Just because the "public" doesn't use it doesn't mean "corporations" don't. I hate to break it to you but you only have access to probably 50% of what's out on the internet at large...
You mean that there's porn out there only for corporations?!
comment about computer security. This comment is incorrect. Here's why it's incorrect:
( ) You have no idea how system security works (x) You assume that popularity is inversely proportional to security ( ) You make improper use of technical language (x) You assume that part of the problem is the entire problem (x) You fail to account for different security models.
Specifically, your comment fails to consider that:
( ) Security flaws can be exploited in an automated manner (x) Not all bugs are security flaws ( ) Security flaws can be exploited manually ( ) Legions of script kiddies use point-and-drool tools ( ) Dedicated black-hats can cause damage using home-designed tools (x) Privilege separation prevents many problems ( ) Some security flaws are strictly theoretical (x) Different systems are inherently more or less vulnerable to exploits ( ) Security flaws can be independently discovered ( ) Security flaw discovery and exploit does not require source code (x) Not all security flaws are of the same severity (x) Running as root is almost always a problem, no matter the system ( ) Not all viruses are transmitted by e-mail ( ) Not all viruses are self-propagating ( ) Not all security flaws are buffer overruns ( ) Stupid people do stupid things
and the following general objections may also apply:
( ) Full disclosure completely informs affected system administrators ( ) Exploit code has legitimate uses (x) Security is by design, not accident ( ) Security isn't magic, and thinking of it that way is harmful ( ) Hackers/crackers aren't evil magicians who can get around anything ( ) Security starts with the user ( ) Why should we trust the government? ( ) Why should we trust you?
Furthermore, this is what I think about you:
(x) Nice idea, but it's been said before and that doesn't make it any more true. ( ) That's an incredibly stupid idea, and you're stupid for suggesting it. ( ) You're a moron, and I'm surprised you have enough brain cells to continue breathing.
Well, it turns out that I was wrong. The original article was updated to include comments from Microsoft, which say that they will not charge for new versions of games. Given that, the difference between emulating the original and distributing recompiled executables is academic for anyone with an internet connection.
Large outcry from that, which wouldn't be past Microsoft. But I'd still say it'd be more likely that they would put these games on the silver network which is officially for game content downloads, extra game content purchases, and little stuff like scoreboards.
This seems reasonable enough, it doesn't quite make sense to me (see below). More interestingly, what do they intend to do about third-party games where Microsoft doesn't have the right to recompile/redistribute the executable?
Especially for semi-popular games that are no longer being sold new, it wouldn't be in Developer Z's interest to recompile, retest, and rerelease the executable.
In short, I feel that there's no way that Microsoft will be able to offer this level of "backwards compatibility" for free. My guess for the price-point of a downloaded compatibility upgrade will be somewhere in the neighborhood of $5 (waived for Halo, and maybe a couple other extremely popular games). It would make sense for XBox Live Gold Super Duper Ultra Level to offer free backwards compatibility, but at $5/game, I'm not sure if Microsoft's accountants could justify including it in the monthly cost.
Although I personally feel that the ESRB system is relinquishing moral judgements into the hands of a distant third party (much like movie ratings), this system does seem to be a reasonable idea.
It is also a very compelling argument in favor of multi-user systems.
The big catch with this idea, though, is that this is restricted to only commercial titles, and ESRB-rated titles at that. While TFA says that this can also be used to control "computer use," it likely won't be able to distinguish web/Java games from web research for the science fair.
I speculate that the software knows executable names for commercially released games, and it can then cross-reference these against the ESRB database. With this in mind, smaller catches are that this software will require regular updates, and the ESRB rating system itself is quite coarse: look at the dearth of adults-only games.
She would collect up all of our work, mix them up, and hand them back to the class, and give us red pens. If anyone got their own page, we had to trade with whoever was next to us. She'd read out the answers, and we'd mark them, and sign our name as the grader. She'd collect them, and do spot checks to make sure we didn't mess up. [someone could also complain if they were unfairly marked].
Not as uncommon nowadays as you think -- this happened to me fairly often in school also, and I went through normal public education. The real killer of this otherwise good idea is records-privacy laws, which consider grades to be educational records; now it could get teachers in trouble if they let any other students know your grade.
To -actually answer the parent's question,- Canadian citizens do not get fingerprinted at land border crossings with the US. Unlike most other first-world countries, Canada isn't a "Visa Waiver" country -- technically speaking, Canadian citizens don't even need passports to enter the US, although they need proof of citizenship and photo-ID.
The same holds true for US citizens visiting Canada.
A hundred years ago, if you thought it might be fun to depict sex with children, people wouldn't have done anything quite as nice as to take away your working materials and send you away for a few months or years. You probably wouldn't have even made it to someplace quite as comfy as a court of law. There probably would have been a mob of people at your front door knocking each other over to be the one that got to string you up by your genitals in the town square.
If you're a peasant, maybe. As a noble, or gentleman, these things were depicted somewhat freely, albeit with a bit of shame attached. I quote:
Book Description
Lewd, bawdy, and sensual, this cult classic is a collection of Victorian erotica that circulated in an underground magazine known as The Pearl from July 1879 to December 1880. Now dusted off and totally uncensored, the journal of voluptuous reading that titillated the eminent Victorians is reprinted in its entirety.
Re:You can find it googling, but here it is anyway
on
Is DOS Gaming Dead?
·
· Score: 1
For example, VDM Sound + 2k console will play Master of Orion just fine, but say Star Control II has an issue with the graphics rendering painfully slow. It's hit-or-miss, but it works surprisingly well.
You probably shouldn't be using a dos-box to play Star Control II when there's Windows/Linux/OSX versions available.
The article was unclear of the exact law involved here. Searching through the Florida statutes gave me this:
202.15 Special rule for users of substitute communications systems.--Any person who purchases, installs, rents, or leases a substitute communications system must register with the department and pay the taxes imposed or administered pursuant to s. 202.12 annually pursuant to rules prescribed by the department.
and
202.11 Definitions.--As used in this chapter: ...
(16) "Substitute communications system" means any telephone system, or other system capable of providing communications services, which a person purchases, installs, rents, or leases for his or her own use to provide himself or herself with services used as a substitute for any switched service or dedicated facility by which a dealer of communications services provides a communication path.
Section 12 says that the tax rate is 6.8% of the sales price, applied yearly.
'The contents of your Gmail account also are stored and maintained on Google servers in order to provide the service. Indeed, residual copies of email may remain on our systems, even after you have deleted them from your mailbox or after the termination of your account.'
Has it occured to anyone that keeping residual copies of e-mails, possibly even for a time after the account is deleted, is necessary, even required, to back up the data? Google's privacy policy is unique in that it tells you what they do with your information, rather than (only) what they'll let other people do with your information.
The other large privacy concern here, that of ad-delivery, requires Google to scan e-mails for keywords. Yep. Big woop. They do that every time you search, you know -- and in the e-mails, their privacy policy specifically says that no humans will read it without specific permission to solve e.g. technical problems.
Well, to start with, the Gentoo page seems to have hard page width formatting so it spills off my screen, forcing the browser into scroller mode. And I have a big Mozilla window open on a 1280 width display.
The LDP page cleanly wrapped the text and everything is there comfortably for me to read. I can even size down the browser window, in case, for example, I am actually trying to do something that the doc pertains to.
Not strictly true. The Gentoo page and LDP page have hard widths in exactly the same way.
The Gentoo page will not shrink beyond the size of the longest code listing (on that page, 36), nor will the LDP page shrink beyond its longest blockquote (the ping). In both cases, this is to prevent wrapping on quoted text that, more often than not, needs linebreaks preserved.
You notice the Gentoo width because that particular example comes from a fdisk session, and the Gentoo people included a couple long lines of comments (the red text).
Each system I've used, ever since my first, has used some components (to a greater or lesser degree) from the one before it.
The only part original to the lot is my keyboard... it was original with a 386 (AMD brand, I believe) computer, with a whopping 8MB of RAM. The keyboard is mighty unusual, also -- it's the only one I've ever seen with diagonal arrow keys [barely functional], an on-keyboard "turbo" button for changing keyboard repeat rate.
Other features include an asterisk between the right control and alt keys, an unlabeled key (that seems to be a backslash) between left control and alt keys, and a giant L-shape enter key [with full-length backspace above it].
No date on the keyboard itself, although identification on the bottom says "Focus Electronic Co, Ltd." with "Made in Taiwan."
It's worth noting here that the benchmarks were all run on files >= 1GB, if I'm reading the table correctly; this stresses the raw IO of the system, and doesn't really take into account the differences in tree-structure between the filesystems.
As for complaints about Reiser's performance -- last I heard, it was more optimized for many small files -- precisely the domain that this thing didn't test.
... and is still partially owned by Chuck Hagel, Republican Senator from Nebraska - who, surprisingly, won unprecedented victories in his state against an incumbent Democrat governor, winning by the largest landslide ever and taking the majority among demographics that had never voted Republican in the past.
Actually, this is one of the times I'd be LEAST likely to suspect election fraud. You seem to forget that any election more attention-getting than local school board is going to be continuously monitored by opinion polling.
If, as you suggest, the landslide was fraudulent, then the election results would have no relation to either the pre-election polls or the exit polling. This would attract an awful lot of attention in the media, and I believe that any fraud on the scale that you suggest would at least be openly accused.
The only place, in my mind, that election fraud would be useful beyond the threat of detection would be in extremely close races -- those that no one has any idea who will win. In those cases, than altering the votes by 1% would still be within the margin of error on even the exit polling, and so wouldn't be immediately suspicious.
Does violating the GPL with one product (the kernel) violate it under another (samba)?
No, because the programs are seperately licensed. However, you could make the argument that SCO's public comments are a rejection of the GPL, probably through their comment that it was illegal. Since I doubt you can agree to a contract that you know is illegal, SCO did not agree to the terms of the GPL and is thus not allowed to distribute Samba.
There are two cases where you win by sticking. Where you are shown #2, and where you are shown #3.
Yes, but the table you draw lists those cases equally with your initial choice, but that is false. Monly's revelation is a seperate event. You're giving what is simply a subchoice the same weight as the main choices.
Let's assume, for the same of simplicity, that you always pick door number one. Since the first choice is completely random, this doesn't affect the final odds -- and it reduces the number of cases we need to look at fromo 15 to 5.
Door one is the correct choice, and you chose it right off the bat. (33%)
Case 1a -- Monty shows you what's behind Door 2, and switching (to Door 3) will result in a loss for you. (50% probability of this Monty revalation)
Case 1b -- Monty shows you what's behind Door 3, and switching (to Door 2) will result in a loss for you. (50%)
Door two is the correct choice, and you pick a losing door. (33%)
Case 2a -- Monty shows you what's behind Door 3, and switching (to Door 2) will result in a win for you. (100% probability of monty showing you Door 3)
Door three is the correct choice, and you pick a losing door. (33%)
Case 3a -- Monty shows you what's behind Door 2, and switching (to Door 3) will result in a win for you. (100% probability of monty showing Door 3)
Now, to determine your overall probability of winning with a police of (not?) switching, we multiply the probability of each of the main cases by its probability of generating a win for you.
Not Switch: Probability of win == 33% * (50%*100% + 50%*100%) [case 1, either of monty's choices means a win for you] + 33% * (0%) [case 2] + 33% * (0%) = 33% chance of winning.
Switch: Probability of win == 33% * (50% * 0% + 50% * 0%) [case 1, either door means a loss for you] + 33% * 100% [case 2] + 33% * 100% [case 3] = 66% chance of winning.
Therefore, in the Monty Hall problem, a policy of switching when offered the choice results in a 66% probability of win for you, while not switching results in a 33% probability for a win -- meaning that it is in your benefit to switch.
I'm curious.. how can they distribute GPL code under an NDA. Linux code is copyright the author, and the only right SCO has to redistribute it is under the GPL.
I'm guessing that it would fall under the 'short quotes' exemption of Fair Use. They're certainly not distributing substantial (or even functional) portions, and their distribution for the slection is trying to show similarity.
That post is from March, 1988 about rumors of the imminent death of ARPANET. There was also a later post ('91) that specifically mentions "death of the net".
That post is from March, 1988 about rumors of the imminent death of ARPANET. There was also a later post ('91) that specifically mentions "death of the net".
If you accept that the Net was generally considered equivalent to USENET, much like now we consider it equivalent to e-mail + the intraweb thingy, then it was supposed to die in Apr. 1983.
Re:Private key != source
on
Linus on DRM
·
· Score: 1
# Code is something that a machine can execute. Code defines a process, a series of events, a set of conditions and outcomes for those conditions. Code may act without any external data, or may make decisions based on external data. Code may output data (an image editor), or it may not output data (a temperature control algorithm).
# Data cannot be executed. By itself, it can do nothing; it must be acted upon by code. Data can be unique, so that a specific piece of code acting on one set of data will behave different when acting on another set of data.
My definitions:
Data is everything that changes, or potentially can change, between executions of the software. In the case of the Linux kernel, the first data that comes to mind is everything changeable in the/proc interfaces, the mount list, and the kernel command line. Data is modifyable at load time, if not at run time.
Code is everything that either cannot change or is part of the executable. I make this either-or here partly because there is such a thing as self-modifying code (although that is not the topic of discussion). Everything used (as data, natch) by the compiler to create the executable is code. A list of instructions read at runtime for execution counts as code (instead of parameters) if the instructions are in a turing-complete language.
In your example of a test suite, the testing list as I'm understanding its function would be data.
In the original example of the signature, the signature itself would be code (by virtue of being part of the executable, or inseparable from it) and the private key would fall under the GPL, probably as part of the 'preferred form of modifications' clause.
Re:Private key != source
on
Linus on DRM
·
· Score: 1
# Plaintiff: "Since I cannot compile a working binary from the source you sent me, you did not release the source code. You are in violation of GPL. You must release the source, replace the OS with a non-GPL OS, or refund our money."
# Defendant: "Yes, I did release the source code. And it works. I just didn't give you my secret key."
# Plaintiff: "No, you did not release the source. Since I cannot build exactly the same binary that you sent me, part of the source must be missing."
# Defendant: "Yes, I did release the source. The binaries you generated function exactly the same as the ones I gave you. Part of their function is to verify that they were created using the same secret key as the server they are trying to connect to."
Au contraire; the program does not function exactly the same as the ones you gave the purchaser. By your own admission, The Fritz chip won't let [his kernel] boot, and So the binaries cannot verify against my server.
Functioning identically implies that it gives the same outputs for identical inputs. Refusing to boot on FritzPlatform2000 is most certainly a different output, and failing key/response is also a different output. You could write either the notbooting xor the key/response as a function of the hardware, but for it to fail in both cases means that the compiled kernel functions differently from the binary one distributed with FritzPlatform.
On the matter, the GPL says:
For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable.
Since your signature-generator is required for the very installation of the kernel [on the FritzPlatform], let alone iidentical behavior, I cosider it a vital module and thus contained in the GPL license when distributed with the binary kernel.
Linus, et al's signatures are quite different; they are decidedly not required for proper functioning of the kernel on any given hardware platform. In addition, further distancing his signatures (and thus private key) from redistribution requirements in the GPL, his signature is not distributed as part of the (source or binary) kernel. From ftp.kernel.org:
File: linux-2.5.68.tar.bz2.sign 1 KB 4/20/2003 3:02:00 AM
This is an independant file not required for the download, compilation, installation, or use of the kernel. It is as much part of the kernel as the announcing press release is.
I feel that the question of whether signing keys are covered by the GPL is still very much open. I've obviously made my feelings clear on the issue, but it will be interesting to see how it pans out.
There seem to be some currently-popular misunderstandings about this article. This TarProxy is not intended to be running on outgoing SMTP servers -- it makes no sense to throttle clients that you're supposed to be monitoring anyway.
Instead, this is meant to be run on the incoming SMTP server, the one that receives the mail. It will only hurt the spammer if he's trying to send a bunch of spam to your domain, but every server running this can help.
Slashdot Mirror
Just because the "public" doesn't use it doesn't mean "corporations" don't. I hate to break it to you but you only have access to probably 50% of what's out on the internet at large...
You mean that there's porn out there only for corporations?!
Your post involves a
( ) technical (x) social ( ) moral
comment about computer security. This comment is incorrect. Here's why it's incorrect:
( ) You have no idea how system security works
(x) You assume that popularity is inversely proportional to security
( ) You make improper use of technical language
(x) You assume that part of the problem is the entire problem
(x) You fail to account for different security models.
Specifically, your comment fails to consider that:
( ) Security flaws can be exploited in an automated manner
(x) Not all bugs are security flaws
( ) Security flaws can be exploited manually
( ) Legions of script kiddies use point-and-drool tools
( ) Dedicated black-hats can cause damage using home-designed tools
(x) Privilege separation prevents many problems
( ) Some security flaws are strictly theoretical
(x) Different systems are inherently more or less vulnerable to exploits
( ) Security flaws can be independently discovered
( ) Security flaw discovery and exploit does not require source code
(x) Not all security flaws are of the same severity
(x) Running as root is almost always a problem, no matter the system
( ) Not all viruses are transmitted by e-mail
( ) Not all viruses are self-propagating
( ) Not all security flaws are buffer overruns
( ) Stupid people do stupid things
and the following general objections may also apply:
( ) Full disclosure completely informs affected system administrators
( ) Exploit code has legitimate uses
(x) Security is by design, not accident
( ) Security isn't magic, and thinking of it that way is harmful
( ) Hackers/crackers aren't evil magicians who can get around anything
( ) Security starts with the user
( ) Why should we trust the government?
( ) Why should we trust you?
Furthermore, this is what I think about you:
(x) Nice idea, but it's been said before and that doesn't make it any more true.
( ) That's an incredibly stupid idea, and you're stupid for suggesting it.
( ) You're a moron, and I'm surprised you have enough brain cells to continue breathing.
Well, it turns out that I was wrong. The original article was updated to include comments from Microsoft, which say that they will not charge for new versions of games. Given that, the difference between emulating the original and distributing recompiled executables is academic for anyone with an internet connection.
This seems reasonable enough, it doesn't quite make sense to me (see below). More interestingly, what do they intend to do about third-party games where Microsoft doesn't have the right to recompile/redistribute the executable?
Especially for semi-popular games that are no longer being sold new, it wouldn't be in Developer Z's interest to recompile, retest, and rerelease the executable.
In short, I feel that there's no way that Microsoft will be able to offer this level of "backwards compatibility" for free. My guess for the price-point of a downloaded compatibility upgrade will be somewhere in the neighborhood of $5 (waived for Halo, and maybe a couple other extremely popular games). It would make sense for XBox Live Gold Super Duper Ultra Level to offer free backwards compatibility, but at $5/game, I'm not sure if Microsoft's accountants could justify including it in the monthly cost.
It is also a very compelling argument in favor of multi-user systems.
The big catch with this idea, though, is that this is restricted to only commercial titles, and ESRB-rated titles at that. While TFA says that this can also be used to control "computer use," it likely won't be able to distinguish web/Java games from web research for the science fair.
I speculate that the software knows executable names for commercially released games, and it can then cross-reference these against the ESRB database. With this in mind, smaller catches are that this software will require regular updates, and the ESRB rating system itself is quite coarse: look at the dearth of adults-only games.
Not as uncommon nowadays as you think -- this happened to me fairly often in school also, and I went through normal public education. The real killer of this otherwise good idea is records-privacy laws, which consider grades to be educational records; now it could get teachers in trouble if they let any other students know your grade.
It works only because you're directly opening the SWF file. I believe that FlashBlock or its ilk only block/hide flash that's embedded in the page.
The same holds true for US citizens visiting Canada.
There's a typo in that number -- checking the page's figures, it should read 118 billion gallons. (1.18*10^11 gal)
If you're a peasant, maybe. As a noble, or gentleman, these things were depicted somewhat freely, albeit with a bit of shame attached. I quote:
You probably shouldn't be using a dos-box to play Star Control II when there's Windows/Linux/OSX versions available.
Section 12 says that the tax rate is 6.8% of the sales price, applied yearly.
Has it occured to anyone that keeping residual copies of e-mails, possibly even for a time after the account is deleted, is necessary, even required, to back up the data? Google's privacy policy is unique in that it tells you what they do with your information, rather than (only) what they'll let other people do with your information.
The other large privacy concern here, that of ad-delivery, requires Google to scan e-mails for keywords. Yep. Big woop. They do that every time you search, you know -- and in the e-mails, their privacy policy specifically says that no humans will read it without specific permission to solve e.g. technical problems.
Tin foil hats can go back in the closet, boys.
The LDP page cleanly wrapped the text and everything is there comfortably for me to read. I can even size down the browser window, in case, for example, I am actually trying to do something that the doc pertains to.
Not strictly true. The Gentoo page and LDP page have hard widths in exactly the same way.
The Gentoo page will not shrink beyond the size of the longest code listing (on that page, 36), nor will the LDP page shrink beyond its longest blockquote (the ping). In both cases, this is to prevent wrapping on quoted text that, more often than not, needs linebreaks preserved.
You notice the Gentoo width because that particular example comes from a fdisk session, and the Gentoo people included a couple long lines of comments (the red text).
The only part original to the lot is my keyboard... it was original with a 386 (AMD brand, I believe) computer, with a whopping 8MB of RAM. The keyboard is mighty unusual, also -- it's the only one I've ever seen with diagonal arrow keys [barely functional], an on-keyboard "turbo" button for changing keyboard repeat rate.
Other features include an asterisk between the right control and alt keys, an unlabeled key (that seems to be a backslash) between left control and alt keys, and a giant L-shape enter key [with full-length backspace above it].
No date on the keyboard itself, although identification on the bottom says "Focus Electronic Co, Ltd." with "Made in Taiwan."
As for complaints about Reiser's performance -- last I heard, it was more optimized for many small files -- precisely the domain that this thing didn't test.
Actually, this is one of the times I'd be LEAST likely to suspect election fraud. You seem to forget that any election more attention-getting than local school board is going to be continuously monitored by opinion polling.
If, as you suggest, the landslide was fraudulent, then the election results would have no relation to either the pre-election polls or the exit polling. This would attract an awful lot of attention in the media, and I believe that any fraud on the scale that you suggest would at least be openly accused.
The only place, in my mind, that election fraud would be useful beyond the threat of detection would be in extremely close races -- those that no one has any idea who will win. In those cases, than altering the votes by 1% would still be within the margin of error on even the exit polling, and so wouldn't be immediately suspicious.
No, because the programs are seperately licensed. However, you could make the argument that SCO's public comments are a rejection of the GPL, probably through their comment that it was illegal. Since I doubt you can agree to a contract that you know is illegal, SCO did not agree to the terms of the GPL and is thus not allowed to distribute Samba.
Yes, but the table you draw lists those cases equally with your initial choice, but that is false. Monly's revelation is a seperate event. You're giving what is simply a subchoice the same weight as the main choices.
Let's assume, for the same of simplicity, that you always pick door number one. Since the first choice is completely random, this doesn't affect the final odds -- and it reduces the number of cases we need to look at fromo 15 to 5.
Now, to determine your overall probability of winning with a police of (not?) switching, we multiply the probability of each of the main cases by its probability of generating a win for you.
Therefore, in the Monty Hall problem, a policy of switching when offered the choice results in a 66% probability of win for you, while not switching results in a 33% probability for a win -- meaning that it is in your benefit to switch.
I'm guessing that it would fall under the 'short quotes' exemption of Fair Use. They're certainly not distributing substantial (or even functional) portions, and their distribution for the slection is trying to show similarity.
Apologies, but that date is supposed to be '89.
That post is from March, 1988 about rumors of the imminent death of ARPANET. There was also a later post ('91) that specifically mentions "death of the net".
If you accept that the Net was generally considered equivalent to USENET, much like now we consider it equivalent to e-mail + the intraweb thingy, then it was supposed to die in Apr. 1983.
# Data cannot be executed. By itself, it can do nothing; it must be acted upon by code. Data can be unique, so that a specific piece of code acting on one set of data will behave different when acting on another set of data.
My definitions:
In your example of a test suite, the testing list as I'm understanding its function would be data.
In the original example of the signature, the signature itself would be code (by virtue of being part of the executable, or inseparable from it) and the private key would fall under the GPL, probably as part of the 'preferred form of modifications' clause.
# Defendant: "Yes, I did release the source code. And it works. I just didn't give you my secret key."
# Plaintiff: "No, you did not release the source. Since I cannot build exactly the same binary that you sent me, part of the source must be missing."
# Defendant: "Yes, I did release the source. The binaries you generated function exactly the same as the ones I gave you. Part of their function is to verify that they were created using the same secret key as the server they are trying to connect to."
Au contraire; the program does not function exactly the same as the ones you gave the purchaser. By your own admission, The Fritz chip won't let [his kernel] boot, and So the binaries cannot verify against my server.
Functioning identically implies that it gives the same outputs for identical inputs. Refusing to boot on FritzPlatform2000 is most certainly a different output, and failing key/response is also a different output. You could write either the notbooting xor the key/response as a function of the hardware, but for it to fail in both cases means that the compiled kernel functions differently from the binary one distributed with FritzPlatform.
On the matter, the GPL says: For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable.
Since your signature-generator is required for the very installation of the kernel [on the FritzPlatform], let alone iidentical behavior, I cosider it a vital module and thus contained in the GPL license when distributed with the binary kernel.
Linus, et al's signatures are quite different; they are decidedly not required for proper functioning of the kernel on any given hardware platform. In addition, further distancing his signatures (and thus private key) from redistribution requirements in the GPL, his signature is not distributed as part of the (source or binary) kernel. From ftp.kernel.org:
File: linux-2.5.68.tar.bz2.sign 1 KB 4/20/2003 3:02:00 AM
This is an independant file not required for the download, compilation, installation, or use of the kernel. It is as much part of the kernel as the announcing press release is.
I feel that the question of whether signing keys are covered by the GPL is still very much open. I've obviously made my feelings clear on the issue, but it will be interesting to see how it pans out.
Instead, this is meant to be run on the incoming SMTP server, the one that receives the mail. It will only hurt the spammer if he's trying to send a bunch of spam to your domain, but every server running this can help.