| |||||||||||||
|
| |||||||||||||
Don't mix you desires with facts!Don't mix you desires with facts!Posted Oct 15, 2008 13:21 UTC (Wed) by khim (subscriber, #9252)In reply to: GPL doesn't restrict usage? True. Is it enough? YMMV. by drag Parent article: Linux Summit will preview new advanced file system (SearchEnterpriseLinux.com)
GPL only matters in distribution And this is because GPL says ... in part ... . Fill the blanks, please. It's the purpose of the GPL. That's it, it's by design. In your head - may be. Not in real life. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or IF YOU MODIFY IT (emphasis mine). You may not copy, MODIFY, sublicense, or distribute the Program except as expressly provided under this License (emphasis mine again). Does it look like "only distribution does matter" type of license to you? You can see the word 'distribution' repeated many times. In GPLv3 it's propagation. But does not matter: distribution is there, in GPLv2, true, but so is modification. It's restricted, too! Heck: the whole license has a name TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION. The thing which is not restricted is the act of running the Program - and that's all. Can you say without the court decision: combination of works created from BSD kernel, GPLed ext3fs and CDDLed ZFS is "the act of running the program" or "modification of program"? The resulting combination can do things which parts can not do (like copy file from ext3fs-formatted partition to ZFS-formatted one and back)... And these modules surely don't implement some "VFS RFC"...
(Log in to post comments)
Don't mix you desires with facts! Posted Oct 15, 2008 13:39 UTC (Wed) by paulj (subscriber, #341) [Link]
The GPL and the modification conditions apply to the preferred form of the sources - not derived binaries (directly at least).
If you got this theory from a lawyer, you need to find a new one, methinks.
GPL does not decide where it's applicable or not - the law does Posted Oct 15, 2008 14:08 UTC (Wed) by khim (subscriber, #9252) [Link] The GPL and the modification conditions apply to the preferred form of the sources - not derived binaries (directly at least). Novel idea. So you claim that I can add any proprietary code to linux as long as I'm not modifying source, but only binary by the way of hex editor? Sorry, if you got this theory from a lawyer, you need to find a new one, methinks. You are mixing cause and consequence. Where and how GPL applies can not and does not written in GPL. Some jurisdictions allow some [limited] form of modifications without license (for example you can reverse-engineer stuff if it's the only way to make it work in Norway - see DVD John), some do not. Read this for example. MDY tried to claim that EULA does not apply to it because it never agreed to its terms, but court decided that they still illegally modified the code in memory - and if it's not allowed without appropriate license then that means GPL should explicitly allow it. It does not. It's pretty clear that GPL does not contain permissions to make such a combination, but do you even NEED such permission? That's the question. It depends on many-many details and can differ from country to country and even from court to court! That's why I never said "yes" about such a thing and never said "no": YMMV. And yes, I've discussed this with actual lawer (it was about combining GPL and CDDL code - though not about linux kernel and ZFS).
GPL does not decide where it's applicable or not - the law does Posted Oct 15, 2008 14:57 UTC (Wed) by paulj (subscriber, #341) [Link] I didn't claim that. :) The binaries are also subject to conditions under the GPL. As for in-ram-linking-as-modification: just look at the modification licence, its framed in terms of conditions that are really only relevant to fixed, distributable copies. It seems very unreasonable, to me, to think the framer had transient modifications in mind. It is then reasonable to think that the alternative is to consider that such transient modifications, made in the course of running, must then be covered by the 'running is unrestricted' permission. That this has been the prevailing wisdom for a long time (e.g. the GPL FAQ, that NVidia are happy to distribute binary-only drivers for users to load with their GPL kernel, etc..) suggests strongly that your lawyer's interpretation is beyond the pale. IANAL, my view is based primarily on assessing the words and actions of those who have (had) significant matters at stake, ICBW, etc.
Prevailing wisdom? May be in your head... Posted Oct 15, 2008 16:48 UTC (Wed) by khim (subscriber, #9252) [Link] As for in-ram-linking-as-modification: just look at the modification licence, its framed in terms of conditions that are really only relevant to fixed, distributable copies. Bingo. And what does it mean? Right: you go back to copyright law. Does it allow this particular modification or not? YMMV. You've been warned. That this has been the prevailing wisdom for a long time (e.g. the GPL FAQ, that NVidia are happy to distribute binary-only drivers for users to load with their GPL kernel, etc..) suggests strongly that your lawyer's interpretation is beyond the pale. GPL FAQ says: if Sun will produce internal ZFS driver for Linux GPL will
not force a disclosure - Sun will be free to use it internally. Good, but
hardly relevant. NVidia does not distribute GPLed code at all - so again,
not a problem. As for "prevailing wisdom"... Who's wisdoms is
this? Not mine - that's for sure. When at least one distribution was closed over GPL troubles,
when Linus says A vendor who
distributes non-GPL modules is _not_ protected by the module interface per
se, and should feel very confident that they can show in a court of law
that the code is not derived, when you can see things like Oh, and at least one major distro
has been served with legal papers due to them shipping closed source kernel
drivers, and more are on the way (I'm not sure if it was Kororaa or
not, but I sure as hell know that Novell wrote Most developers of the kernel community consider non-GPL
kernel modules to be infringing on their copyright. Novell does respect
this position, and will no longer distribute non-GPL kernel modules as part
of future products) it sure looks to me like "prevailing wisdom" says
the following: Basically it means: CDDLed ZFS can not be included in kernel. No way, no how. It'll be infringment - plain and simple. Separate distribution is probably Ok, but watch your back. Oh, and distributors can not include it in distribution too. Do you really think there are a future for ZFS in a position like this? Permanently ostracized without any hope to be ever included in the trunk and without any support from kernel developers? I sincerly doubt it. If your lawers claim that when you are interpreting the license you can safely ignore opinion of guys who actually wrote the code under said license and still keep the copyright... then you are beyond caring, I afraid... Linus and Co always said one thing and thing only: we can not force GPL on non-derived work. No way, not how. And this is the only reason why non-GPLed modules can exist.
Prevailing wisdom? May be in your head... Posted Oct 15, 2008 17:24 UTC (Wed) by paulj (subscriber, #341) [Link]
Firstly, I don't speak for Sun at all, at all... I have never talked to Sun legal about the CDDL - so there aren't even lines to read between. I have no special knowledge, bar a modicum of knowledge of the internal culture.
Your long post is interesting, and I agree that Linux contributors can not incorporate GPL-incompat code because of GPL licence concerns - presuming they can not agree to allow themselves to do so. To be clear: at no stage have I argued that point, and I have no interest in discussing it.
My main point here was to counter the FUD that Sun would sue for CDDL infringement.
Anyway, dinner :)
|
|
||||||||||||