Slashdot Mirror
Torvalds "Pretty Pleased" With Latest GPLv3
Novus Ordo Seclorum writes "According to CNet, Linus Torvalds is 'pretty pleased' with the current GPL v3 draft. He said, 'Unlike the earlier drafts, it at least seems to not sully the good name of the GPL any more.' After his earlier criticism, some had questioned whether such controversies would lead to rifts in the community, especially if the kernel ended up under a different license than the GNU tools. But with the latest revisions, Linus will entertain moving the kernel over to the GPL v3."
Nobody's requested transfer of copyright to any code in the kernel. For the kernel, in practical terms it has never mattered what Torvalds' thoughts on the GPL is, since they'd need individual permission from every contributor to do so (or rewrite the parts that get no permission or where the contributor or their estate recipients can no longer be found). It'd be the mess of mozilla licensing all over again, but even worse.
Trust the Computer. The Computer is your friend.
Hurd has gone nowhere because imho, it lacks a central genius like Linus. BSD gets 1/10 the development effort of Linux (if that). Many developers are willing to work with a BSD license but most aren't. Most people aren't willing to publish their hard work just so some big company can sell it back to them. Without the GPL, Linux would be a poor second cousin to BSD but neither would be as widespread as Linux has become. Both the GPL (Stallman's creation) and Linus were necessary conditions for the success of Linux.
Actually, I would add another real-world genius: Eban Moglen, the Columbia University law professor who is the legal brains behind the GPL.
http://en.wikipedia.org/wiki/Eben_Moglen
What's to persuade anyone to use J. Random Forkoff's kernel, rather than the Linus kernel? FOSS is littered with the corpses of dead forks.
If you were blocking sigs, you wouldn't have to read this.
So, if you were a company that GPLv3 punished, then you'd be punished when dealing with these new kernels, even though most of the kernel didn't have a GPLv3 specific license.
Now, this assumes that Linus wants GPLv3, which so far he does not. If he doesn't want GPLv3, somebody could attempt to sneak in some patches/new code with a GPLv3-only license, and if Linus put them into the kernel, then the kernel would then have the same GPLv3 baggage. But I suspect that Linus would reject any such patches for now, and if one was snuck in, it would probably be removed if found later.
In any event, even if the kernel remains non-GPLv3, we may find some commonly used packages going GPLv3-only -- and I'm thinking of things like gcc, binutils, fileutils, textutils, etc. If this happens (and it sound very likely), then anybody who doesn't want to be restricted by the GPLv3 restrictions will not be able to distribute updated versions of these packages. In the short term, this won't be such a big deal, but in the long term, it certainly will be.
I appreciate what the FSF is trying to do with GPLv3, but I suspect that it's going to cause the `free software movement' a lot of pain, as companies will probably try to move to BSD from Linux (and even then they won't really get away from the GPL, as the BSDs use gcc as their compiler. Perhaps there will be another gcc fork, with the official GPLv3 version and the fork still being GPLv2 or GPLvwhatever?)
That's a pretty slanted view of what happened.
The FSF's issues concerned more than just TiVo, software patents were another major issue. Fundamentally though, the main issue is that there's a lot of licensing forking going on. The GPL is broken. If it wasn't, the MPL, CDDL, and custom licenses for projects like Apache, wouldn't exist.
And the biggest problem with multiple copyleft licenses is that it undermines software freedom. If you can't mix code from one free software project in another, then how are they both "free"?
Critics of the FSF countered that this was all the GPLs fault, because, erm, it's copyleft and the others are n... well, erm, the problem is copyleft, and the GPL invented that, and nobody else wante... oh wait, well, er, RMS is a dirty smelly hippy! Yeah!
The FSF recognized there is an issue, and went forward and tried to create a set of licenses (it's important to note that both the GPL and LGPL are being modified here) that anyone who believes in free software could find common ground with.
Whereupon Torvalds threw a fit, because he'd fucked up. Seriously fucked up. Early on in Linux development, he settled on the GPLv2, but didn't like one commonly included licensing mechanism, the ability to use future versions of the GPL. By itself, that's fine, trusting a third party to always put out fair licenses is a massive mistake, but where Torvalds screwed up was in not replacing it. He just took it out. It's like seeing:
i = int_add_function(i, 1);
in some code, and deciding that it sucks, and it's hideous, and it's really going to have side effects and stuff that are unpredictable, and God knows why someone would put it in, and deciding to remove the damned line instead of replacing it with "i++;"
Essentially Torvalds replaced a clause allowing for future upgrades with nothing whatsoever, which means that it's going to be very, very, very hard indeed to ever upgrade the license of the Linux kernel, no matter how necessary.
So he made up some spurious complaints about the draft. They were nonsense. In some cases his complaints were even that it was somehow a violation of the spirit of the GPL to outlaw ways of making it illegal to modify software (such as use of the DMCA.) The FSF has had to seriously water down one important clause, and rewrite another so it's obvious even to a anti-FSF zealot that Torvalds was full of shit.
After the revision, even Linus has realized that he's going to be laughed at if he makes the same complaints, so now he's trying to look magnanimous while simultaneously dissing it. Yes, contrary to the headline, he's saying he still doesn't like it.
Meanwhile, the rest of the free software (and open source) communities look on with amazement and sadness. Looks like there's less chance than ever of us settling on common hard-copyleft and soft-copyleft licenses for software we feel should be copylefted. Instead it's more likely that GPLv3 will just add to the mess of licenses that aren't compatible with other licenses.
And if Torvalds had just said from the start, "Hey guys, great idea and everything, but just to let you know, I fucked up with the Linux kernel licensing so it's not going to happen with my project", the project itself may, instead of being compromised by the FSF jumping through hoops to satisfy one egomaniac with no great interest in the free software movement to begin with, be the universal set of licenses we wanted to begin with.
You are not alone. This is not normal. None of this is normal.
The question is, if the FSF decides to take a hard stance on this and move their entire toolchain, utilities, and apps to the GPLv3, will their developer base follow them? I personally wouldn't see a reason to, and if I had a FSF-owned project, I would most certainly fork it.
He only f'd up the licensing in YOUR view. Maybe he didn't want the license to change. No added restrictions. I don't consider GPLv3 an 'upgrade'. If you do, great, license your software with it, but don't shove it down the throat of folks who choose not to use it.
> Essentially Torvalds replaced a clause allowing for future upgrades with nothing whatsoever,
> which means that it's going to be very, very, very hard indeed to ever upgrade the license
> of the Linux kernel, no matter how necessary.
Not necessarily. It all depends on how code in the kernel is licensed. There are several files in the kernel that are "GPL 2 or above" and several that are MIT/BSD licensed and several that are LGPL.
Currently, the kernel is "GPL 2 only" because mixing a single "GPL 2 only" file with any of the other licenses mentioned above makes the whole kernel "GPL 2 only".
The key question is: What percentage of the code is GPL 2 only? (I believe LWN.net did an analysis a few months back, but unfortunately I can't find a reference. Does anyone have one?)
If the percentage of GPL 2 only code is small (say 5%) and it's in a noncritical area or can be rewritten quickly or relicensed by the original authors (i.e. they're still around like Linus is) or replaced with other sources like the FreeBSD code or the Solaris kernel (when it goes GPLv3), then changing over to GPL v3 (or at least GPL 2 or above) should be fast.
But even if this were the case, I wouldn't expect any immediate changes. The GPL v3 needs to be out in the field and kernel developers need to feel comfortable with it and see advantages for it (e.g. Solaris-Linux code sharing) before they'd even consider a switch. That could take a few years.
There's two dangers to the "any other version" clause
1) An overly permissive future license allows other people to use and distribute your code in their product without providing source or with restrictions you find repugnant. Not very likely, but consider if the FSF got itself sued for software patent violations or something and Microsoft actually obtained control of it.
2) A more restrictive future license allows other people to use and distribute your code in their product without allowing you to use their code without those new restrictions. This is much more likely.
I don't think it's insane for the FSF to recommend the "and all future versions" clause; they trust themselves, after all. But I don't see why anyone else should.
I think parent's question is why the FSF assumes that everybody who has written a GNU utility and assigned it to the FSF necessarily agrees with the FSF on everything including the revised GPLv3.
I suspect at least a few utility developers agree with Linus that "if it ain't broke, don't fix it - and especially don't fix it in braindead ways".
Since the FSF seems to want to cripple one of the five major Linux distros (SUSE) to "punish" Novell for making a pointless and irrelevant deal with Microsoft (which is unlikely to have ANY significant effect on Linux, OSS OR "free" software now or in the future), it would be amusing if they decided to go that route - and then discovered that half their developers refused to support the FSF anymore, thus damaging the GNU utilities suite's ability to be moved to GPLv3. (I may be wrong, however, in that I think the FSF has full control over all utilities written under their banner, unlike Linux where the individual parts are not so assigned.)
In any event, the willingness to damage one of the major corporate-supported Linux distros merely to "punish" Novell clearly shows the fanaticism of the FSF and Stallman in particular.
Richard Steven Hack - This sig is TOO GODDAMN SHORT TO DO ANYTHING USEFUL WITH! MORONS!
Having volunteered many hours of my time to the FSF, I can say that I'm no stranger to them, and I think GNU has contributed immensely to the goal of software freedom. I think that making Free Software a "movement" was a good idea, and it attracted many people, like myself, who think that we should remove artifical barriers to computation. However, I feel that with the GPLv3, the FSF has strayed from their goal. Let's face it, the GPLv3 puts more restrictions on what you can do with software, not fewer. I am of the opinion that once a piece of software is written, it shouldn't have to be rewritten except to improve or adapt it. GPLv3 forces certain people to rewrite because they aren't ideologically aligned with the FSF.
When I write a piece of software, I want it to be free. That's why I don't use the GPL anymore, and that's why I won't be using the GPLv3 in the future. Presently, I use the ISC license and the WTFPL.
The early versions of Linux, that contained a lot of Linus's code, were absolute crap.
Sure, Linus can write code, but he can't engineer his way out of a wet paper bag.
But he's a great manager, and Linux succeeded due to Linus's leadership abilities,
not due to his engineering abilities.
Stallman, on the other hand, is a great engineer.
Emacs and GCC are testiments to this fact.
But Stallman is a lousy manager, based on what I've read of the history of the Emacs project.
Doug Moen
I have written a truly remarkable program which this sig is too small to contain.