway, VLIW is NOT RISC. In RISC you have small simple instructions, in VLIW you have large instructions.
The trick to getting a VLIW fast is to get all "slots" filled. Transmeta seems to have a dynamic optimization. They profile the code, and those parts that get executed often, are optimized.
So, in effect, they translate the X86 instructions to a set of simple (RISC-like) instructions that map say on the left 20% of the VLIW. However, if the code is executed more often, the code goes into the optimizer, and a higher utilization of the VLIW is achieved. If they do this very well, that utilization will reach around 80% after optmization.
Anyhow, while Crusoe is not 'just a RISC chip', it's native instruction set must be simpler than x86 CISC, since you can't reasonably go in the opposite direction.
Nope. I don't think your estimate of "reasonable" is invalid. As we all know, transmeta doesn't stay within the "reasoable" boundaries. 1W for a 700MHz processor is NOT reasonable. That's outrageous.
[I just read the crusoetechwp.pdf file from the transmeta site (after writing the above!) and they show examples of the inital 20% translation and one where the eventual fill-level approaches 80%.]
Now that the decryption has been discovered to be pretty pathetic to begin with, the companies just need to roll with it and not worry.
The encryption was pathetic. This had two causes. One: They made a mistake and used a pretty bad encryption algorithm such that 14 out of 40 key bits are "given away" too easily. That leaves only 26 key bits. And 64M of ram for a big table is entirely feasible. And iterating over 64M keys is also easily done.
However, all this wouldn't have mattered if they had used 128 bit keys. In that case a serious flaw in the encryption: giving out one third of the bits of the key would've reduced the 128bit key to an 85 bit one. Try allocating that multi billion terabyte array in memory then. Try iterating over the only 3*10^25 remaining keys. Yeah right.
This all comes down to that the United States Government forced them to use unsafe 40bit keys. They should sue the United States Government for damages. And actually, those are actually the only guys with enough money to pay up the losses....
Xing's dvd player was reverse-engineered, which was against Xing's EULA, which makes any information obtained therein illegal to use
In Europe under whose jurisdiction the reverse engineering happened, the law stating that reverse engineering is allowed for compatibility reasons, overrides all EULAs that the user may have signed.
Also, IIRC, Xing's dvd player did not encrypt Xing's key, which violated the contract between Xing and the MPAA.
Let me start by saying that it sounds as if you know what you are talking about.
Further, there is ample case law that states that only the OWNER of the copy has an archival right. Re-read your license to discover that you are merely the bailee of the vendor's copy, and you are SOL.
So for about 20 years now, every software program you've bought has only been licenced to you. And some of the licences allow you to make one (1) archival copy. And you think that the judge is going to decide that this simple trick on the part of the software vendors is going to overturn a previous decision that you're allowed to make normal archival copies, depending on how your backup system works?
Suppose I have a large tape drive and make a level-0 (full backup) every day. I keep the tapes for two weeks, and then I recycle all tapes except the sunday one. Once they get older than 6 months, I recycle every one except the first-sunday-of-the-month one. Am I held to prevent the backup of all the licenced products on my backups? That's outrageous! I chose to buy that large tape drive to have the comfort of being able to insert one recovery boot flop, and the single tape that I want restored.
You can make backup copies, however your backup policy is formulated. This holds even if the licence says otherwise, provided of course that you don't use this to restore the software onto multiple other computers without a licence. However, it is pretty likely that the software companies know better than to sue someone over making full backups every day or week. So this probably hasn't been tested in court.
Now THIS is the reason that contracts/licences always say that if one clause is held unenforceable, all others remain in effect. This one clause ("No backups" or "Just one backup") will be ruled "unenforceable" when its time comes. They don't want the whole licence to fall into the water.
DeCSS is just an example... if they win, you can bet they'll go after css-auth and everything similar...
But in the case of css-auth, it is suddenly pretty clear that the intent is not to copy DVDs, but to play them under Linux. So at least initially, they can't make a case for a judge that "css-auth" is a bad thing.
So, they do have a much better case if they go after deCSS instead of css-auth.
You forgot to mention that at 8 years old you'll understand less than half of the jokes. But most of them will still be fun.
Reread the books when you're 20 or so, and you'll have double the fun!
Nobody, repeat NOBODY "gets" more than 80 or 90% of the jokes in a Terry Pratchett book. You really need a whole army of people to find them all. I once read a TP book, just after reading a Dawkins book. Only then do you understand that TP had just been reading that same Dawkins book just before writing his book. This goes COMPLETELY past you if you haven't read the other book less than a few months before....
Everybody seems to express disbelief that this will work.
If they are right, in their "assertion" that the spams are coming through open newsservers of their clients, then their measure will certainly work very very well!
However, I do doubt their assertion. But who am I to do that?
Anyway, I'd give them the benefit of doubt, and allow them to prove that they are right. Wouldn't it be nice if they were right, and their assertion would be true?
The weakest part of the case is that it all depends on trade secrets.
and XING is the one who spilled the beans. Trade secrets are just that. You keep them secret.
For example, I don't want you to know how much I'm charging for the new driver for the Specialix SX+ multiport card. So, I'm not telling.
Now if a guy named "Linus" needs to know, and I'd want to tell him, but I don't want him to tell everyone else, I'd make a contract stating that I'm going to provide the party "Linus" with information that is considered a trade secret and that this information is only provided that he won't tell anybody else.
Now if Linus DOES put this info on his website, the contract states what happens next. Usually the terms "irrepairable damages" occur in those sections of the contract. Sometimes a "fine" is named. Sometimes not.
Rumour has it, that the fine in the DVD contract is one million dollars. That's not much for Microsoft, but quite a lot for a startup. Damages are really much more in this case.
The information avaliable is legal, it was not stolen and its not patented. Nobody that I am awear of is being prosecuted for copyright theft involoving the use of the DeCSS program.
There is ONE party who has broken a law/contract. That's XING. Their contract with the DVD association states that they will keep the algorithm and keys secret, even though they are allowed to distribute a software-only version of the DVD decryption algorithm.
XING broke that contract.
The law in the EU states that even if you signed a contract not to reverse engineer a program, that condition in the contract is invalid. You can legally ignore that clause! Thus the reverse engineering was legal. Of course, the distribution thereafter of legally obtained information cannot be prosecuted.
Whatever XING did or didn't do to make it harder, that's their problem: they promised the DVD club that it wouldn't be done.
My guess is that XING has already had a serious talk with the DVD club, and, for example, their fate may depend on the results of this court case. e.g. if the court case succeeds, XING may get off with a reasonable fine. However, as things stand I expect they will be fined an unreasonably high fine, and go bankrupt. (but then again, I don't know their financial status... )
My site was up on the rollover, and according to the logs, nothing out of the ordinary happened.
To those that pulled plugs: The worst that would've happened is that you ended up "offline" for a period of time .
If you didn't want anybody to watch the system at the rollover, you should've sent everybody home. If, now 1/1/00, you're sending someone in to reboot the system, you could've done that anyway. However, if you'd left your system running he'd have gone for nothing: just to see if everything was ok. If you're satisfied with the system going back online on 03/01/00, fine. It might have survived.
Now the only thing that can be worse than just outage is material damage. A power dip won't damage much equipment. A power "jump" might. But how likely is that? That costs SERIOUS energy, and where is that going to come from. And keep in mind that power companies have lived with parts getting disconnected on short notice before.
"cal 9 1752" is THE way to check if your "cal" program gets the difficult things right. The correct answer is the one you quoted.
The "bug" was in the calendar, which had 365.2500 days/year instead of the 365.2425 that we use now. A guy named Gregorius (or something like that) in the 1750-ies figured it out and they decided to drop 11 days out of september 1752....
Riots broke out because people felt robbed of 11 days of their life: They thought they were destined to die on say december fifth, and that this way they'd live 11 days shorter.....
Different countries dropped the 11 (or 13 by the time everyone caught up) days at different times. Russia, I believe only joined the rest of the world after the october revolution (1917).
So actually, calendars between september 1752 and 1917 need a country-of-validity. The switchover in september 1752 was probably "vatican".
Hmm. I don't think so. But still no reason to stockpile though. If there is something going wrong, it's going to be one type of shops say "A&P" being "down" (in the computer sense) in the next millennium. That however doesn't mean you can't get food, even if you normally buy all your food at A&P.
Slashdot Mirror
It's much easier to find one copy out of ten linke than it is to find one copy out of nothing at all.
Users have to find ONE copy. Then they stop looking and have fun with the software.
MPAA wants to find ALL copies. They want to name everyone with deCSS on their homepage in the lawsuit.
However, the intention is that the judge will laugh them out of court if their list contains 50% or even 10% "legit" deCSS copies. Moreover, I'd recommend everyone to switch to publishing the other deCSS the moment his name shows up on an MPAA list....
Roger.
Why the least brain power, and not the most?
Simple. The sensors mistake muscle activity for brain activity. And muscle activity shows up with say 10 times more amplitude than brainactivity.
So to make the machine give a high reading, you just have to move about.
Roger.
And if you do the math....
.03 * 16 + .03 *13.5 + .10 * 11.5 + .31 * 9.5 +.21 * 8 + .15 * n
the result is that the average slashdotter works between 6.6 and 7.7 hours a day, depending on how you estimate "less than 8 hours". Put it at 4, and the average slashdotter works 7.2 hours a day....
Roger.
I tried installing a new machine with the GUI (I swear I didn't know I was going to get a GUI. I just popped in the first CD like I always do...).
It omits a few questions that I find essential. I fogot which. One thing I do remember is that it tells me that the root password that I've been using for way too long now has illegal characters. Jee, and I've been using those illegal characters all those years, and NOW it tells me!
Booted the other CD and installed the system.
Roger.
I just saw the show. It's fascinating.
They have been trying to get things to walk, just by letting "trial and error" evolve. This is an advancement.
Now they are a bit further by having a 14-motor ape swing from rung to rung on a ladder. Some people may be afraid for AIs, but then they should just protest. I mean, this is just another step on the big ladder.
Roger.
It seems to me that there are some tricky special relativity thingies that these guys are forgetting.
Roger.
The first time a chunk of code is run, there's a performance hit as it's translated to the native instruction set for the particular chip on which it's running.
Nope. the first few times it is executed it takes a performance hit as if it is interpreted. As that's exactly what the transmeta CPU is doing.
Only after a few times the code is going into the compiler. And after a few hundred (my guess) times the optimizer is kicked into action....
Roger.
If you think of a cache like a window on a portion of the code, and then you can imagine that only a small portion of the code (less than 1K) would be kept in the cache. The TM3120 has 108KB of cache, but the TM5400 has 400KB cache, which sounds like a good amount.
Crusoe uses about seven to fifteen Megabytes of code-cache. The 1M codemorpher is copied from ROM to sdram, and then the system uses 8 to 16Mb of
system ram which is not accessible to the x86 code.
The word cache refers to several different things. Just like a screw can be in the dashboard of your car, on the rear wheel, or in the motor.
Roger.
(Elsewhere the paper implies that this ROM software is or can be copied to RAM at boot up for faster execution.)
It is. The ROM is serial and about 8Mbit. So the CPU will spend a significant number of milliseconds of emptying the serial ROM into the SDRAM.
Roger.
So, you would still need a crusoe to crusoe translator to resolve dependancies and do OOO.
No. You need a compiler. Some things are hard to do. For example:
*p = a * *b;
*q = a * *b + 1;
here a normal compiler will have to assume that a and *b can change by the assignment to *p. It therefore will have to reevaluate the a * *b expression for the assignment to *q. However, on a TM CPU, you can ask the CPU to raise an exception on the case that that *p happens to hit on a or *b!
There are several different areas where a compiler would benefit just as much as code-morpher does...
But the power of the Transmeta CPUs is that they have an efficient bytecode that they pretend to interpret which in fact they compile, optimize and cache. That bytecode happens to be industry standard x86 code.... They can change the internal architecture without having to bother with a large installed base of programs. There is just one 64k instruction program that really matters: The code morpher.
Roger.
These instructions on the Crusoe, are in almost all certainty RISC rather than CISC, since it's easier (drawing on intuition) to decompose CISC instructions into sequences of RISC instructions.
Some of the boundaries are slippery. For example: Is an intel 960 a RISC CPU?
Anyway, VLIW is NOT RISC. In RISC you have small simple instructions, in VLIW you have large instructions.
The trick to getting a VLIW fast is to get all "slots" filled. Transmeta seems to have a dynamic optimization. They profile the code, and those parts that get executed often, are optimized.
So, in effect, they translate the X86 instructions to a set of simple (RISC-like) instructions that map say on the left 20% of the VLIW. However, if the code is executed more often, the code goes into the optimizer, and a higher utilization of the VLIW is achieved. If they do this very well, that utilization will reach around 80% after optmization.
Anyhow, while Crusoe is not 'just a RISC chip', it's native instruction set must be simpler than x86 CISC, since you can't reasonably go in the opposite direction.
Nope. I don't think your estimate of "reasonable" is invalid. As we all know, transmeta doesn't stay within the "reasoable" boundaries. 1W for a 700MHz processor is NOT reasonable. That's outrageous.
[I just read the crusoetechwp.pdf file from the transmeta site (after writing the above!) and they show examples of the inital 20% translation and one where the eventual fill-level approaches 80%.]
Roger.
Now that the decryption has been discovered to be pretty pathetic to begin with, the companies just need to roll with it and not worry.
The encryption was pathetic. This had two causes. One: They made a mistake and used a pretty bad encryption algorithm such that 14 out of 40 key bits are "given away" too easily. That leaves only 26 key bits. And 64M of ram for a big table is entirely feasible. And iterating over 64M keys is also easily done.
However, all this wouldn't have mattered if they had used 128 bit keys. In that case a serious flaw in the encryption: giving out one third of the bits of the key would've reduced the 128bit key to an 85 bit one. Try allocating that multi billion terabyte array in memory then. Try iterating over the only 3*10^25 remaining keys. Yeah right.
This all comes down to that the United States Government forced them to use unsafe 40bit keys. They should sue the United States Government for damages. And actually, those are actually the only guys with enough money to pay up the losses....
Roger.
Xing's dvd player was reverse-engineered, which was against Xing's EULA, which makes any information obtained therein illegal to use
In Europe under whose jurisdiction the reverse engineering happened, the law stating that reverse engineering is allowed for compatibility reasons, overrides all EULAs that the user may have signed.
Also, IIRC, Xing's dvd player did not encrypt Xing's key, which violated the contract between Xing and the MPAA.
Jep, So they should be sueing XING.
Roger.
Let me start by saying that it sounds as if you know what you are talking about.
Further, there is ample case law that states that only the OWNER of the copy has an archival right. Re-read your license to discover that you are merely the bailee of the vendor's copy, and you are SOL.
So for about 20 years now, every software program you've bought has only been licenced to you. And some of the licences allow you to make one (1) archival copy. And you think that the judge is going to decide that this simple trick on the part of the software vendors is going to overturn a previous decision that you're allowed to make normal archival copies, depending on how your backup system works?
Suppose I have a large tape drive and make a level-0 (full backup) every day. I keep the tapes for two weeks, and then I recycle all tapes except the sunday one. Once they get older than 6 months, I recycle every one except the first-sunday-of-the-month one. Am I held to prevent the backup of all the licenced products on my backups? That's outrageous! I chose to buy that large tape drive to have the comfort of being able to insert one recovery boot flop, and the single tape that I want restored.
You can make backup copies, however your backup policy is formulated. This holds even if the licence says otherwise, provided of course that you don't use this to restore the software onto multiple other computers without a licence. However, it is pretty likely that the software companies know better than to sue someone over making full backups every day or week. So this probably hasn't been tested in court.
Now THIS is the reason that contracts/licences always say that if one clause is held unenforceable, all others remain in effect. This one clause ("No backups" or "Just one backup") will be ruled "unenforceable" when its time comes. They don't want the whole licence to fall into the water.
Roger.
DeCSS is just an example... if they win, you can bet they'll go after css-auth and everything similar...
But in the case of css-auth, it is suddenly pretty clear that the intent is not to copy DVDs, but to play them under Linux. So at least initially, they can't make a case for a judge that "css-auth" is a bad thing.
So, they do have a much better case if they go after deCSS instead of css-auth.
Roger.
> The Discworld series by Terry Pratchett.
You forgot to mention that at 8 years old you'll understand less than half of the jokes. But most of them will still be fun.
Reread the books when you're 20 or so, and you'll have double the fun!
Nobody, repeat NOBODY "gets" more than 80 or 90% of the jokes in a Terry Pratchett book. You really need a whole army of people to find them all. I once read a TP book, just after reading a Dawkins book. Only then do you understand that TP had just been reading that same Dawkins book just before writing his book. This goes COMPLETELY past you if you haven't read the other book less than a few months before....
Roger.
Everybody seems to express disbelief that this will work.
If they are right, in their "assertion" that the spams are coming through open newsservers of their clients, then their measure will certainly work very very well!
However, I do doubt their assertion. But who am I to do that?
Anyway, I'd give them the benefit of doubt, and allow them to prove that they are right. Wouldn't it be nice if they were right, and their assertion would be true?
Roger.
The weakest part of the case is that it all depends on trade secrets.
and XING is the one who spilled the beans. Trade secrets are just that. You keep them secret.
For example, I don't want you to know how much I'm charging for the new driver for the Specialix SX+ multiport card. So, I'm not telling.
Now if a guy named "Linus" needs to know, and I'd want to tell him, but I don't want him to tell everyone else, I'd make a contract stating that I'm going to provide the party "Linus" with information that is considered a trade secret and that this information is only provided that he won't tell anybody else.
Now if Linus DOES put this info on his website, the contract states what happens next. Usually the terms "irrepairable damages" occur in those sections of the contract. Sometimes a "fine" is named. Sometimes not.
Rumour has it, that the fine in the DVD contract is one million dollars. That's not much for Microsoft, but quite a lot for a startup. Damages are really much more in this case.
IANAL.
Roger.
The information avaliable is legal, it was not stolen and its not patented. Nobody that I am awear of is being prosecuted for copyright theft involoving the use of the DeCSS program.
There is ONE party who has broken a law/contract. That's XING. Their contract with the DVD association states that they will keep the algorithm and keys secret, even though they are allowed to distribute a software-only version of the DVD decryption algorithm.
XING broke that contract.
The law in the EU states that even if you signed a contract not to reverse engineer a program, that condition in the contract is invalid. You can legally ignore that clause! Thus the reverse engineering was legal. Of course, the distribution thereafter of legally obtained information cannot be prosecuted.
Whatever XING did or didn't do to make it harder, that's their problem: they promised the DVD club that it wouldn't be done.
My guess is that XING has already had a serious talk with the DVD club, and, for example, their fate may depend on the results of this court case. e.g. if the court case succeeds, XING may get off with a reasonable fine. However, as things stand I expect they will be fined an unreasonably high fine, and go bankrupt. (but then again, I don't know their financial status... )
Roger.
After about 3:10 EST, you can go here and see the result.
Nice MPEG movie. However, my mpeg_play and xanim seem to get something related to "interlacing" wrong. What do you guys use for playing mpeg movies?
Roger.
My site was up on the rollover, and according to the logs, nothing out of the ordinary happened.
To those that pulled plugs: The worst that would've happened is that you ended up "offline" for a period of time .
If you didn't want anybody to watch the system at the rollover, you should've sent everybody home. If, now 1/1/00, you're sending someone in to reboot the system, you could've done that anyway. However, if you'd left your system running he'd have gone for nothing: just to see if everything was ok. If you're satisfied with the system going back online on 03/01/00, fine. It might have survived.
Now the only thing that can be worse than just outage is material damage. A power dip won't damage much equipment. A power "jump" might. But how likely is that? That costs SERIOUS energy, and where is that going to come from. And keep in mind that power companies have lived with parts getting disconnected on short notice before.
Roger.
Roger.
Ok. Cal is English. Gregory was two centuries earlier.
"cal 9 1752" is THE way to check if your "cal" program gets the difficult things right. The correct answer is the one you quoted.
The "bug" was in the calendar, which had 365.2500 days/year instead of the 365.2425 that we use now. A guy named Gregorius (or something like that) in the 1750-ies figured it out and they decided to drop 11 days out of september 1752....
Riots broke out because people felt robbed of 11 days of their life: They thought they were destined to die on say december fifth, and that this way they'd live 11 days shorter.....
Different countries dropped the 11 (or 13 by the time everyone caught up) days at different times. Russia, I believe only joined the rest of the world after the october revolution (1917).
So actually, calendars between september 1752 and 1917 need a country-of-validity. The switchover in september 1752 was probably "vatican".
Roger.
EVERYTHING IS GOING TO WORK.
Hmm. I don't think so. But still no reason to stockpile though. If there is something going wrong, it's going to be one type of shops say "A&P" being "down" (in the computer sense) in the next millennium. That however doesn't mean you can't get food, even if you normally buy all your food at A&P.
9600 9599...
Roger.
I see that they usually mention full URLs in the injunction order. How about everyone changing a few letters of the URL?
Can't hurt can it?
Roger.