There are several ports to other languages underway, including my own work on libbt, a C-language implementation intended to be suitable for use as a library.
Reading SCO's response to Novell is highly amusing. We concede all of your points, but we're going to keep going the way we have been. These guys need a reality check. But then so does their market cap.
>HST must consist of unfathomably cool (literally
>and figuratively) electronics.
Do you really mean `literally cool`?
The only thing worse than a grammar pedant is a grammar pedant who doesn't read the article he is criticizing. Since uneven heat flow would lead to spurious readings from a normal CCD; yes, I did 'really' mean 'literally cool'.
Imagine a Beowulf... um. Seriously, how do you cope with reciprocity failure in a 3.5 day exposure. I would have thought that stray heat or electron flow would turn the whole image to static with such a long exposure. HST must consist of unfathomably cool (literally and figuratively) electronics.
Re:What is an acceptable risk?
on
Shuttle Politics
·
· Score: 4, Insightful
I would easily say that 1/62.5 is acceptable. In fact, I'm quite impressed that it's not 1/2. It's a really amazing accomplishment to do it at all. Back in the early days (even well into the Apollo program) it was pretty much given that this is a major risk to the lives of the astronauts.
Strangely enough, I had never considered combining the first and second failed
shuttle missions into a single statistic. The space shuttle program is a system, with a failure rate that varies over time, not a single 20 year long experiment. I would rather say that the failure rate at the time of 51L was a little under 10%, and that the system now has a failure rate of a little over 1%, although good statistics don't really apply to such small sample sizes. Still I would hestitate a long time before replacing the known failure rate of
a 20 year old system, with a new and unproven system which still has all of
its bugs intact. Nor is NASA interested, if I guess rightly.
In part I think that this is what annoys Joe Barton among others. It isn't that NASA is too risky, but too conservative. There are no new systems coming on line, and the old system isn't sexy any more. In its current state the STS is incredibly manpower intensive, and a lot of the reliability of the system depends on the training and full staffing of the shuttle program. If NASA were less risk adverse, they might be able to reset and design a new system, which over twenty years could approach the reliability of STS, but at a fraction of the cost in time and manpower.
But thinking that way will make the system less reliable, not more; at least until the bugs have been worked out.
And won't spammers be able to circumvent whatever is in place for spam prevention?
In this case, yes. The system described relies on downgrading people
with certificates who abuse their privelege and send spam anyway. The gap
in time between when they start sending spam, and when their certificate
is downgraded presents the same problem that double spends do for e-cash,
but in the case of spammers, nifty techniques for revealing the identity
of the double spender are irrelevant; the merchandise (the delivery right
of email) has already been received and consumed a thousand fold.
Fixing the incentive behind spam is an enormous and intractable problem, and this
'solution' gets no closer to solving it. Indeed the distinction between this
and PKI is minor as far as I can tell -- it merely transfers the signature before the content thereby allowing the MTA to reject the message before the
entire message has been received. If PKI were the solution to spam, then it
would long ago have had an effect.
At $17k we're talking about a semester or two of tuition at many schools. Enough to make the kids, or more likely their parents, take notice. But
in no way enough to be worth filing for bankruptcy. I very much doubt that
the RIAA cares about the $17k as actual money. And driving the kids into bankruptcy would have an even more chilling effect than the money for most people.
Advertising is an offer to redirect the user into a new information stream.
Paradoxically, the better the data on a site is organized the less likely a user is to want to break out of the information flow they are in. But decreasing the quality of information only has the effect of making it less valuable to visit that site in the first place.
Arguably, to be effective advertising should be located near where the user makes information flow decisions, and the information flow decisions cannot be
so consistent that the user learns to ignore any alternatives that are presented. For a Blog style site like your nancies page, that would mean interspersing advertising into the news stream as articles. You don't want to confuse readers, so the advertising should be in offset type and layout (mp3.com does a good job of this), but
the position of advertising within the news list should be altered from day to day.
While we are on the subject though, I should probably mention a pet peeve
that I have with online advertising. Just because you can change which advert is delivered with each page doesn't mean that you should. Varying the content from user to user is fine, but having seen the page with a specific layout once, the advertising should be left the same the next time that I view that page. Swapping out ads messes up my information flow; I have to backtrack to see if I missed something that I really wanted to read, and I may have lost a link to an ad that I really wanted to follow later in my browsing.
Re:Signed binaries subvert the GPL
on
Linus on DRM
·
· Score: 1
Binaries that "don't work right" if they have the wrong signature, effectively subvert the intent of the GPL. The GPL exists to empower users by making them not be hostages to any other party.
While I think you are right in your interpretation of the intent of the GPL, I don't think that signed binaries have the effect that you've assigned them. If
the software didn't work right because it was unsigned then you would simply
recompile the sources with a modification that allowed them to work correctly when unsigned. That isn't the case that is interesting.
The problem that Linus is addressing is whether or not it is legal to create hardware that only runs signed binaries. In essence this question asks; is it legal to run GPL'd software on top of a proprietary base. In lieu of signed binaries, you could consider the case where Linux is burned into ROM. Is burning
Linux into ROM illegal if the ROM chips are epoxied to the motherboard?
Adding code signature verification to the motherboard firmware is analogous to the epoxy. It inhibits you from changing out the software on this particular piece of hardware, but not from changing the software in general assuming you can find another piece of hardware to run it on that is open.
There isn't really any other way to see it than as a permitted use under the GPL. Otherwise any element which inhibits another element from running; the UNIX privilege system for example (when control of root access is in the hands
of someone else), could be considered a violation of the GPL. While there have been arguments about this in the FSF's past, the GPL itself has never been interpreted that way even by the FSF.
The only reason this is a problem is because the technology has made it dirt cheap to copy/steal material and distribute it, to the point where a whole generation thinks its OK. I don't think the answer is to charge a tax on everyone just because some place no value on the ownership and law. And, it remains to be seen if technology goes the direction of letting the authorities more easily watch what I'm doing or quickly track down servers pumping out illegal copies of songs.
So you have two alternatives. Either you make criminals of this 'generation [which] thinks its OK', or else you change the
law so that it actually is OK. I prefer the latter. Beyond that we'll
just have to agree that our opinions differ.
I am perfectly happy with the content that I can legally download right now. Why in the world should I have to pay a fee so that the rest of you can get your N-Sync fix?
The current system, in which companies are trying to protect their digital assets through a combination of legislation such as the DMCA, lawsuits such as the ones posted last week, and copy protection systems that prevent fair use have a net negative impact on society that goes far beyond whether you would personally benefit from being able to download N-sync. I personally have not either purchased or copied music from an RIAA member in at least two years, but that doesn't mean that I'm not impacted by societal impact of current policies.
The solution to the current intellectual property problem is simple. The copyright holders should target abusers and prosecute the hell out of them. If people are faced with the prospect of being prosecuted for distributing copyrighted works illegally then the current problems will clear themselves up.
So you don't care about N-Sync... do you want your kid, or your mom, or your brother thrown in jail for a year and fined $500k? Do you really think that taking these people out of circulation will make the US a safer and better place to live? I hardly know anyone who has never gotten a copy of an album taped by a friend, or more commonly these days, a CD-R. These are regular people you are talking about, not the criminal fringe. I am amazed that you would advocate
locking them all up over pulling a couple bucks out of your own pocket so that they could do as they please. You'll pay far more in taxes to imprison half the population than you'd ever pay to give them the liberty to share their music.
If you want positive legislation in this area, the best hope is a compulsory license, which would charge all users of the Net a small, mandatory fee. In exchange for this, it would become totally legal to distribute and use any audio and video content on the Net. The revenue from the fees would be split up among the creators according to a formula, based on how many times each work was downloaded or played. If you do the math, the fees can be pretty small while still replacing the revenue the music and movie industries would otherwise lose.
Indeed, I've come to believe the same thing. There has to be some (possibly voluntary) way of assessing what kind of content is being copied, and in what quantities in order to assign proportional royalty payments and thereby preserve
profit motive among the content creators. But freely copyable data is the only system that really makes sense to me at a societal level. Data ownership just doesn't scale in proportion with data networks and immeasurable value is lost in this way.
By distributing binaries from code mixed with GPL, you've agreed to the terms of the GPL. Opening up your sources may be the only way to keep the other copyright holder(s) from suing you. What would really happen is up to the courts to decide, as there have been no precedent setting cases.
This is basic contract law. By implication you agreed to be bound by the GPL when distributing GPL source code; and then you failed to meet the requirements of the contract. The contract cancellation clauses come in to effect at this point, and they prohibit you from any further distribution of any GPL'd code. There is no mechanism in the GPL for forfeiture of property rights. If you decide voluntarily to GPL your own sources so that you are again in compliance with the GPL then that is your choice, but is not required by any contract language in the GPL.
The only thing you've lost is the ability to distribute GPL'd code, but since you really never had that right in the first place without complying with the GPL, you have really lost nothing at all.
The worst case scenario is that you never saw or agreed to the GPL, but decided to copy the program anyway even though you were aware that it was covered by copyright. In that case you could have to forfeit the statutory copyright violation fines as well as court and lawyers fees. So far all companies violating the GPL have recognized the folly of this approach and
have either voluntarily stopped distributing GPL'd software, or have voluntarily open sourced the software that put them in non-compliance with the GPL.
He didn't say steal. He said lose. You write some code from scratch. This is now your intellectual property. You combine the code with code licensed under the GPL and distribute it. You have now "lost" your intellectual property. Sure, you still own the copyright on the code, but now everyone can see it and use it.
The original poster claimed that by distributing binary device drivers with Linux that they would eventually lose their IP. If you've kept the sources and
the source code is your own property, then nothing in the GPL can make you open up your source code. The GPL can only deny you the right to distribute software covered by the GPL -- you may end up deciding that distributing e.g Linux is more valuable to you than your source code is, or you may not, but the choice is up to you.
The court isn't going to come along and say; hey you distributed your binaries with Linux, they are obviously now GPL'd, so hand over the code.
what is your opinion on their view of binding GPL'd and non-GPL'd software together into an image (as is often done with embedded systems) ?
I haven't seen either of them express an opinion on precisely that topic, but I think that a Judge would have a hard time seeing something like a Tivo as a whole, as a derived work of Linux. Instead it would probably be seen for what it is, a specialized kind of computer with software loaded on it. Likewise, as long as the software that is loaded on the system respects the GPL and the generally accepted GPL barriers, I can't see why they would have any licensing issues.
Indirectly RMS has commented on systems like these as something he would like the next revision of the GPL to address. The copyright that the FSF plans to leverage is the right to control public performance, and probably applies more to web sites than to embedded consumer equipment.
When someone eventually sues over the binary drivers packaged with the Linux kernel, people will lose their intellectual property.
This is the worst sort of paranoid scaremongering. There is no mechanism in the GPL, nor could there be any mechanism, for stealing IP from you. In a worst case scenario you might be barred from, or even fined for distributing other peoples IP without their consent, but your own IP will always remain yours. In practical terms it hasn't ever even been necessary to fine or enjoin the distribution of GPL'd software as all companies so far have been able to come to an agreement with the FSF in mediation.
Right, but they aren't derivative of the GPL. They just created a driver. Why does "only distributing the driver" protect them
By only distributing a driver that isn't a derived work, they don't need anyone elses permission to do so.
Unfortunately for them, that isn't true of Linux. Linux is a copyrighted work, and the copyrights are retained by their authors, so nVidia has no right to copy, distribute, modify, or publically perform that work. The authors of Linux will allow
people to copy, and distribute Linux under certain conditions and those conditions are spelled out in the GPL, which among other things enumerates certain responsibilities that a company must accept in order to distribute Linux. One of those responsibilities is that Linux must not be linked with proprietary libraries that are distributed under terms more restrictive than the GPL itself. Since nVidia cannot comply with that responsibility, it cannot distribute Linux.
Thus as long as nVidia only distribute their own non-derived drivers they are perfectly within their rights. If they were to open source their drivers then they would be within their rights as granted by the GPL license to also distribute Linux. But as long as the drivers are maintained as a proprietary product that is linked to the GPL'd kernel, they have no right to distribute Linux.
What's the difference between A) Having two pages, one where a person downloads Linux with the source, and another page where one downloads binary-only drivers - and B) A single page where you download Linux with it's source and the binary-only drivers..
It isn't a question of where, but a question of who. The same 'who' cannot offer you both Linux, and these drivers, and remain compliant with both licenses.
In case you are interested, here is some more detail on the mechanics of the GPL. The use of software isn't a right that is reserved to the copyright holder; that is you don't have to contact the copyright holder to run a program, read a book, or any other use that you might want to make of the copy righted item. The GPL does not attempt to force you to accept the GPL until you want to do something that would be a reserved right (like copying and distributing), so it isn't really that nVidia is out of compliance, so much as that they would not be able to comply with the requirements of the GPL if they were to distribute Linux. Also the GPL can be read in a few different ways; in this post I'm representing the GPL the way it is understood by Eben Moglen and the FSF, and as interpreted by Linus Torvalds in their respective writings on the issue. There are other stricter interpretations but it is IMHO scaremongering to talk about those interpretations when the original author of the license is on record with looser interpretations. Likewise there are people who attempt to interpret symbolic linking as distinct from linking and thus a valid barrier between GPL and non-GPL code; while this approach hasn't been tested in court, it goes completely counter to the interpretation of the FSF and thus will land you in legal hot water if you insist on that interpretation.
Okay, this post made absolutely no sense. The whole idea of being "out of compliance with the GPL" is nonsense, completely. There is nothing stopping them from packaging Linux and their driver together and distributing them. The GPL, under which the Linux kernel is placed, states that if they distribute the Linux kernel, they have to provide source... for the kernel!
Actually the GPL states that if you distribute Linux, then you must also distribute the source code to Linux, and anything that was linked with GPL'd software, under a license which is no more restrictive than the GPL. Their code being non-GPL does not give them the right to distribute GPL'd products, only the GPL does that, and they aren't complying with the provisions under which the GPL allows them to distribute that software, so they must not do so.
As well, Linus himself has stated that binary drivers which are linked to the kernel do not "inherit" the GPL. Does that mean they can't package GPLd and non-GPLd code together?
While nVidia's drivers are not covered by the GPL, Linux is. To distribute Linux you must comply with the GPL, and linking binary drivers with the GPL kernel is non-compliant. The GPL does not permit you to distribute GPL'd software if you do not distribute the complete sources, including any code that is linked to the GPL'd code. The only exception provided is for libraries which are a part of the base operating system on which the GPL'd code runs. Since Linux is the operating system, there are no non-GPL'd libraries that Linux links with that are a part of the operating system the Linux runs on.
Plenty of distributions have done this in the past (Redhat has demos of various closed source apps, IIRC, as part of the CD package). The GPL applies only to the code it's been applied to (along with derivative works, etc... and no, "derivative" does not include being packaged together on the same CD).
This is a completely different issue. If you comply with the terms of the GPL license then you have the right to distribute GPL'd software. Since e.g: Netscape doesn't link to the Kernel (it only links to LGPL'd code, not GPL'd code), Netscape and Linux can easily be distributed by the same company (assuming that they have the right to distribute Netscape) since there is no
conflict.
If nVidia's drivers were derivative of GPL'd source code then they would have no right to distribute them at all, with or without the Kernel.
No, NVidia get away with distributing their drivers by not distributing Linux. Distributing their drivers puts them out of compliance with the GPL, and thus they have no right to distribute Linux, but since their drivers aren't GPL'd, being
out of compliance with the GPL doesn't affect their ability to distribute their
own product.
That doesn't work so well when the end result is an embedded product and the seller will have to distribute both Linux and the proprietary drivers at the same time.
I think your Lawyers are basically correct in this. It isn't really a matter of protecting your own IP, but of complying with the license you must agree to in order to distribute someone elses IP, namely Linux. If you write a proprietary device driver that links directly into the kernel, then so long as you own the entire proprietary source code to that driver, you don't have to comply with the GPL -- the reason you don't have to comply with the GPL is because you aren't distributing GPL'd code; you are only distributing your own code.
As soon as you start distributing GPL'd code in addition to your proprietary drivers you will be in trouble. The GPL only allows you to distribute the GPL'd code if any code that you link to it is likewise GPL'd or under an equivalently free license. At that point you are out of compliance with the license, and thus have no right to distribute the GPL'd code.
So I'm sorry, but I agree with your lawyers on this one.
Ah yes, Monday night baby brigade at the Parkway is a blast. No one looks at you funny if your kid is crying (although mostly mine sleeps), and there are couches to sit on instead of chairs if you get in early enough. And the pizzas are
priced reasonably as well. On the other hand $5/seat is on the high end for B run
theaters, but I'm never aware that they are late in the run anyway since I never get out to the first run theaters.
I've read the complete Chronicles of Thomas Covenant, and the story is certainly epic, but so tragic I hesitate to ever recommend it to anyone. Life in the Realm
starts out bad, and ends up worse, followed by disastrous, etc. Additionally the second trilogy was IMHO poorly planned out, and only stumbled to a discordant conclusion. Likewise his through the looking glass series was not at all interesting to me. So, yes, good character development, but not all enjoyable reading despite that.
I haven't read Iain Banks though, and thanks for the recommendation.
Not because I find his books especially well written, but because I'm a Sci-Fi junkie. His characters tend to feel like cardboard propups. Not that I'm defending Lucas' crap either.
What exactly do you consider good character development? Anne McCaffrey? I read Brin specifically because I think his character development is better than most of the other SF authors. The only comparable authors I know of are Larry Niven and Scott Card (and maybe Stephen Brust although I'm not fond of his nihilism.) Benford's science is better but his characters are paper thin. McCaffrey's characters stir up emotion, but she recycles the same character and plot over and over again. Greg Bear is just awful in about every way I can think of. I like Neal Stephenson, but I wouldn't read him for his characters. Pournelle? Brooks? Haldeman? Gibson? Robinson? Bujold?
Only two authors that come to mind are better than Brin in character development, and those are Frank Herbert, and JRR Tolkien. The rest are
either thin, carbon copies, or serialized with no transformation from one
novel to the next. But if you have recommendations, please tell.
Slashdot Mirror
There are several ports to other languages underway, including my own work on libbt, a C-language implementation intended to be suitable for use as a library.
Reading SCO's response to Novell is highly amusing. We concede all of your points, but we're going to keep going the way we have been. These guys need a reality check. But then so does their market cap.
>and figuratively) electronics.
Do you really mean `literally cool`?
The only thing worse than a grammar pedant is a grammar pedant who doesn't read the article he is criticizing. Since uneven heat flow would lead to spurious readings from a normal CCD; yes, I did 'really' mean 'literally cool'.
Imagine a Beowulf... um. Seriously, how do you cope with reciprocity failure in a 3.5 day exposure. I would have thought that stray heat or electron flow would turn the whole image to static with such a long exposure. HST must consist of unfathomably cool (literally and figuratively) electronics.
Strangely enough, I had never considered combining the first and second failed shuttle missions into a single statistic. The space shuttle program is a system, with a failure rate that varies over time, not a single 20 year long experiment. I would rather say that the failure rate at the time of 51L was a little under 10%, and that the system now has a failure rate of a little over 1%, although good statistics don't really apply to such small sample sizes. Still I would hestitate a long time before replacing the known failure rate of a 20 year old system, with a new and unproven system which still has all of its bugs intact. Nor is NASA interested, if I guess rightly.
In part I think that this is what annoys Joe Barton among others. It isn't that NASA is too risky, but too conservative. There are no new systems coming on line, and the old system isn't sexy any more. In its current state the STS is incredibly manpower intensive, and a lot of the reliability of the system depends on the training and full staffing of the shuttle program. If NASA were less risk adverse, they might be able to reset and design a new system, which over twenty years could approach the reliability of STS, but at a fraction of the cost in time and manpower.
But thinking that way will make the system less reliable, not more; at least until the bugs have been worked out.
And won't spammers be able to circumvent whatever is in place for spam prevention?
In this case, yes. The system described relies on downgrading people with certificates who abuse their privelege and send spam anyway. The gap in time between when they start sending spam, and when their certificate is downgraded presents the same problem that double spends do for e-cash, but in the case of spammers, nifty techniques for revealing the identity of the double spender are irrelevant; the merchandise (the delivery right of email) has already been received and consumed a thousand fold.
Fixing the incentive behind spam is an enormous and intractable problem, and this 'solution' gets no closer to solving it. Indeed the distinction between this and PKI is minor as far as I can tell -- it merely transfers the signature before the content thereby allowing the MTA to reject the message before the entire message has been received. If PKI were the solution to spam, then it would long ago have had an effect.
At $17k we're talking about a semester or two of tuition at many schools. Enough to make the kids, or more likely their parents, take notice. But in no way enough to be worth filing for bankruptcy. I very much doubt that the RIAA cares about the $17k as actual money. And driving the kids into bankruptcy would have an even more chilling effect than the money for most people.
Paradoxically, the better the data on a site is organized the less likely a user is to want to break out of the information flow they are in. But decreasing the quality of information only has the effect of making it less valuable to visit that site in the first place.
Arguably, to be effective advertising should be located near where the user makes information flow decisions, and the information flow decisions cannot be so consistent that the user learns to ignore any alternatives that are presented. For a Blog style site like your nancies page, that would mean interspersing advertising into the news stream as articles. You don't want to confuse readers, so the advertising should be in offset type and layout (mp3.com does a good job of this), but the position of advertising within the news list should be altered from day to day.
While we are on the subject though, I should probably mention a pet peeve that I have with online advertising. Just because you can change which advert is delivered with each page doesn't mean that you should. Varying the content from user to user is fine, but having seen the page with a specific layout once, the advertising should be left the same the next time that I view that page. Swapping out ads messes up my information flow; I have to backtrack to see if I missed something that I really wanted to read, and I may have lost a link to an ad that I really wanted to follow later in my browsing.
While I think you are right in your interpretation of the intent of the GPL, I don't think that signed binaries have the effect that you've assigned them. If the software didn't work right because it was unsigned then you would simply recompile the sources with a modification that allowed them to work correctly when unsigned. That isn't the case that is interesting.
The problem that Linus is addressing is whether or not it is legal to create hardware that only runs signed binaries. In essence this question asks; is it legal to run GPL'd software on top of a proprietary base. In lieu of signed binaries, you could consider the case where Linux is burned into ROM. Is burning Linux into ROM illegal if the ROM chips are epoxied to the motherboard?
Adding code signature verification to the motherboard firmware is analogous to the epoxy. It inhibits you from changing out the software on this particular piece of hardware, but not from changing the software in general assuming you can find another piece of hardware to run it on that is open.
There isn't really any other way to see it than as a permitted use under the GPL. Otherwise any element which inhibits another element from running; the UNIX privilege system for example (when control of root access is in the hands of someone else), could be considered a violation of the GPL. While there have been arguments about this in the FSF's past, the GPL itself has never been interpreted that way even by the FSF.
So you have two alternatives. Either you make criminals of this 'generation [which] thinks its OK', or else you change the law so that it actually is OK. I prefer the latter. Beyond that we'll just have to agree that our opinions differ.
The current system, in which companies are trying to protect their digital assets through a combination of legislation such as the DMCA, lawsuits such as the ones posted last week, and copy protection systems that prevent fair use have a net negative impact on society that goes far beyond whether you would personally benefit from being able to download N-sync. I personally have not either purchased or copied music from an RIAA member in at least two years, but that doesn't mean that I'm not impacted by societal impact of current policies.
The solution to the current intellectual property problem is simple. The copyright holders should target abusers and prosecute the hell out of them. If people are faced with the prospect of being prosecuted for distributing copyrighted works illegally then the current problems will clear themselves up.
So you don't care about N-Sync... do you want your kid, or your mom, or your brother thrown in jail for a year and fined $500k? Do you really think that taking these people out of circulation will make the US a safer and better place to live? I hardly know anyone who has never gotten a copy of an album taped by a friend, or more commonly these days, a CD-R. These are regular people you are talking about, not the criminal fringe. I am amazed that you would advocate locking them all up over pulling a couple bucks out of your own pocket so that they could do as they please. You'll pay far more in taxes to imprison half the population than you'd ever pay to give them the liberty to share their music.
Indeed, I've come to believe the same thing. There has to be some (possibly voluntary) way of assessing what kind of content is being copied, and in what quantities in order to assign proportional royalty payments and thereby preserve profit motive among the content creators. But freely copyable data is the only system that really makes sense to me at a societal level. Data ownership just doesn't scale in proportion with data networks and immeasurable value is lost in this way.
Isn't it fraudulent to sell someone a perpetual license for software that you've knowingly designed to stop working after two years?
The title of Czar seems lately to have taken on some of the same connotations as buffoon.
This is basic contract law. By implication you agreed to be bound by the GPL when distributing GPL source code; and then you failed to meet the requirements of the contract. The contract cancellation clauses come in to effect at this point, and they prohibit you from any further distribution of any GPL'd code. There is no mechanism in the GPL for forfeiture of property rights. If you decide voluntarily to GPL your own sources so that you are again in compliance with the GPL then that is your choice, but is not required by any contract language in the GPL.
The only thing you've lost is the ability to distribute GPL'd code, but since you really never had that right in the first place without complying with the GPL, you have really lost nothing at all.
The worst case scenario is that you never saw or agreed to the GPL, but decided to copy the program anyway even though you were aware that it was covered by copyright. In that case you could have to forfeit the statutory copyright violation fines as well as court and lawyers fees. So far all companies violating the GPL have recognized the folly of this approach and have either voluntarily stopped distributing GPL'd software, or have voluntarily open sourced the software that put them in non-compliance with the GPL.
The original poster claimed that by distributing binary device drivers with Linux that they would eventually lose their IP. If you've kept the sources and the source code is your own property, then nothing in the GPL can make you open up your source code. The GPL can only deny you the right to distribute software covered by the GPL -- you may end up deciding that distributing e.g Linux is more valuable to you than your source code is, or you may not, but the choice is up to you.
The court isn't going to come along and say; hey you distributed your binaries with Linux, they are obviously now GPL'd, so hand over the code.
I haven't seen either of them express an opinion on precisely that topic, but I think that a Judge would have a hard time seeing something like a Tivo as a whole, as a derived work of Linux. Instead it would probably be seen for what it is, a specialized kind of computer with software loaded on it. Likewise, as long as the software that is loaded on the system respects the GPL and the generally accepted GPL barriers, I can't see why they would have any licensing issues.
Indirectly RMS has commented on systems like these as something he would like the next revision of the GPL to address. The copyright that the FSF plans to leverage is the right to control public performance, and probably applies more to web sites than to embedded consumer equipment.
This is the worst sort of paranoid scaremongering. There is no mechanism in the GPL, nor could there be any mechanism, for stealing IP from you. In a worst case scenario you might be barred from, or even fined for distributing other peoples IP without their consent, but your own IP will always remain yours. In practical terms it hasn't ever even been necessary to fine or enjoin the distribution of GPL'd software as all companies so far have been able to come to an agreement with the FSF in mediation.
By only distributing a driver that isn't a derived work, they don't need anyone elses permission to do so.
Unfortunately for them, that isn't true of Linux. Linux is a copyrighted work, and the copyrights are retained by their authors, so nVidia has no right to copy, distribute, modify, or publically perform that work. The authors of Linux will allow people to copy, and distribute Linux under certain conditions and those conditions are spelled out in the GPL, which among other things enumerates certain responsibilities that a company must accept in order to distribute Linux. One of those responsibilities is that Linux must not be linked with proprietary libraries that are distributed under terms more restrictive than the GPL itself. Since nVidia cannot comply with that responsibility, it cannot distribute Linux.
Thus as long as nVidia only distribute their own non-derived drivers they are perfectly within their rights. If they were to open source their drivers then they would be within their rights as granted by the GPL license to also distribute Linux. But as long as the drivers are maintained as a proprietary product that is linked to the GPL'd kernel, they have no right to distribute Linux.
What's the difference between A) Having two pages, one where a person downloads Linux with the source, and another page where one downloads binary-only drivers - and B) A single page where you download Linux with it's source and the binary-only drivers..
It isn't a question of where, but a question of who. The same 'who' cannot offer you both Linux, and these drivers, and remain compliant with both licenses.
In case you are interested, here is some more detail on the mechanics of the GPL. The use of software isn't a right that is reserved to the copyright holder; that is you don't have to contact the copyright holder to run a program, read a book, or any other use that you might want to make of the copy righted item. The GPL does not attempt to force you to accept the GPL until you want to do something that would be a reserved right (like copying and distributing), so it isn't really that nVidia is out of compliance, so much as that they would not be able to comply with the requirements of the GPL if they were to distribute Linux. Also the GPL can be read in a few different ways; in this post I'm representing the GPL the way it is understood by Eben Moglen and the FSF, and as interpreted by Linus Torvalds in their respective writings on the issue. There are other stricter interpretations but it is IMHO scaremongering to talk about those interpretations when the original author of the license is on record with looser interpretations. Likewise there are people who attempt to interpret symbolic linking as distinct from linking and thus a valid barrier between GPL and non-GPL code; while this approach hasn't been tested in court, it goes completely counter to the interpretation of the FSF and thus will land you in legal hot water if you insist on that interpretation.
Actually the GPL states that if you distribute Linux, then you must also distribute the source code to Linux, and anything that was linked with GPL'd software, under a license which is no more restrictive than the GPL. Their code being non-GPL does not give them the right to distribute GPL'd products, only the GPL does that, and they aren't complying with the provisions under which the GPL allows them to distribute that software, so they must not do so.
As well, Linus himself has stated that binary drivers which are linked to the kernel do not "inherit" the GPL. Does that mean they can't package GPLd and non-GPLd code together?
While nVidia's drivers are not covered by the GPL, Linux is. To distribute Linux you must comply with the GPL, and linking binary drivers with the GPL kernel is non-compliant. The GPL does not permit you to distribute GPL'd software if you do not distribute the complete sources, including any code that is linked to the GPL'd code. The only exception provided is for libraries which are a part of the base operating system on which the GPL'd code runs. Since Linux is the operating system, there are no non-GPL'd libraries that Linux links with that are a part of the operating system the Linux runs on.
Plenty of distributions have done this in the past (Redhat has demos of various closed source apps, IIRC, as part of the CD package). The GPL applies only to the code it's been applied to (along with derivative works, etc... and no, "derivative" does not include being packaged together on the same CD).
This is a completely different issue. If you comply with the terms of the GPL license then you have the right to distribute GPL'd software. Since e.g: Netscape doesn't link to the Kernel (it only links to LGPL'd code, not GPL'd code), Netscape and Linux can easily be distributed by the same company (assuming that they have the right to distribute Netscape) since there is no conflict.
If nVidia's drivers were derivative of GPL'd source code then they would have no right to distribute them at all, with or without the Kernel.
That doesn't work so well when the end result is an embedded product and the seller will have to distribute both Linux and the proprietary drivers at the same time.
As soon as you start distributing GPL'd code in addition to your proprietary drivers you will be in trouble. The GPL only allows you to distribute the GPL'd code if any code that you link to it is likewise GPL'd or under an equivalently free license. At that point you are out of compliance with the license, and thus have no right to distribute the GPL'd code.
So I'm sorry, but I agree with your lawyers on this one.
Ah yes, Monday night baby brigade at the Parkway is a blast. No one looks at you funny if your kid is crying (although mostly mine sleeps), and there are couches to sit on instead of chairs if you get in early enough. And the pizzas are priced reasonably as well. On the other hand $5/seat is on the high end for B run theaters, but I'm never aware that they are late in the run anyway since I never get out to the first run theaters.
I haven't read Iain Banks though, and thanks for the recommendation.
What exactly do you consider good character development? Anne McCaffrey? I read Brin specifically because I think his character development is better than most of the other SF authors. The only comparable authors I know of are Larry Niven and Scott Card (and maybe Stephen Brust although I'm not fond of his nihilism.) Benford's science is better but his characters are paper thin. McCaffrey's characters stir up emotion, but she recycles the same character and plot over and over again. Greg Bear is just awful in about every way I can think of. I like Neal Stephenson, but I wouldn't read him for his characters. Pournelle? Brooks? Haldeman? Gibson? Robinson? Bujold?
Only two authors that come to mind are better than Brin in character development, and those are Frank Herbert, and JRR Tolkien. The rest are either thin, carbon copies, or serialized with no transformation from one novel to the next. But if you have recommendations, please tell.