Slashdot Mirror
Why Torvalds is Sitting out the GPLv3 Process
lisah writes "Linus Torvalds has a lot of reasons for not wanting to participate in drafting the third version of the GNU General Public License (GPL): He doesn't like meetings, says committees don't make sense, has philosophical differences with the Free Software Foundation, and seems to be generally distrustful of the whole drafting process. Though Torvalds prefers the GPLv2, he says if others prefer the GPLv3, they ought to support it because 'it's not like it kills and eats small children for breakfast, and must never be allowed.'" Linux.com and Slashdot are both owned by OSTG.
If you want to contribute to the GPLv3, you can. The FFII, for example, proposed some changes that would clarify the GPLv3 with respect to patent law in Europe (the current draft is too US-biased).
Torvalds doesn't need to contribute, but I'm glad he's moved to a more neutral stance. The GPLv2 is old and out of date and though it still works today, will start to crumble in a few years.
In every new project my firm does, we end up adding our own conditions onto the GPL3 (for instance for patents) and it'd be far better to have these defined as standard.
It's good to be critical of processes that aren't clear, and it's entirely possible that the FSF won't be able to produce a worthy successor to GPLv2, which is an incredibly important document in the history of software, but we should give them the benefit of the doubt.
My blog
FTFA...
For Torvalds, the controversy over the different versions of the GPL is ultimately very simple: If "I can just go back to 1992, when I relicensed Linux under the GPLv2, and ask myself: If I had the choice of licenses back then that I have today (including the GPL3 draft), which one would I have chosen? And the answer simply isn't the GPLv3. It might have been the Open Software License, though. But, most likely, it would still be the GPLv2."
At heart, he's a developer. He'd rather be coding or debugging than getting involved in legal debate. And that's a good thing for us. I'd much rather him spend an hour working on Linux than disputing some clause of some license. It's just a more productive use of his time.
for the kills small children and eats them for lunch license. Dinner would be OK, too, but not breakfast.
"To those who are overly cautious, everything is impossible. "
Torvalds may not like the GPLv3. However, I think that is orthogonal to why he is sitting out the process. At heart, the man is an engineer/coder. How many people work as software engineers/programmers/code monkeys/whatever and jump at the shot to sit in the "politcal" meetings? Seriously. As a general rule, engineers and programmers would rather be engineering and programming. They don't care so much about marketing. They don't care so much about the political undercurrents of the organization. They just want to do their job well.
Have a chair.
I have actively participated in the debate about the clause 0 defect of the GPLv2 ("that is to say...") and I became satisfied that it is in good shape for the GPLv3.
It's better to be the foot on the boot than the face on the pavement. ~~ tkx Kadin2048
Your analogy makes no sense.
Forking the FSF license creation process is not like forking Linux; it would be like forking the Linux development model, which is equally impossible.
Forking Linux would be like forking the GPL itself, which is not only possible, but trivially easy: all you have to do is re-write it however you like, and rename it (e.g., "ACPL," for "Anonymous Coward's Public License").
"[Regarding the 'cloud,'] ownership was what made America different than Russia." -- Woz
For those that didn't see it (because my submission to slashdot was rejected, between other reasons), An Ode to GPLv2:
"One of the reasons I didn't end up signing the GPLv3 position statement that James posted (and others had signed up for), was that a few weeks ago I had signed up for writing another kind of statement entirely: not so much about why I dislike the GPLv3, but why I think the GPLv2 is so great.
Rest of the post
Second, ability to run the program, but not see the source code. Case in point, Google. It is beyond question that Google are using all kinds of GPL applications, from the kernel to webservers to highly modified filesystem drivers. All of it GPLed and none of the code available for you to see, despite the fact that Google allow you to use all these services online, you'll never see a line of the modified code.
Both these cases violate not the letter of the GPLv2 licence, but the spirit of it. That spirit being the ability to run the program, modify the source, and run the changed program. This is happening on small scales today. It could soon be happening on a huge scale, and that would undermine the whole FOSS community. GPLv3 will be needed in the future.
May the Maths Be with you!
Linus doesn't have a choice in the matter: since the kernel is under GPLv2 without an "or later" clause, he can't change it. The Linux kernel is stuck with GPLv2. For him to argue one way or the other is pointless and sounds like post-hoc rationalizing.
Personally, I think the GPLv2 will sooner or later kill the Linux kernel. Some highly successful embedded Linux systems like the WRT54G only became hackable because the manufacturers made a mistake. Evidently, embedded users of Linux just don't get the benefits of openness, and they'll get better and better at circumventing the GPLv2; the GPLv2 will turn more and more into a kind of encumbered BSD license, and you can see how well BSD did with that.
Of course, I'm not too concerned. I think we really need a successor to the Linux kernel anyway, yet the industry is happy to keep running a 30 year old kernel design. If being increasingly the target of GPL circumvention is what it takes to motivate people to move to a new kernel, that's fine with me, too.
I recognize the importance of software licenses, especially the GPL, but I've come to the inescapable conclusion that many of the major players in OSS have an insatiable need to spend enormous amounts of time bickering about licensing minutiae.
Every week brings a new drama-bomb in the endless pissing contests and personal rivalries/vendettas. If half the energy expended to one up, or argue with another developer was put into the development process, an untold number of projects might be a bit further along. One thing you can say about closed-source software is that the financial pressures end up stifling a great deal of the petty childishness that seems to pervade the OSS community, and taints its image in the process.
Don't get me wrong, you still get this sort of crap on the closed-source side of things - "I don't want to use your standard...I want to reinvent the wheel for this app..." etc, but it's not at the forefront. Human nature dictates that you will find these problems everywhere, but in the corporate, closed-source enviroment, it comes down to one conclusion - eventually the project needs to get done.
If OSS wants to gain more acceptance, it needs to put this sort of thing aside and get back to the core issue - it's the code, dammit. None of the present issues with the community are insurmountable, but direct action needs to be taken, these problems are not going to going away on their own. Rampant egoism, Not Invented Here Syndrome, coder-centric, not user-centric development methodologies...these all slow the pace of progress and paint open source in a very bad light.
OSS has a large community of smart people, and I just think it can do a whole lot better.
I think Linus's difference with the FSF is quite simple.
The FSF is concerned with users. The whole thing started when Richard Stallman couldn't fix the printer driver that he was a user of. The FSF's goal is to ensure that everyone who uses software, ever, has the technical and legal right to modify the software they are using.
Linus seems more concerned with developers. If someone comes along and contributes some sweet code to the Linux kernel, he thinks it's only fair that any developer gets the opportunity to use that code too, in their own project. But he's not concerned that an end user can't install a modified version of Linux on their Tivo.
Linus has such a dislike for the FSF that he rants on these things that he doesn't even know about and what's worse, uses his position to spread his ignorance like a cancer, a malignant ignorance. Consider that he did not even know the 'meetings' took place over email and IRC. Or his repeated claims of having to give up his private key, which is shown wrong over and over by legal experts. Or saying committees don't take responsibility for decisions and then complaining that they didn't just blindly agree to whatever his kernel developers wanted.
What's interesting to me is when Torvolds says the GPL2 is where companies and open source people can meet in perfect harmony, as if companies like the GPL2. No company likes the prospect of having to open up their product because some 'tard put in GPL code without their knowledge. They put up with it because they have to, because it's a reality they can't escape. I know I have had many heated arguments about making code GPL when others on a project wanted BSD to be more 'corporate friendly'. Perfect harmony? Wtf world is he living in? Use GPLv3 and they will come and work with that too (even though they don't want to) and for the same reasons.
I think the real question is, as an open-source developer, why wouldn't you choose GPLv3 over v2? Because you want some company to use your program and then sue you because you made use of their patents? Or you want your software to make DRM devices cheaper to create? Or you want your license to be worded in a way that is ambiguous in some regions? I wonder why Linus wants linux to be licensed without patent protections, with ambiguous language, and in a way that supports DRM?
I keep going over and over this, and I still can't figure out why Linus would want Linux to be able to be Tivo-ized, but not want it BSD-licensed. Can you explain to me what it is about these specific loopholes that makes them so much more desirable than people taking your code wholesale and making it into a proprietary program?
Don't thank God, thank a doctor!
As an engineer at heart, he understands computers and software systems very well, but likely avoids and despises legal systems.
Or, put simpler: I think he simply doesn't understand it. And yes, I know that sounds arrogant, but if you remember his posts on Groklaw, he demonstrated again and again that he thought the GPLv3 demanded things that it didn't, and that he had completely missed the point of what it's actually trying to do. For instance, he actually brought out that old FUD about how disabling DRM will prevent certain security measures, which it doesn't.
I don't think Linus and PJ actually disagree, but I do think PJ actually knows her stuff, and Linus should stick to the actual coding, organizing, and benevolent dictating of the kernel itself.
That, or sometime fairly soon, we're going to actually squeeze a statement from Linus that, given the choice, he'd go with BSD or public domain. They seem more in line with his ideals.
Don't thank God, thank a doctor!
about "free software". What they are truly fundamentally about is
creating a comprehensive category of software which is completely free from
corporate/business control, and which individual users can completely control in
all aspects as they wish.
His fundamental motivation is an anti-corporation, pro-individual/community
point of view. The fact that the mechanism for enabling his version of
"free software" is the GPL and a common pool of open source is
secondary. If he could have gotten a global law enforced that all corporations
must release all their source code freely on the Internet, that's what he would
have done, instead of GNU and GPL.
RMS is an absolutist on this point. He truly sees this as good vs. evil, and as
a belief system about which there can be no question.
To help understand this, http://www.zmag.org/content/showarticle.cfm?ItemI
read this interview.
This is where the insistence that DRM and "Trusted Computing" and
software patents must be abolished comes from. These are all tools that
corporations use to protect their property. RMS does not believe they should
have property like this... that it should all be made available to users with no
control by corporations.
Linux is also licensed under the GPL (v2), but comes from a completely
different motivation than RMS. Torvalds simply believes the open-source
development model is the most effective way to create excellent software.
Torvalds is just fine with corporations and businesses using Linux for profit,
even if that means "controlling" some aspects of its use. He
certainly has opinions on DRM, patents, and "Trusted Computing", but
he's not going to let those get in the way of Linux development.
So now starts the struggle for control of "what is the meaning of free
software". RMS is clearly trying to re-establish his vision of the
principles involved by pushing through GPL v3, because he's seen GPL v2 used in
ways that offend his principles deeply. Is it too late? Has the FOSS movement
taken off to an extent that he no longer controls it? Stay tuned.
Simply with BSD-licensed code you don't have to give your changes back, but with GPL v2 you have to Tivio or not. And that's the whole difference, simply getting the code back.
I can't tell you what Linus is thinking, but I can tell you why I think that way.
/seeing the code/. I care that I can then use that code (or more likely ideas and tricks from that code) in my own projects. I don't care about making my consumer-grade router outpace the Cisco gear I use at work. I care about being able to make my own software on par with IOS.
/learn/ from that system that I want. Yes, learning could perhaps be easier if I could run modified code on the device, but ultimately, simply having access to the source is what I really care about.
I am a programmer. I am not a tinkerer. I care about
The ability to tinker with a system just isn't that important to me. It's the ability to
It's always a long day... 86400 doesn't fit into a short.
It's true. Engineers, scientists, programmers, mathmeticians, etc, would rather engineer than participate in meetings and organizational politics. Often, this is accompanied by an inability to play well with others--which I suspect is the case in this instance.
There are so many cases on the record where LT beats a hasty retreat after his arguments are demonstrated to have poor logic. Let's hope LT learns to moderate his penchant for hyperbole. Let's all be glad he codes better than he discusses policy.
First of all, I'm not speaking for the FSF.
Second, I have no doubt that they're trying to accomodate everyone as much as possible. However, they're not about to do something completely contrary to their stated goal, which is to make software that's free for the user. Fundamentally, the GPL exists to serve the FSF's goals; therefore, no matter how touchy-feely you try to make the process, the bottom line is that it's going to be what the FSF wants.
And before you complain about this, think for a minute and you'll realize that it's the same for every human organization, from the US Government to the Linux kernel to Bob's Fine China and Firearm Emporium, Inc. Deal with it.
"[Regarding the 'cloud,'] ownership was what made America different than Russia." -- Woz
But you still need a device to run the code on. Consider the fact that in many cases (especially embedded) there isn't a good substitute device. What good does having the code do you in this case? What benefit do you get out of the GPL that you wouldn't get out of, say, MS's "shared source" licenses?
More importantly, why should the hardware device maker get the benefit of the GPL for themselves, without having to give back? I mean, let's assume you like the GPL in the first place. In that case, you already belive it's reasonable to require software writers to open the code they write that's combined with GPL code, because being able to use the combined whole is the point. So, similarly, isn't it reasonable to require hardware makers to open the hardware they make that's combined with GPL code, because using the combined whole is still the point?
"[Regarding the 'cloud,'] ownership was what made America different than Russia." -- Woz
I keep going over and over this, and I still can't figure out why Linus would want Linux to be able to be Tivo-ized, but not want it BSD-licensed. Can you explain to me what it is about these specific loopholes that makes them so much more desirable than people taking your code wholesale and making it into a proprietary program?
With GPLv2, people who take your code and alter it have to publish the alterations. This adds to the store of knowledge generally available to the human race. Good ideas that improve your code can be incorporated into your own project or into others. This doesn't happen with a BSD license.
And again, the problem is that there may not be a substitute device. That is, there could be amazing, incredible innovation in some GPL'd code, that would be utterly useless to you without an open mp3 player. Now, that case is actually irrelevant now -- there's a player that encourages rockbox, and you can make Linux run on an iPod -- but it's still a valid point.
If you look at your history, I think RMS will back me up here. The whole free software movement was inspired by a printer driver without source code -- not because RMS particularly wanted to see how it worked, but because it didn't work, and he wanted to fix it.
Alright, then. I still think it's asinine of people to lock down a device so I can't run custom software on it. It is their right to develop such a device, but I do not want to help them, so I don't think it's their right to use my code in such a device.
Don't thank God, thank a doctor!
Linus, you're never going to have a successful career in politics with an attitude like that.
When all you have is an axe, everything looks like a grindstone.
Linus just doesn't care all that much. He hasn't learned the proper lesson from the bitkeeper incident yet apparently.
evil is as evil does
it's not like it kills and eats small children for breakfast
GPL3 may not look like that, but Stallman does!!
Software patents are not valid in europe - now why would anyone on this side of the atlantic want to license anything under GPLv3?
Does anyone know if software patents are valid in Asia?
What is stopping you compiling it and running it ?
Your hardware doesn't run (by design) programs that you compile ?
Well, get some that does then. Why did you get non-user-modifiable hardware if you wanted to modify it ?
I can't modify the software on (for one example) my phone, do I care ? no. Because if I _wanted_ to modify it, I'd have bought a phone which supported me modifying it.
You can take every modification Tivo has made to GPL software and use it in your own PVR (or in something completely different) - even compete with Tivo if you wish, and even if you eat their lunch in the marketplace, they _still_ have to give you every improvement they make to the GPL software. You can benefit from _all_ their GPL software R&D, for free.
Provided, of course, that you reciprocate, so they benefit from all yours too (_that_ is the core of the GPL - nothing to do with hardware).
So, why is this code not useful ? It isn't useful to me, as I don't want to build a PVR, but there's loads of GPL code that I have no use for - that doesn't mean it isn't useful.
That doesn't make any sense unless you selectively interpret it to fit what you want it to. There's no Freedom guaranteeing that a modified program will run on the original hardware. For one thing, the modified code might be buggy. For another, it might be too large. For a third, the hardware might be buggy.
Running modified code on the original hardware is convenient. It's probably what you want to do. It might even have been your inital goal. But calling it a "freedom" is just silly.
It's always a long day... 86400 doesn't fit into a short.
This may be true with all methods of machine counts that you're aware of, but that's the ludicrous statement of a luddite to claim that no machine count can ever be trusted. Humans are much more easily bribed and manipulated than machines. Humans from both parties.
It costs more money, though, and it does take some amount of time more. That said, I was pretty annoyed that Kerry conceded before the votes were actually counted.
Quick Google search shows population of Australia to be 20,090,437. Population of US is 295,734,134. That's a significant difference. Now, if I could find out how many actual votes were cast...
Which is also thousands of opportunities to make mistakes, intentional or not.
Oh, I agree, but it would be nice if we could have neither the delay nor the fraud.
Aside from the sheer cost, time, number of volunteers, and potential innaccuracy... No, I don't think electronic voting is wholly uncalled for. I do think that I don't know strong enough words for the level of negligence with which the US has treated voting in general -- from the chads and the butterfly ballot to the Diebold machines.
It's been said over and over again -- we treat our electronic slot machines and poker machines with much more scrutiny than we treat our voting machines. Frankly, I'd feel better trying to win the jackpot from a slot machine than simply trying to have my vote counted by a Diebold machine.
Don't thank God, thank a doctor!
that might be true if everyone voted, but not when you don't have compulsory voting...and not when you don't even get close to 100% turnout of those who are registered to vote.
no, it's not a luddite statement. it's an analysis of the risks vs the benefits. an election is FAR more important than a personal signature, it matters a lot more whether it is fraudulent or not because the scale of damage is far greater.
and, just as significantly, moving from written signatures to electronic signatures isn't a significant reduction in the security of the system because written signatures aren't very secure to start with. by contrast, moving from a many-eyes manual count to an electronic count is a massive reduction in the security of the system.
the miniscule benefits of electronic counting are greatly outweighed by the risks.
because they're going to be affected by the result of the election, regardless of whether they supported the winner or not. the winning candidate is supposed to represent *everyone* in their electorate whether they voted for them or not, they are elected to do a job - to represent the people in their area. it is a peculiarly american attitude that somehow you don't count, your right to be represented is void if you don't vote, or if you didn't vote for the winning candidate.
and, no, the extremists don't know what they're voting for either. for the most part, they're just like football fans voting for their side, regardless of what their policies are. non-compulsory voting means that voting is done by those most sucked in by the hype.
in practice, with compulsory voting, what happens most of the time is that people vote for those who either a) promise to have the least damaging policies, and/or b) offer the most to them (e.g. hospitals, schools, whatever other things are needed in the local community). there's certainly potential for problems here, but it's far better than candidates simply ignoring what the majority of people in their electorate want, to concentrate exclusively on what the nutters and corporate interests want (i.e. appease the nutters to get their vote, and do what their corporate masters tell them to do).