Messages in this thread | | | Date | Tue, 9 Nov 1999 18:34:45 -0500 (EST) | From | "Mike A. Harris" <> | Subject | Re: Reiserfs licencing - possible GPL conflict? |
| |
On Tue, 9 Nov 1999, Matthew Kirkwood wrote:
>> ---- Reiserfs licence ------------------------------------------ >> Reiserfs is hereby licensed according to the Gnu Public License, >> but with the following special terms: you may not integrate it >> into any kernel (or if not added to a kernel, into any software >> system) which is not also a GPL kernel (software system) without >> obtaining from Hans Reiser an exception to this license. > >There is no GPL conflict here. It's Hans' code, so he may choose >its licence. Since this licence is trivially GPL compatible, there >is not problem with including it with Linux.
Perhaps not a conflict, but why guess... fix the wording so it is not ambiguous...
>> The text above from the reiserfs readme file states "Gnu public >> licence". Strictly speaking - legally - no such licence exists to my >> knowledge. A court of law would probably not equate "GNU public >> licence" to be the same as "GNU General Public Licence", and as such I >> would like to see this clarified if possible. > >Of course it would. Pedantic correctness is even better, but courts of >law are not, in general, sufficiently stupid that they would actually >fail to protect reiserfs on the basis of this.
Perhaps. There is no guarantee however. Why play with assumptions. I agree with you that it would _probably_ be interpreted as per the intention of the author. By rewording it however, ambiguity is removed and no interpretation or assumption is needed. This is my intention.
>> The GPL is referenced later in the statement, which makes me believe >> that reiserfs is in fact intended to be under the GPL licence. >> However, the first paragraph goes on to mention what appear to be >> "further restrictions" which are explicitly prohibited by the GPL >> licence. > >No. The GPL disallows _downstream_ recipients from imposing further >restrictions. Since Hans' (and others? Who owns Chris' code?) is the >copyright holder, he may apply whatever terms he chooses to the code. > >Look at it not as the more common "GPL, but we'll sell you exceptions" >but as "Modified GPL". In this case, the difference is small, but enough >to cause your concern and thus, perhaps, worthy of clarification.
The GPL is the GPL. If you solely create a piece of software, and licence it under GPL, and then in the "licence" file, state that "this program is licenced under the GNU GPL, with the following restrictions - you may not do this, you may not do that" - then YOU the AUTHOR are violating the GPL yourself. Your work is NOT protected by the official GPL if you modify it. If you modify the licence text, or impose restrictions - which you are more than free to do - being the copyright holder, then you MUST change the NAME OF THE LICENCE to something other than General Public Licence, because the GPL licence itself is copyright by Richard Stallman I believe. Modifications to the GPL licence can only be done by RMS or FSF I believe.
At any rate, who cares. The point is reiserfs licence needs rewording to be unambiguous.
>> Further inspection seems to show that the "restrictions" are not >> really so - even though worded as such, because any GPL work must >> remain GPL anyways. So basically the "following special terms" are >> not necessary to begin with because the GNU GPL allready explicitly >> forbids inclusion in a non GPL work. > >No it doesn't. In practical terms yes, that is its effect, however, >there are many cases where it is possible to link GPL code with code >under another licence, provided that _the union of restrictions from >both licences is no larger than those imposed by the GPL_.
That is more or less correct, however, back to the point - which is that the licence is ambiguous, and can be fixed in a few minutes time to not be so.
>> One other thing: Hans, et al. are perfectly free to licence their >> code under any number of different licencing schemes, as stated more >> or less in the above blurb from the readme. This is perfectly ok, and >> doesn't in any way muck with GPL issues. >> >> What is certainly unclear however is WHAT the EXACT licencing of this >> filesystem code is, and not in writing that is open to interpretation. > >reiserfs is not under the GPL. It is under a licence slightly more >restrictive than the GPL (one couldn't link it with a kernel under the >free-BSD licence, or the MIT licence, where GPL software in general is >compatible).
This PROVES my point. The licence that reiserfs is under IS AMBIGUOUS. I read it, and see the author's intention to be that it IS under the GNU GPL licence, however you state that it is NOT under the GPL, but rather one that is more restrictive.
Thus 2 different people have interpreted the author's licencing statement COMPLETELY differently, and thus the licencing statement needs to be cleared up properly. That is my point entirely.
>Actually, this has potential bad interactions with the kernel - if >reiserfs is linked with the kernel then _as a whole_ the kernel's licence >become more restrictive than the GPLs, and thus reiserfs can no longer >be linked with it. I think that perhaps Hans, Linus and some lawyers >could do to discuss this further.
That may or may not be necessary. If Hans intention is to have the Linux reiserfs code be licenced under GPL, and simultaneously available under an alternative licence, then the GPL licenced reiserfs does not violate anything at all. It just needs to be stated more clearly, and include an actual copy of the legalese of the licence.
If however, reiserfs under alternative licencing from the author, is used with a Linux kernel, then there is definitely a problem. Anyone distributing executable code such as a linux kernel, which has been built from the GPL Linux kernel sources, and with reiserfs (or any other addon code) which is licenced under a more restrictive licence, or closed-source, then that particular distribution of code or executable violates the licence indeed.
The only legal way that "may" be possible in this case is code that is standalone from the kernel source, and does not modify any kernel source files whatsoever, and which compiles cleanly without incorporating any kernel code internally or linking with it. If this were made available as a binary only module, then it would be sufficiently isolated. I seem to remember something about binary only addons to the kernel may or may not be "ok" though. If it is 100% standalone, then it is ok.
I beleive it is stated that if the purpose of a GPL'd driver is to export internal kernel stuff to a proprietary binary only driver or userland app, that this violates the kernel licence. People could just write a minimal interface, GPL it, stick it in the kernel or make it freely avail, and fully GPL compliant, but which the purpose of is just to use their proprietary binary only code with, and "bypass" the GPL restriction - then this is not allowed. I believe the MOSIX people tried to do something like this before...
>> [LICENCE] This software is hereby licenced under the GNU General >> Public Licence version 2 or later. Please see the file "COPYING" >> which should have accompanied this software distribution for details >> of that licence. >> >> Further licencing options are available for commercial and/or other >> interests directly from the author at: <email address> > >The issue here is that this is in no way the intent of Hans' licence. >Otherwise, some unscrupulous company could link it with FreeBSD, turn the >prouct into a NetApp filer-style appliance (say) and give back only the >FreeBSD linkage. Which the FreeBSD people would have little use for >anyway, since they don't like GPL code in their core. > >It seems that tyou have spotted a problem - not with Hans' licence (which >he is entitled to choose) but with its potential linkage with the Linux >kernel.
Hmm... That may be true, but I didn't realize that until now. I hope that reiserfs indeed is GPL compliant, and that the licence wording will be corrected.
>Personally, I think that Hans' licence unifies the intent of the GPL with >some fairly practical modern additions, and would have no particular >problem seeing the whole of the kernel under these conditions.
I think once the licence is corrected (if it is - and I think it was more of an oversight and will be fixed), then the only possible violation that I can see, would be if someone contacted Hans for an alternative licence to the code, and then linked that particular bit of code to the Linux kernel source. That would violate the GPL that the kernel itself is under because the combined kernel binary could not be re-released as a derivitive work under GPL because the reiserfs code licence that was "arranged" would not be GPL compatible ( or would it?) I guess it would depend on the wording of that licence too.
At any rate, some lawyers should help Hans out if there is indeed any problem like this.
-- Mike A. Harris Linux advocate Computer Consultant GNU advocate Capslock Consulting Open Source advocate
Join the FreeMWare project - the goal to produce a FREE program in which you can run Windows 95/98/NT, and other operating systems.
http://www.freemware.org
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |