News for hackers. Stuff that matters.
Re:ZFS Licensing and Linux by ajross on Friday February 19, 2016 @12:49PM (#11135075)
> And zfs.ko, as a self-contained file system module, is clearly not a derivative work of the Linux kernel but rather quite obviously a derivative work of OpenZFS and OpenSolaris

I don't see at all why that follows. Why not both? Building a kernel module requires including headers and writing glue code to conform to the kernel API, which is licensed only under the GPLv2. It strains reason to argue that a ELF binary with Linux metadata and Linux module entry points registering functions to implement a Linux filesystem is "clearly not" derivative of Linux.

Re:ZFS Licensing and Linux by ryao on Friday February 19, 2016 @01:31PM (#11135426)
The Linux community already had this discussion over the Andrew Filesystem, which was a proprietary port. That is partly why we have symbols designated as being okay for non-GPL code. As for using headers, few consider them to be subject to copyright and those that do would likely find a fair use case for using them for compatibility. However, complex macros and inline functions are a concern. There are not many that are quite so large that they would be considered non-trivial. So far, I have yet to see any that should pose a problem for legal review, although I have only paid attention to ones used by ZFS.

Also, if it is not clear, this was raised inside Gentoo. There was a discussion involving this with myself, the ZoL project lead, the concerned developer and a member of the licensing team who had final call on the matter. The descriptions of what was being done and why it should be okay passed review.

Re:ZFS Licensing and Linux by rodgerd on Friday February 19, 2016 @01:53PM (#11135613)
> As for using headers, few consider them to be subject to copyright and those that do would likely find a fair use case for using them for compatibility.

Except as it stands, Oracle have now won cases in the US courts that designate them copyrightable, have they not?

Re:ZFS Licensing and Linux by zerocrates on Friday February 19, 2016 @02:11PM (#11135771)
Oracle v. Google did break Oracle's way on API copyrightability, but the precedent is from the Federal Circuit (because there were also-appealed patent claims), and they wouldn't end up hearing most subsequent copyright cases... so it's all pretty murky at best.

Despite their occasional enthusiasm for overturning the Federal Circuit, the Supreme Court denied cert in the case, so this probably won't be settled any time soon.

Re:ZFS Licensing and Linux by ryao on Friday February 19, 2016 @02:11PM (#11135761)
That is restricted to the US, involved whether a compatible implementation could be made using them and offered the possibility that all such implementations were actually fair use. It really is not relevant unless you are discussing making a Linux kernel replacement under a different license using the Linux kernel headers. As far as I last heard from lawyers on this, it is not considered to affect anyone until the legal battle finishes and as far as I can tell, if a oracle manages to say this is not okay, they will be in far deeper trouble over their DTrace module that circumvents GPL symbol exports than ZFSOnLinux that restricts itself to an interface intended to allow proprietary modules.
Re:ZFS Licensing and Linux by dsp1234 on Friday February 19, 2016 @01:02PM (#11135184)
It strains reason to argue...

So then the question is, why is "nvidia.ko"'s binary blob not a derivative of Linux? What's the logic that says one binary blob is a derivative work, and the other is not when they both are 'ELF binary with Linux metadata and Linux module entry points'.

Re:ZFS Licensing and Linux by ajross on Friday February 19, 2016 @01:06PM (#11135211)
The NVIDIA driver has been long subject to exactly this controversy, which is one reason why distros don't like to ship it.
Re:ZFS Licensing and Linux by wnoise on Friday February 19, 2016 @01:09PM (#11135247)
Many people think it is, but apparently no one with authority to go to court over it has done so.
Re:ZFS Licensing and Linux by dsp1234 on Friday February 19, 2016 @01:22PM (#11135346)
It certainly would be interesting to see a "Software Freedom Law Center" vs "Free Software Foundation" legal battle. But if no one wants to fight over a clearly more egregious graphics blob, then why would they go after Canonical for something seemingly less of an issue.