Not logged in
Log in now
Create an account
Subscribe to LWN
LWN.net Weekly Edition for May 23, 2013
An "enum" for Python 3
An unexpected perf feature
LWN.net Weekly Edition for May 16, 2013
A look at the PyPy 2.0 release
People provide exceptions for this kind of thing all the time.
Grand, but be clear: then we're no longer talking about a GPL licence.
Open is Open.
Posted Nov 10, 2005 23:30 UTC (Thu) by bojan (subscriber, #14302)
Of course we are. For instance, here is a part of the text from such an exception (see: http://www.gnu.org/cgi-bin/license-quiz.cgi for an example use, as written by FSF):
If you modify this file, you may extend these exceptions to your version of the file, but you are not obligated to do so. If you do not wish to do so, delete one or both of these exception statements from your version.
So, the recipient of the software can relicense the software under the pure GPL if he/she so wishes. But the point remains - this software can be included, without any problems, in _any_ software licenced under the _pure_ GPL. In effect, it is licensed under the GPL.
PS. This text is from the copyright statement of my own software, so I didn't just pull it out of thin air.
Posted Nov 11, 2005 9:22 UTC (Fri) by paulj (subscriber, #341)
Of course we are.
No.. You're talking about GPL + exceptions. Not at all disputing the validity of exceptions (I think I mentioned in another post the ability for GPL authors to give themselves exceptions so they could link to CDDL code).
However it is still not quite the GPL, it's "GPL + X". It's disingenuous to reply to me and claim that Sun could have chosen the GPL, when really you mean "GPL + X". Further, the "+ X" part would be the "specify patent pool and MAD on litigation", which would apply additional restrictions on patent rights not specified in the GPL. Hence that would make this "GPL + X" licence incompatible with the GPL.
So your suggestion essentially is no different to Sun having used the CDDL.
Posted Nov 12, 2005 3:50 UTC (Sat) by bojan (subscriber, #14302)
On the contrary. My suggestion is _qualitatively_ different to what Sun have done. It is _illegal_ to include CDDL licensed code with the GPL licensed software. It is perfectly legal to do so with the code licensed under the GPL + Exception. And that is because GPL + Exception _is_ essentially GPL, for people that choose not to pass on the exceptions. Nothing like that is possible with the CDDL.
As for patents, I addressed that before to a degree, but I'll elaborate. Bottom line is this: Sun are unwilling to guarantee (in any legal sense, like IBM) that they won't sue open source developers over patents Sun holds. In other words, I can get in trouble for using patents from OpenSolaris in my open source software. I cannot get in trouble for using IBM's patents they promised not to sue over. So, I think this behaviour on behalf of Sun is more disingenuous that anything I wrote here, particularly coming from the company that supposedly not only understands but also "invented" open source (luckily, they didn't patent that invention :-).
We can all pretend here, but it is quite clear that CDDL was designed and chosen _precisely_ because it isn't compatible with the GPL, in order to prevent Solaris technology from going to Linux (which is likely, given the size of Linux v. OpenSolaris development communities). Ditto the patent promises, which only apply to CDDL licensed software. I just wish Sun were open about it and told us so in the first place.
I did watch the launch of Galaxy servers where Red Hat (and by extension Linux) supposedly became Sun's friend again. I just wonder why that was? Could it possibly have anything to do with the fact that most Sun AMD based servers run an operating system other than Solaris? Or was it a sign or "real" friendship? Hmm, I'm inclined to go with the former, but that's just my cynical self ;-)
Posted Nov 12, 2005 15:41 UTC (Sat) by paulj (subscriber, #341)
On the contrary. My suggestion is _qualitatively_ different to what Sun have done. It is _illegal_ to include CDDL licensed code with the GPL licensed software. It is perfectly legal to do so with the code licensed under the GPL + Exception. And that is because GPL + Exception _is_ essentially GPL, for people that choose not to pass on the exceptions. Nothing like that is possible with the CDDL
This is about the patent-pool, patent MAD clauses of the CDDL. The problem is that if you add those to the GPL, as you suggest, then they are not exceptions but additional restrictions. Additional restrictions would make this theoretical "Sun GPL" incompatible with the GPL anyway. Authors of GPL software *can* make use of CDDL patent grants, just by giving themselves to link to the CDDL software to which the grant applies.
Bottom line is this: Sun are unwilling to guarantee (in any legal sense, like IBM) that they won't sue open source developers over patents Sun holds.
Ha, ha. Go and look at the patents IBM donated to the Linux *kernel* (not to opensource in general btw). I'm sure the linux kernel is far safer now that IBM won't sue it for things like patents on gel-packs, and such.
Sun executives BTW are on record that Suns' interest in patents is primarily defensive. And the patents Sun has given Free Software (which is what code under the CDDL is, according to *RMS*) are those relevant to the code at hand. Unlike IBMs publicity stunt.
We can all pretend here, but it is quite clear that CDDL was designed and chosen _precisely_ because it isn't compatible with the GPL, in order to prevent Solaris technology from going to Linux (which is likely, given the size of Linux v. OpenSolaris development communities).
Your powers of mind-reading are obviously far greater than mine. I don't see how that is clear at all. I see it as a possibility, a very unlikely possibility given various other factors, as I've already explained. Eg. that the CDDL simply has restrictions in it which can never be GPLv2 compatible (but maybe could be GPLv3 compatible). Or e.g. that OpenSolaris and Linux kernel code simply is *way* too different for code to transfer across anyway. As for OpenSolaris apps, you can port those across no problem, CDDL licence has no problem with that. If you want to implement DTrace on some other kernel, you're free to use the OpenSolaris Dtrace userland tools. (Someone is working on this for FreeBSD btw).
Are there reasonable things which you could criticise Sun for? Sure. Are there omissions or deficiencies in the CDDL, some claim there are (see other threads). I'm pretty sure Sun and the OpenSolaris community would like to hear them (join the lists), they might even be fixable if valid.
But if your arguments start to reduce down to mind-reading, I'll have to bow out I'm afraid.
Posted Nov 13, 2005 3:03 UTC (Sun) by bojan (subscriber, #14302)
I never suggested _that_. I said that Sun could have licensed with GPL + Exception for _linking_ with other software. I addressed patents before (twice) and I won't do it again.
> donated to the Linux *kernel* (not to opensource in general btw)
Please get your facts straight. IBM _explicitly_ promised not to sue _any_ open source project over about 500 patents they have (applies to all OSI approved licences). Not just the kernel.
> Sun executives BTW are on record that Suns' interest in patents is primarily defensive.
Which is legally meaningless.
> Unlike IBMs publicity stunt.
> Your powers of mind-reading are obviously far greater than mine.
I studied under Yoda himself ;-)
Posted Nov 13, 2005 6:42 UTC (Sun) by paulj (subscriber, #341)
I never suggested _that_. I said that Sun could have licensed with GPL + Exception for _linking_ with other software.
Ok, fair enough. However, all my arguments here before about why the GPL would not have sufficed have revolved mostly around the patent pool+MAD clauses of the CDDL, linkage was just another example. Yes, the linkage thing possibly could be dealt with via exceptions. My point regarding the patent clauses stands though.
Ah, fair enough. I must have been confused. However, go and *look* at the list of patents. Quite a few have absolutely no relevance to software (CPU, token ring, ATM, etc. hardware patents), others do not have much value (though, that's probably more a comment on the state of the patent system) and at least two of the patents don't have anything to do with computers (tamper proof screws, DNA reading assembly).
A unilateral patent grant still does not do anything to try setup a patent pool and patent "peace" around a software project, as the CDDL does. As I said in another post, be careful what you criticise the CDDL for, when there's a fair chance the GPLv3 will try do similar tomorrow (Eben Moglen has stated patent threats are something the GPLv3 may try address). The CDDL's clauses are intended to encourage the build-up of a pool of grants to a project on all relevent patents held by the initial and contributing developers. That's a worthwhile goal IMHO.
IBM have *many* more patents than those 500. They can almost certainly dig up patents from their portfolio to sue almost anyone they wanted to, regardless of those 500. So their grant is not *that* meaningful either. So lets be honest: The primary reasons big corporates involved fruitfully in Free Software, like IBM and Sun, neither have nor likely will (in the short to mid-term at least) sue any Free Software developers or projects is because they have absolutely nothing to gain from it and a *lot* to lose.
Anyway, did Sun want to do something different than the GPL? Yes, I've listed some *good* reasons why. Disagreeing with the logic of those reasons is fine. However clinging to the view that Sun chose the CDDL out of sheer perverseness (reasoning I can never really refute, since it hinges implicitely on Sun having hidden motives), when it isn't that hard to come up with more rational reasons is perverse in itself, if not sheer paranoia. Particularly when the objectionable result of this "hidden motive" theory, incompatibility with the GPL, lies *not* in the CDDL but in the GPL and can easily be fixed in the next revision of the GPL, which is due in the next year or two. Further, any GPLv2 copyright holder can *today* avail of CDDL code by simply giving their own GPL code an exception. If you have suggestions on improving the CDDL, you are welcome to discuss them on the opensolaris-discuss list, the CDDL *can* be up-revved too.
I studied under Yoda himself ;-)
Posted Nov 13, 2005 8:31 UTC (Sun) by cventers (subscriber, #31465)
Posted Nov 13, 2005 19:26 UTC (Sun) by bojan (subscriber, #14302)
I actually agree with you here (and I'm not saying this to be sarcastic). I don't trust Sun after all the flip-flopping they've gone through in regards to open source and Linux. And it's all right there in Jonathan's blog (if he doesn't speak for the company, I don't know who does). All this change of heart makes me nervous. Others, particularly IBM (note: I don't work for IBM, don't own shares of it, never purchased a computer from them and where I work the whole organisation has zero IBM equipment) have been more straightforward, while still quite openly supporting other (proprietary) things and generally hedging their bets, but without "change of heart" every second day.
I would say that possibly the biggest trust busting factor of all is Java for me. If there ever was a vibrant open source effort, it is around Java (e.g. project Jakarta). Some of the best Java stuff out there is open source. And yet, I cannot find any _valid_ reason for Java to be closed source. I know Sun exectives quoted compatibility, but that is just so easy to achieve through tradmarks, that is completely unbelievable. Unfortunately there is no scientific way (yet) to measure what would happen to the platform if it were open sourced, but I think that mere fact that it would get included as a standard part of evey Linux distribution speaks for itself, not to mention all the JDK bugs that have been outstanding for months (in cases years) being fixed quickly. Why Sun doesn't want that to happen, I don't understand.
Posted Nov 18, 2005 3:37 UTC (Fri) by robilad (guest, #27163)
Posted Nov 13, 2005 6:23 UTC (Sun) by bojan (subscriber, #14302)
Just one more comment here, if I may. You seem to be reading too much of Jonathan's blog and therefore mocking all of those patents. The patents IBM pledged to OSS are here:
To the majority of open source developers, if only one of those patents is something an OSS developer would ever use, IBM's pledge is just about infinitely better than whatever Sun promised in relation to OpenSolaris. That's because Sun's promise relates to (mosly likely unmodified) OpenSolaris code only.
Copyright © 2013, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds