It is disingenuous to portray a reasonable interpretation as a ‘minority position’.
if ZFS were statically linked with Linux and shipped as a single work, few would argue it was not a “work based on the Program
Equating linking to being a derivative work, a long standing position of RMS and FSF, is something that was counter-intuitive to even Stallman himself (See the last mail). It may be the case that a kernel module is a derivative work of the kernel, but RMS argues that even using readline as an optional dependency makes your whole program (in this case a CL implementation) a derivative work. A laughable claim. I hope this goes to trial and that the beginnings of a reasonable interpretation of derivative work for computer programs starts to be formed.
Edit: typo main -> mail
Well, it appears to be a minority of lawyers. They think judges would consider that user-does-the-link are subterfuges and would not look favourably upon someone trying to sidestep the GPL via user-does-the-link. It is the opinion of judges that matters.
This wasn’t RMS’s argument, but his lawyer’s argument. That’s why RMS was initially surprised by this.
It may be the case that a kernel module is a derivative work of the kernel, but RMS argues that even using readline as an optional dependency makes your whole program (in this case a CL implementation) a derivative work. A laughable claim. I hope this goes to trial and that the beginnings of a reasonable interpretation of derivative work for computer programs starts to be formed.
If this gets overturned in court GPL loses teeth overnight. I don’t think linking against a library is a derivative work inherently unless you’re using library instructions. In which case you are using GPL licensed instructions therefore making a derivative work but IANAL.
It also means that using proprietary libraries and linking to them optionally is not a derivative work, which weakens copyrights for libraries as a whole. Meaning that using libraries is fair game regardless of their licencing status or means of acquisition as long as you have some form of stubs making the proprietary library optional.
It’s really surreal to read this bureaucratic stuff. We all want to help users, but we’re going to avoid doing something that helps users because some paperwork says not to?
“helping users” is up to interpretation. The FSF stance is that the ultimate help for users is that they have full ability to help themselves and reuse the software anyway they want (under the provisions of the GPL). Sure, the stance is debatable, especially in the context of an optionally loadable module, but it is not bureaucratic, just because it is subtle.
OK, is anyone even a LITTLE bit surprised that Oracle isn’t leaping to make their OSS software GPL compatible? :)
To be fair though, they’re the ones who started Btrfs as well. (Before they acquired the copyright for ZFS though.)
I don’t really low licencing, wouldn’t it be the case, in all cases, that a third-party entity holds a magic wand that can instantly resolve the situation?
I believe this is specifically because Oracle are the majority copyright holders, as mentioned later in the article.
Relicensing zfs under gpl would probably be pretty terrible for the health of zfs itself though. The newly licensed zfs would no longer be usable in illumos and derivatives (which most of the authors of zfs work on), nor freebsd. Any changes in the interim (since original release) would also be unusable (released under a CDDL which being incompat is the whole point of this article). Maybe if zfs were relicensed to BSD or MIT or something similar, but then people would probably just do a hostile fork anyway.
They don’t have to relicense to GPL, just to a GPL-compatible license. The ones you mentioned are GPL-compatible. SFC seems to have said several times “GPL-compatible” and only once said “GPL”.
They could dual-license it under whichever licenses they want, they don’t have to limit themselves to one license.
I agree. I don’t really see the GPL per se as being all that important. OSS is OSS to my unsophisticated eyes (This is where RMS or one of his followers comes leaping to the fore :)
I just grimaced a bit when I read that part on the original article though. I get that SFC has an agenda, but their asking Oracle to relicense zfs specifically under GPL, and not something like “GPL, MPL, or a more liberal BSD/MIT license”, came off as a bit slimly when I read it. Not that Oracle ever gave even a single shit about what anyone else wanted!
I don’t know, I understood them saying just that
we again ask Oracle to respect community norms against license proliferation and simply relicense its copyrights in ZFS under a GPLv2-compatible license.
Which is even broader than these you mentioned.
While I agree that distributing both together would be a licensing violation, I don’t see how distributing them in separate packages necessarily are, unless the zfs.ko module itself includes GPL code? And even then you could get around it by using a separate ZFS dkms package.
You’ll find zfs.ko automatically built and installed on your Ubuntu systems.
From https://insights.ubuntu.com/2016/02/16/zfs-is-the-fs-for-containers-in-ubuntu-16-04/ linked in the article.
Fair enough, though that only solves part of the equation.