"I get an email with a paragraph of words that are both fairly common AND fairly unlikely to appear in spam, then the spam plug...you could flag this message as spam, but then you slowly destroy half of what makes Baysian filtering work: The list of words that are not in spam."
This will not necessarily happen. These words will become neutral and not affect the score either way. There will still be words that appear in your email correspondence but not in the spam. First off, a Bayesian filter will in effect build up a whitelist of addresses to accept, so you need to be more worried about what words will show up in messages from random strangers who try to email you. They will be mailing about your web page, your business, or topics related to mailing lists and usenet groups you participate in. This should provide a more specialized set of words that are common in your mail but less likely to be chosen at random by a spam generator.
A fascinating feature of these filters is that they are tuned for specific characteristics your email. There is no general way of choosing content of a spam so it will slip through everyone's Bayesian filter.
"I don't really feel sorry for the people that are going to loose large with the influx of daytraders that see SCO as a good inventment and powerful force that can shake even the mighty IBM."
It's lose, dammit! But to the point. Do we know who is buying the SCO stock and sending its price to absurd heights? It could be day traders, but it could just as well be some other arm of the Canopy Group buying SCO stock to pump the price so the SCO execs get richer off their options. Sounds like a perfect way for a bunch of interlocking companies to funnel money to their execs in an indirect way that won't show up on anyone's quarterly report. "Follow the money."
"this particular grievance has been addressed in Mozilla. Highlight the URL you want to copy, go to Mozilla, click the middle button anywhere on the page"
This worked in Netscape 4. This has been around for a long time. Better than that, in my.emacs file I have (global-set-key [f5] 'browse-url-at-point)
so hitting F5 when the cursor is over a url in emacs automatically loads the page in whatever browser is running.
"I never really understood the difference between/opt and/usr/local"
/opt is for when you want to keep all the files of a program together. For example, if you want to install a 30-day test version of a portland group compiler, you don't want it to spread stuff around/usr/local/lib and/usr/local/bin when it is going to stop working in a few weeks anyway./usr/local should mimic the/usr hierarchy, so things like/usr/local/netscape are now no-nos.
A lot of this doesn't matter as much when things are managed by a packaging system.
"/var: Stuff that changes often. Mail & printer spools, log files,..."
/var is for stuff that has to be writeable and local for the system to work. The idea is that you could run with/usr,/etc and so on mounted read only or over a network, and/var on a local disk mounted read-write.
From SCO's response: If IBM wants customers to accept the GPL risk, it should indemnify them against that risk. The continuing refusal to provide customer indemnification is IBM's truest measure of belief in its recently filed claims.
I would say this shows IBM's faith that IBM will still exist a year from now and will have to honor any contracts it makes. SCO can offer any indemnification they want to if they plan on going bankrupt is 3 months.
"SCO's license is actually no license at all - if it really is found that the Linux kernel contains any infringing code, the GPL forbids everyone who possesses a copy from using it at all. No one would be allowed to continue using Linux until the improper code was removed."
Could you explain the principle behind this statement? The GPL does not affect usage of the programs under it. It is not a user license, it only applies to modification and distribution.
The the kernel contains infringing code, the GPL will prohibit any modification and redistribution by anyone (including SCO, which they don't appear to understand). It will also be true that the previous distribution of the code will have been done illegally.
The whole strength of the GPL is that is based on giving extra rights that are otherwise restricted by copyright. The GPL doesn't take away any rights at all. How could the GPL disallow running the kernel?
"...except that the end users didn't steal anything, so have no need for a SCO licence."
The only possible situation in which Linux users could need a license is if the kernel infringed on patents owned by SCO. You need permission to use something covered by a patent regardless of how you got it. This is why Unisys could demand money for making gifs using patented LZW compression. However, patents are the one form of IP that so far don't seem to be involved here.
"Maybe SCO will claim ownership of everything right down to the ideas of pipes, inodes, and accessing devices as files."
Notice that SCO said "the problems inherent in Linux" (emphasis mine). They are implying here that they are aiming for a much bigger target than just JFS, NUMA, etc. from IBM, because none of those is an inherent part of Linux. It appears that they have a plan for trying to extort money from Linux distributers and users continuously for the forseeable future. None of the specific allegations to this point is large enough to have a long term effect on Linux, so they must be planning to eventually claim control over the whole Linux kernel. On what possible grounds?
Even if it is at heart a pump-and-dump scheme, they still need plausible deniability; they have to offer some other justification for their actions. They must have an exit strategy that will get them away from the SEC, but I don't see what it is. Being bought out (by e.g. IBM) seems impossible now. Where do they really think they're going with this?
"What I'm really curious about is Lindows. They apparently have the right to distribute Linux..."
If they have the right to distribute Linux, everyone has the right to distribute Linux. If the whole kernel is under the GPL, then anyone can distribute it. If the whole kernel is not under the GPL, then no one has the right to distribute it. There is no gray area and no third option.
The only thing SCO can possibly do (if their claims are valid) is sell licenses to permit people who have kernels to use the SCO IP in the kernel. These kernels will still have been distributed illegally. If hypothetically SCO's claims are valid, the affected kernels cannot be distributed at all period until SCO's IP is removed or SCO releases it under the GPL. (Which they already have, apparently, but that's a different issue.) (Lindows appears to use the 2.4 kernel. It is possible but doubtful that they have a special kernel with alleged SCO IP removed.)
"Basically, I want to be sure that if the code which SCO showed me is removed from the Linux kernel, that there is absolutely no reason to think that I had anything to do with it."
Let's assume for a moment that they have lots and lots of examples of "stolen" code. All the people who signed the NDAs might not have been shown the same code. Perhaps one offending example was shown only to Ian. If this example is leaked, then they would know who it came from, and Ian would be in trouble up to his eyeballs.
I think this applies to the original authors of a program. A program can only be licensed via the GPL if all the copyright owners agree to release it under the GPL. After a program is initially released under the GPL, it stays under the GPL, regardless of whether future contributers include their own copyright notices or not.
It is not as if each line in a program is separately licensed. If I add a new function to someone else's GPLed main.c, I don't need to put a separate copyright and GPL notice for that new function. My additional code is automatically placed under the GPL if I ever distribute the new main.c.
"The only way they feel you could "clean" Linux is to revert to kernel 2.2 and restart development from there, but none of the existing developers or even Linux users could work on development because they've already been tainted"
I will take the code, though I do not know the way...
In a way, it is good that they would say this, because it is obviously overbroad. It is another statement that fails the smell test. How could someone who has never read the kernel code know their IP?
It also shows again how they are twisting the law to generate FUD. Their supposed rights to any derivative code are based on their contract with IBM. Which Linus did not sign.
Under what legal philosophy can they be talking about "tainted" code? It could only be trade secrets. Can't be copyrights. Authors read books written by other people, just as programmers see copyrighted code all the time. If you read a copyrighted implementation of a hash table, that doesn't mean you can never write your own hash table. Not patents, because then it wouldn't matter where the code came from.
Even if they can claim some ownership over e.g. the code written by IBM, that does not necessarily make that code a trade secret owned by SCO. They would need to argue its status as "secret" on top of justifying their ownership claims.
"To me the difference between a 7 (or 7.0) and a 7.1 (etc.) is that the x.0 series always contained large changes that affected the way the base system operated. And could be expected to be a bit unstable. But the bugs that inevitably crept in would be exorcised during the upgrades to the higher point releases."
That is a very good point. I was looking forward to RH 8.1 for that very reason. Maybe this is part of the problem: too many people avoiding x.0 releases because of the perception (true or not) that they aren't quite finished yet. They will need to imply somehow that these releases are better than the old x.0 releases (and I hope it is true).
"Basically, the error you are making is that just because you have already have a copy of the code means that there is no distribution involved in licensing the SCO code. But, it doesn't work that way."
Who says it doesn't work that way? Can you explain how that could possibly be considered distribution?
This makes me think of Unisys/GIF. People bought software that made GIFs. This was illegal. Unisys offered licenses which allowed people to legally use the softare they already had. What's the difference?
"A new kernel version is a definite good reason for a whole version increment. But when they called it 9 instead of 9.0, I considered it a very bad sign."
It may not be that complicated. How big is the difference between a Redhat 7.1 with all the updates and Redhat 7.3 with all the updates? Not very much. But it must be expensive to support so many versions concurrently. I bet they will still sell CDs. They would be insane to assume every home user will download the isos. OTOH it is not so far out to let people download all the updates and dispense with x.0, x.1, x.2... Or even offer a CD of all the updates for $3, new one every month, for those who don't want to download.
Here's the part I don't understand: This happened a long time ago, in the 80's for macs and in the 90's for windows. It's one of the main reasons some of us switched to Linux in the first place! To know what's going on and to be able to fix it.
Every time you add another layer of abstraction, you add the possibility that some wacky confusing problem is in the new layer, in its interpretation of what is happening at the lower levels. (Kind of like a program reporting "file not found" if it fails to open a file for writing because the file exists but the user lacks write permission.) If these things get bad, then you lose the ability to fix things and you get to the stage where repairs consist of wipe the system and reinstall from scratch.
Why is this? Because companies like Redhat and Ximian want to make money off of Linux by using it to replace windows (not saying that's bad, just true). For people who don't want a windows-like environment, these changes are heading in the wrong direction. (please don't tell me I have to switch to debian:-).
"Why don't the linux distros just file suit, now that SCO has crossed the line, and subpeona what exactly the line numbers of these fabled bits of SCO code in the linux kernel are?"
My guess is that they are waiting for IBM to smash them first. IBM has the money and the lawyers to work on proving the facts of the case. Once these are proven, then distros etc. can go after SCO without have to prove what SCO did. All they would need to show is what loss they suffered. SCO would lose every such lawsuit almost automatically; the only question would be how bad.
"If, on the other hand they say, "your existing license is invalid, here have another." Then they are in the right."
Yes yes yes! Pay attention people! This is a license for using the kernel, for running the code. The GPL has nothing to say about who can run the code; it is a distribution license, not a user license. If indeed the 2.4 Linux kernel contains stolen SCO code, then distribution of it is illegal. Period. The SCO part of the code can be used iff you have a license from SCO to do so.
There is another wrinkle. If you buy a license, you are implicitly accepting that SCO owns rights to some of the kernel, and any copying or distribution of the kernel is illegal. Anyone who buys a SCO license and then downloads, installs, or distributes a copy of the kernel is committing an act which he has effectively admitted is illegal.
Of course, SCO knowingly distributing the kernel with the GPL intact probably means the code isn't stolen, in which case SCO has no case and none of this applies.
"The GPL is a redistribution license, not a usage license. I think as long as SCO doesn't distribute Linux, they can grant a license to use their code in Linux without the GPL interfering."
I think you are hitting on a subtle and possibly very important point. Assume hypothetically that SCO is correct in its assertions. I have a kernel that I bought from Redhat. This redistribution by RH would then turn out to have been illegal. But I have the kernel, and the GPL has no effect on me if I use it and don't redistribute, so the GPL can't stop me from using the kernel. But it would be illegal to use the proprietary code "stolen" from SCO. A license from SCO would allow me to run the kernel, but not redistribute it, because that would violate the GPL. I think if SCO's assertions were valid, then they could indeed sell licenses that would allow people to use kernels they already have.
This of course ignores the implications of SCO's continued distribution of kernels under the GPL.
"less then 10% of RH's income is from the distro and they would drop it if they could. It was only a marketing tool for them."
Advertising doesn't generate any income at all; it's just marketing, yet companies still advertise. Even if they just break even on the boxed sets, it seems crazy to drop them. I would guess that often when a business spends a gazillion dollars on RH support and training, the decision was driven by some guys who have a row of RH boxed sets at home.
Despite what many people here have said, I think that there are a lot of people who don't have broadband at home (or CD burners at work) and (believe it or not) there are companies that have only dialup net access. These people (e.g. me) will apparently no longer be Redhat customers. Although I would guess the mystery announcement will also say that RH will still sell CDs through Amazon or something. Still, they seem to be cutting off their grassroots support.
"The fact that MS was willing to let Munich unbundle office is indicative that people dont want to pay for huge monster suites that they arent using most of. And in a govt organization this is even more true. Your average memo writing paper pusher doenst need to use excel."
Doesn't somebody (joelonsoftware ?) claim that the most common use for excell is making lists? Hard to say in advance what a user will need, b/c people don't use programs in the way you think they will (e.g. if some PHB decides to start sending out meeting agendas as xls, everyone will need to be able to view them). You could also say that the unbundling is clever marketing, because anyone who gets a machine w/o excell or whatever and then later needs to have it it will have to pay more for it.
"The study claimed that for the pure linux solution they would have to buy a lot of new peripherials (card readers, printer etc) which makes the linux solution more expensive."
Just shows that they don't understand open source as well as they could. Why replace millions of dollars worth of unsupported hardware when you could get Linux drivers written for what you have? Either by hiring programmers or contracting it out or for a fee from the manufacturer. (Since the whole thing was 30 some million, if the hardware costs mattered they must have been in the millions.) This is a different way of looking at the problem than what they are used to, but it is one idea of how open source is supposed to work.
"Listening to him attempt to code (in Perl) was pretty funny, e.g. saying "twiddle" for "~" and such."
It's a tilde, dammit! Sound like a bunch four year olds saying "twiddle". Any coding would be like this... "Left-angle h-t-m-l right-angle enter left-angle exclamation-point dash dash..." or "right-parenthesis right-parenthesis right-parenthesis right-parenthesis..."
Slashdot Mirror
This will not necessarily happen. These words will become neutral and not affect the score either way. There will still be words that appear in your email correspondence but not in the spam. First off, a Bayesian filter will in effect build up a whitelist of addresses to accept, so you need to be more worried about what words will show up in messages from random strangers who try to email you. They will be mailing about your web page, your business, or topics related to mailing lists and usenet groups you participate in. This should provide a more specialized set of words that are common in your mail but less likely to be chosen at random by a spam generator.
A fascinating feature of these filters is that they are tuned for specific characteristics your email. There is no general way of choosing content of a spam so it will slip through everyone's Bayesian filter.
It's lose, dammit! But to the point. Do we know who is buying the SCO stock and sending its price to absurd heights? It could be day traders, but it could just as well be some other arm of the Canopy Group buying SCO stock to pump the price so the SCO execs get richer off their options. Sounds like a perfect way for a bunch of interlocking companies to funnel money to their execs in an indirect way that won't show up on anyone's quarterly report. "Follow the money."
This worked in Netscape 4. This has been around for a long time. Better than that, in my .emacs file I have
(global-set-key [f5] 'browse-url-at-point)
so hitting F5 when the cursor is over a url in emacs automatically loads the page in whatever browser is running.
A lot of this doesn't matter as much when things are managed by a packaging system.
"/var: Stuff that changes often. Mail & printer spools, log files,..."
I would say this shows IBM's faith that IBM will still exist a year from now and will have to honor any contracts it makes. SCO can offer any indemnification they want to if they plan on going bankrupt is 3 months.
Could you explain the principle behind this statement? The GPL does not affect usage of the programs under it. It is not a user license, it only applies to modification and distribution.
The the kernel contains infringing code, the GPL will prohibit any modification and redistribution by anyone (including SCO, which they don't appear to understand). It will also be true that the previous distribution of the code will have been done illegally.
The whole strength of the GPL is that is based on giving extra rights that are otherwise restricted by copyright. The GPL doesn't take away any rights at all. How could the GPL disallow running the kernel?
That would be /usr/bin/finger --middle
The only possible situation in which Linux users could need a license is if the kernel infringed on patents owned by SCO. You need permission to use something covered by a patent regardless of how you got it. This is why Unisys could demand money for making gifs using patented LZW compression. However, patents are the one form of IP that so far don't seem to be involved here.
Notice that SCO said "the problems inherent in Linux" (emphasis mine). They are implying here that they are aiming for a much bigger target than just JFS, NUMA, etc. from IBM, because none of those is an inherent part of Linux. It appears that they have a plan for trying to extort money from Linux distributers and users continuously for the forseeable future. None of the specific allegations to this point is large enough to have a long term effect on Linux, so they must be planning to eventually claim control over the whole Linux kernel. On what possible grounds?
Even if it is at heart a pump-and-dump scheme, they still need plausible deniability; they have to offer some other justification for their actions. They must have an exit strategy that will get them away from the SEC, but I don't see what it is. Being bought out (by e.g. IBM) seems impossible now. Where do they really think they're going with this?
If they have the right to distribute Linux, everyone has the right to distribute Linux. If the whole kernel is under the GPL, then anyone can distribute it. If the whole kernel is not under the GPL, then no one has the right to distribute it. There is no gray area and no third option.
The only thing SCO can possibly do (if their claims are valid) is sell licenses to permit people who have kernels to use the SCO IP in the kernel. These kernels will still have been distributed illegally. If hypothetically SCO's claims are valid, the affected kernels cannot be distributed at all period until SCO's IP is removed or SCO releases it under the GPL. (Which they already have, apparently, but that's a different issue.) (Lindows appears to use the 2.4 kernel. It is possible but doubtful that they have a special kernel with alleged SCO IP removed.)
Let's assume for a moment that they have lots and lots of examples of "stolen" code. All the people who signed the NDAs might not have been shown the same code. Perhaps one offending example was shown only to Ian. If this example is leaked, then they would know who it came from, and Ian would be in trouble up to his eyeballs.
I think this applies to the original authors of a program. A program can only be licensed via the GPL if all the copyright owners agree to release it under the GPL. After a program is initially released under the GPL, it stays under the GPL, regardless of whether future contributers include their own copyright notices or not.
It is not as if each line in a program is separately licensed. If I add a new function to someone else's GPLed main.c, I don't need to put a separate copyright and GPL notice for that new function. My additional code is automatically placed under the GPL if I ever distribute the new main.c.
I will take the code, though I do not know the way...
In a way, it is good that they would say this, because it is obviously overbroad. It is another statement that fails the smell test. How could someone who has never read the kernel code know their IP?
It also shows again how they are twisting the law to generate FUD. Their supposed rights to any derivative code are based on their contract with IBM. Which Linus did not sign.
Under what legal philosophy can they be talking about "tainted" code? It could only be trade secrets. Can't be copyrights. Authors read books written by other people, just as programmers see copyrighted code all the time. If you read a copyrighted implementation of a hash table, that doesn't mean you can never write your own hash table. Not patents, because then it wouldn't matter where the code came from.
Even if they can claim some ownership over e.g. the code written by IBM, that does not necessarily make that code a trade secret owned by SCO. They would need to argue its status as "secret" on top of justifying their ownership claims.
That is a very good point. I was looking forward to RH 8.1 for that very reason. Maybe this is part of the problem: too many people avoiding x.0 releases because of the perception (true or not) that they aren't quite finished yet. They will need to imply somehow that these releases are better than the old x.0 releases (and I hope it is true).
Who says it doesn't work that way? Can you explain how that could possibly be considered distribution?
This makes me think of Unisys/GIF. People bought software that made GIFs. This was illegal. Unisys offered licenses which allowed people to legally use the softare they already had. What's the difference?
It may not be that complicated. How big is the difference between a Redhat 7.1 with all the updates and Redhat 7.3 with all the updates? Not very much. But it must be expensive to support so many versions concurrently. I bet they will still sell CDs. They would be insane to assume every home user will download the isos. OTOH it is not so far out to let people download all the updates and dispense with x.0, x.1, x.2... Or even offer a CD of all the updates for $3, new one every month, for those who don't want to download.
Or, heaven help us, Redhat X.
Here's the part I don't understand: This happened a long time ago, in the 80's for macs and in the 90's for windows. It's one of the main reasons some of us switched to Linux in the first place! To know what's going on and to be able to fix it.
Every time you add another layer of abstraction, you add the possibility that some wacky confusing problem is in the new layer, in its interpretation of what is happening at the lower levels. (Kind of like a program reporting "file not found" if it fails to open a file for writing because the file exists but the user lacks write permission.) If these things get bad, then you lose the ability to fix things and you get to the stage where repairs consist of wipe the system and reinstall from scratch.
Why is this? Because companies like Redhat and Ximian want to make money off of Linux by using it to replace windows (not saying that's bad, just true). For people who don't want a windows-like environment, these changes are heading in the wrong direction. (please don't tell me I have to switch to debian :-).
My guess is that they are waiting for IBM to smash them first. IBM has the money and the lawyers to work on proving the facts of the case. Once these are proven, then distros etc. can go after SCO without have to prove what SCO did. All they would need to show is what loss they suffered. SCO would lose every such lawsuit almost automatically; the only question would be how bad.
Yes yes yes! Pay attention people! This is a license for using the kernel, for running the code. The GPL has nothing to say about who can run the code; it is a distribution license, not a user license. If indeed the 2.4 Linux kernel contains stolen SCO code, then distribution of it is illegal. Period. The SCO part of the code can be used iff you have a license from SCO to do so.
There is another wrinkle. If you buy a license, you are implicitly accepting that SCO owns rights to some of the kernel, and any copying or distribution of the kernel is illegal. Anyone who buys a SCO license and then downloads, installs, or distributes a copy of the kernel is committing an act which he has effectively admitted is illegal.
Of course, SCO knowingly distributing the kernel with the GPL intact probably means the code isn't stolen, in which case SCO has no case and none of this applies.
I think you are hitting on a subtle and possibly very important point. Assume hypothetically that SCO is correct in its assertions. I have a kernel that I bought from Redhat. This redistribution by RH would then turn out to have been illegal. But I have the kernel, and the GPL has no effect on me if I use it and don't redistribute, so the GPL can't stop me from using the kernel. But it would be illegal to use the proprietary code "stolen" from SCO. A license from SCO would allow me to run the kernel, but not redistribute it, because that would violate the GPL. I think if SCO's assertions were valid, then they could indeed sell licenses that would allow people to use kernels they already have.
This of course ignores the implications of SCO's continued distribution of kernels under the GPL.
Advertising doesn't generate any income at all; it's just marketing, yet companies still advertise. Even if they just break even on the boxed sets, it seems crazy to drop them. I would guess that often when a business spends a gazillion dollars on RH support and training, the decision was driven by some guys who have a row of RH boxed sets at home.
Despite what many people here have said, I think that there are a lot of people who don't have broadband at home (or CD burners at work) and (believe it or not) there are companies that have only dialup net access. These people (e.g. me) will apparently no longer be Redhat customers. Although I would guess the mystery announcement will also say that RH will still sell CDs through Amazon or something. Still, they seem to be cutting off their grassroots support.
Doesn't somebody (joelonsoftware ?) claim that the most common use for excell is making lists? Hard to say in advance what a user will need, b/c people don't use programs in the way you think they will (e.g. if some PHB decides to start sending out meeting agendas as xls, everyone will need to be able to view them). You could also say that the unbundling is clever marketing, because anyone who gets a machine w/o excell or whatever and then later needs to have it it will have to pay more for it.
Just shows that they don't understand open source as well as they could. Why replace millions of dollars worth of unsupported hardware when you could get Linux drivers written for what you have? Either by hiring programmers or contracting it out or for a fee from the manufacturer. (Since the whole thing was 30 some million, if the hardware costs mattered they must have been in the millions.) This is a different way of looking at the problem than what they are used to, but it is one idea of how open source is supposed to work.
It's a tilde, dammit! Sound like a bunch four year olds saying "twiddle". Any coding would be like this... "Left-angle h-t-m-l right-angle enter left-angle exclamation-point dash dash ..." or "right-parenthesis right-parenthesis right-parenthesis right-parenthesis..."