Now that Debian has added zfs support to their experimental branch; https://ftp-master.debian.org/new/zfs-linux_0.6.4.2-1.html

I understand. My thought was, that there seems to be a push to add support in Debian and Ubuntu. Would it now be possible/make sense to create maybe a nodebug kernel package based on a stable kernel supported by ZFS and zfs/spl packages maybe even as a copr repo?

-- devel mailing list devel(a)lists.fedoraproject.org http://lists.fedoraproject.org/admin/lists/devel@lists.fedoraproject.org

Here is a simple if then for figuring out how ZFS support may ever get into Fedora:

As a rule, I try not to take legal licensing interpretations from a CTO who's trying to sell me the thing they're talking about the licensing of. We certainly could send that interpretation of CDDL/GPL and the kernel to the legal team... but I'm not sure they'd agree with it.

Am 14.01.2016 um 19:57 schrieb Gerald B. Cox: > > On Thu, Jan 14, 2016 at 10:45 AM, Bill Nottingham <notting(a)splat.cc > <mailto:notting@splat.cc>> wrote: > > As a rule, I try not to take legal licensing interpretations from a > CTO > who's trying to sell me the thing they're talking about the > licensing of. > > We certainly could send that interpretation of CDDL/GPL and the > kernel to the > legal team... but I'm not sure they'd agree with it. > > > Well, if Lawrence Livermore is doing it, and Canonical apparently plans > to do it, it probably would be a good idea to get a determination from > the legal team who is "Lawrence Livermore"?

likely i did much more research than you can even imagine long before  that thread started

CDDL is incompatible with GPLv2 - period

On Thu, 2016-01-14 at 20:24 +0100, Reindl Harald wrote: > likely i did much more research than you can even imagine long > before > that thread started I find this challenging to believe. > CDDL is incompatible with GPLv2 - period Did you read the web site at all? The argument is that it can be used as a kernel module without constituting a derived work. Many developers believe this is not a GPL violation. Many believe otherwise. This is a well-known, open controversy. It's to be expected that different sets of lawyers will have different opinions on the risk depending on business requirements and their company's risk profile. Michael

> On 14 Jan 2016, at 11:39, Neal Gompa <ngompa13(a)gmail.com&gt; wrote: > > On Thu, Jan 14, 2016 at 2:35 PM, Michael Catanzaro <mcatanzaro(a)gnome.org&gt; wrote: >> On Thu, 2016-01-14 at 20:24 +0100, Reindl Harald wrote: >>> likely i did much more research than you can even imagine long >>> before >>> that thread started >> >> I find this challenging to believe. >> >>> CDDL is incompatible with GPLv2 - period >> >> Did you read the web site at all? The argument is that it can be used >> as a kernel module without constituting a derived work. Many developers >> believe this is not a GPL violation. Many believe otherwise. This is a >> well-known, open controversy. It's to be expected that different sets >> of lawyers will have different opinions on the risk depending on >> business requirements and their company's risk profile. >> >> Michael > > As far as I know, that's why the kernel has symbol export feature to > indicate which ones are covered by the GPL-ness (GPL_ONLY symbol > export). > The distinction between EXPORT_SYMBOL and EXPORT_SYMBOL_GPL is minimal. The theory [1] is that EXPORT_SYMBOL_GPL indicates that the kernel community believes that EXPORT_SYMBOL_GPL symbols are so core to Linux that you cannot use them without creating a derived work under copyright law. Thus using an EXPORT_SYMBOL_GPL symbol from a GPL-incompatible module is deliberate infringement, with all that that implies in terms of the legal system; using an EXPORT_SYMBOL symbol from a GPL-incompatible module *might* be non-infringing (if the work using it is legally separate in terms of copyright law), or might be accidental infringement (if you didn't realise what you were doing carried legal risk). In all cases, you need to talk to your copyright expert lawyer about distributing GPL-incompatible modules for the Linux kernel. Copyright law has some sharp edges, and you can get hurt if you ignore them; for Fedora, Red Hat Inc take on that liability, and they'll not want to do anything that puts them at risk of harm. [1] https://lwn.net/Articles/154602/ -- Simon Farnsworth -- devel mailing list devel(a)lists.fedoraproject.org http://lists.fedoraproject.org/admin/lists/devel@lists.fedoraproject.org

Well - If there was to be a plan, it would have to start with RH legal making that determination would it not? Could FESCO or the other council (sorry it escapes me ATM) take this up as a meeting item? Is it worth presenting for a legal determination?

In my mind, if it was approved by legal we would have to; A - determine with we were going to start building a kernel with nodebug turned off AND if it needs to be maintained as a separate kernel package to make sure that the kernel version tracked is supported by ZFS.

B - Be prepared to support a filesystem that needs modules build by DKMS

C - Find maintainers ( I would volunteer - I'd have to learn packaging)

D - Plan what release/testing, etc....

E - Decide if it when/if it would be supported in the installer and make those changes as well

F - do other stuff...

Should this be presented on another list?

Good Discussion all...

>> C - Find maintainers ( I would volunteer - I'd have to learn packaging) > > > >I'd certainly be willing to assist if it were allowed. I will be honest and say I do not foresee this being allowed in an official capacity. People are better off using the existing ZFS repo or contacting rpmfusion about it. To be clear, maintainers would need to be willing to deal with any and all kernel issues reported with ZFS loaded. We cannot staff the existing "next-generation" filesystem that is already upstream, let alone another that never will be. Also, willing maintainers should be item A:).

>> who is "Lawrence Livermore"? > > Lawrence Livermore National Laboratory is an organization founded by > the University of California to do research and development for > academic and government purposes. The US Department of Energy > commissioned them to port ZFS to Linux quite a long time ago[0], which > is the foundation of the current ZFS on Linux codebase. and they build a large, genral purpose, linux distribution?

> Please do some research before actually saying things likely i did much more research than you can even imagine long before that thread started

Does Oracle include ZFS in their ISO by default?

On Sat, Jan 16, 2016 at 3:00 PM, Chris Murphy <lists(a)colorremedies.com&gt; wrote: > On Sat, Jan 16, 2016 at 12:39 PM, Luya Tshimbalanga > <luya(a)fedoraproject.org&gt; wrote: >> Does Oracle include ZFS in their ISO by default? > > No, and as far as I know they don't contribute to ZFS on Linux. There > is a distinction between ZFS and OpenZFS that's kinda important. ZFS > on Linux is based on OpenZFS, not ZFS. There're incompatible features > since pool version 28 in each, so they're essentially diverging. I > don't know if that qualifies them as defacto forks (either from each > other, or from ZFS pool version 28). Anyway, Oracle only includes ZFS > in Solaris. And they continue to contribute to Btrfs. They do, however, include DTrace in their distribution, which remains CDDL licensed in their distribution

On Sat, Jan 16, 2016 at 4:18 PM, Reindl Harald <h.reindl(a)thelounge.net&gt; wrote: > > > Am 16.01.2016 um 22:07 schrieb Neal Gompa: >> >> On Sat, Jan 16, 2016 at 3:00 PM, Chris Murphy <lists(a)colorremedies.com&gt; >> wrote: >>> >>> On Sat, Jan 16, 2016 at 12:39 PM, Luya Tshimbalanga >>> <luya(a)fedoraproject.org&gt; wrote: >>>> >>>> Does Oracle include ZFS in their ISO by default? >>> >>> No, and as far as I know they don't contribute to ZFS on Linux. There >>> is a distinction between ZFS and OpenZFS that's kinda important. ZFS >>> on Linux is based on OpenZFS, not ZFS. There're incompatible features >>> since pool version 28 in each, so they're essentially diverging. I >>> don't know if that qualifies them as defacto forks (either from each >>> other, or from ZFS pool version 28). Anyway, Oracle only includes ZFS >>> in Solaris. And they continue to contribute to Btrfs. >> >> They do, however, include DTrace in their distribution, which remains >> CDDL licensed in their distribution > > stop that FUD - CDDL is not the problem - MIXING is the topic > and don't bring "cdrecrord" to the topic, the author is the problem > > back to topic: > even if DTrace would be closed source Oracle could include it because tehy > are the *copyright holder* and can release it under *every* license they > like to do DTrace is a kernel module that is CDDL licensed in Oracle Linux. So, not FUD

On 14 January 2016 at 12:20, Neal Gompa <ngompa13(a)gmail.com&gt; wrote: > On Thu, Jan 14, 2016 at 2:14 PM, Reindl Harald <h.reindl(a)thelounge.net&gt; wrote: >> >> >> Am 14.01.2016 um 19:57 schrieb Gerald B. Cox: >>> >>> On Thu, Jan 14, 2016 at 10:45 AM, Bill Nottingham <notting(a)splat.cc >>> <mailto:notting@splat.cc>> wrote: >>> >>> As a rule, I try not to take legal licensing interpretations from a >>> CTO >>> who's trying to sell me the thing they're talking about the >>> licensing of. >>> >>> We certainly could send that interpretation of CDDL/GPL and the >>> kernel to the >>> legal team... but I'm not sure they'd agree with it. >>> >>> >>> Well, if Lawrence Livermore is doing it, and Canonical apparently plans >>> to do it, it probably would be a good idea to get a determination from >>> the legal team >> >> >> who is "Lawrence Livermore"? >> > > Lawrence Livermore National Laboratory is an organization founded by > the University of California to do research and development for > academic and government purposes. The US Department of Energy > commissioned them to port ZFS to Linux quite a long time ago[0], which > is the foundation of the current ZFS on Linux codebase. > > Please do some research before actually saying things. > Might be good if you did also :). Lawrence Livermore National Laboratory is a US Government/Department of Energy laboratory that UC Berkeley has a hand in managing. It was not founded by the University of California but came out of the post Manhatten project to build the hydrogen bomb. The US DOE did not commission the lab to make the port. The port was done as part of a project to have a large scale filesystem work for certain computer clusters. That work got special permission from Sun but those permissions look to be sealed. Just because a port was done and then released does not mean that Lawrence Livermore is using it in any large way etc. There are thousands of projects that come out of these labs which maybe 3-20 students, grad students and phd's worked on but aren't in any use. Most of them will get that file attached to them which doesn't mean that LLNL sactioned it, DOE commission it etc etc. It just means that it was done at the lab as part of some project and the law requires that the file is attached to it.

LLNL is still actively involved in the ZFS on Linux project, so they are still doing something with it.

Lawrence Livermore National Laboratory is a US Government/Department of Energy laboratory that UC Berkeley has a hand in managing. It was not founded by the University of California but came out of the post Manhatten project to build the hydrogen bomb. The US DOE did not commission the lab to make the port. The port was done as part of a project to have a large scale filesystem work for certain computer clusters. That work got special permission from Sun but those permissions look to be sealed.

Just because a port was done and then released does not mean that Lawrence Livermore is using it in any large way etc. There are thousands of projects that come out of these labs which maybe 3-20 students, grad students and phd's worked on but aren't in any use. Most of them will get that file attached to them which doesn't mean that LLNL sactioned it, DOE commission it etc etc. It just means that it was done at the lab as part of some project and the law requires that the file is attached to it.

On Thu, Jan 14, 2016 at 11:14 AM, Reindl Harald <h.reindl(a)thelounge.net <mailto:h.reindl@thelounge.net>> wrote: ZFS cannot be included in the GPL-licensed Linux kernel, because it is licensed under the GPL-incompatible CDDL Harald, you missed the point

We all understand it cannot be included in the kernel - we're talking about whether or not it can be included in the distribution. Fedora already includes software that has incompatible GPL licenses...

https://www.gnu.org/licenses/license-list.html#GPLIncompatibleLicenses In fact, the CDDL is an approved Fedora license: https://fedoraproject.org/wiki/Licensing:Main?rd=Licensing

Just because something is incompatible with the GPL doesn't in and of itself blacklist it from being included in the distribution

Right. See also: https://fedoraproject.org/wiki/Forbidden_items#cdrtools for another case where an upstream attempted mixing GPL and CDDL code, and Red Hat Legal's and the FSF's stance on it.

It depends on exactly what FSF knows and how Canonical is planning to do this. It is not safe to assume FSF is even aware of all the details here. If you want FSF's opinion, they have a public contact address for licensing questions and respond pretty quickly. Note that FSF's opinion although influential does not determine what Fedora legally can ship. That has to be approved by Red Hat legal as well. And ... Fedora does not ship third party kernel modules.

Gerald B. Cox wrote: > Fedora has it's own rules and can ship or not ship what they want. I'm > perfectly fine with that. As I previously stated, IMO BTRFS is a much > better choice. My point was simply that I don't believe saying it would > be a GPL violation to include ZFS in a Linux distribution is one of them. > If it were, I can't imagine Canonical would be doing it. Canonical also has no qualms shipping the NVidia driver, which has the exact same licensing issue. They decided that they don't care.

On Sat, Jan 16, 2016 at 10:13 AM, Kevin Kofler <kevin.kofler(a)chello.at&gt; wrote: > Gerald B. Cox wrote: >> Fedora has it's own rules and can ship or not ship what they >> want. I'm perfectly fine with that. As I previously stated, IMO >> BTRFS is a much better choice. My point was simply that I don't >> believe saying it would be a GPL violation to include ZFS in a >> Linux distribution is one of them. If it were, I can't imagine >> Canonical would be doing it. > > Canonical also has no qualms shipping the NVidia driver, which has > the exact same licensing issue. They decided that they don't care. > Canonical ships everything as source code, so their justification likely is that they aren't doing binary distribution, for whatever that's worth.

The benchmark is probably what Debian and their team thinks of it, because Debian and Fedora have similarly strict guidelines for stuff like this because they *do* care.

On Sat, Jan 16, 2016 at 9:55 AM, Kevin Fenzi <kevin(a)scrye.com <mailto:kevin@scrye.com>> wrote: The benchmark if it's legal to include something in Fedora is what Fedora Legal says. I basically would agree with everything you stated, except I would change the sentence to read: "The benchmark if it's permissible..." Fedora has it's own rules, but when you use the term "legal" the connotation is that the other distributions which are distributing ZFS are breaking the law

Fedora cannot make that determination

blanket statements like that lead to FUD and misquotes. If Fedora chooses not to distribute ZFS for any reason, that is perfectly fine - however, the fact that we choose not to do so doesn't make it illegal

I can't image anyone misinterpreting my statement that way, but yes, I was not trying to suggest anything anyone else does is legal or not, simply that any inclusion in Fedora would need approval of Fedora legal and continuing to post about it or speculate is just a waste of time.

On Thu, Jan 14, 2016 at 10:57 AM, Gerald B. Cox <gbcox(a)bzb.us <mailto:gbcox@bzb.us>> wrote: On Thu, Jan 14, 2016 at 10:45 AM, Bill Nottingham <notting(a)splat.cc <mailto:notting@splat.cc>> wrote: As a rule, I try not to take legal licensing interpretations from a CTO who's trying to sell me the thing they're talking about the licensing of. We certainly could send that interpretation of CDDL/GPL and the kernel to the legal team... but I'm not sure they'd agree with it. Well, if Lawrence Livermore is doing it, and Canonical apparently plans to do it, it probably would be a good idea to get a determination from the legal team. I don't care one way or another, I use BTRFS - but we shouldn't be saying there are license issues if there aren't. I also found this: https://wiki.ubuntu.com/ZFS ZFS is licensed under the CDDL <http://en.wikipedia.org/wiki/CDDL>;, a popular and widely used OSI-approved open source license <http://opensource.org/licenses/category>;, that is recognized by the FSF <https://www.gnu.org/licenses/license-list.html#CDDL> as a free software license, but is incompatible with the GNU GPL. Because of that ZFS cannot be added to the Linux kernel directly. It can, however, be distributed as a DKMS package separate from the main kernel package

I understand. My thought was, that there seems to be a push to add support in Debian and Ubuntu. Would it now be possible/make sense to create maybe a nodebug kernel package based on a stable kernel supported by ZFS and zfs/spl packages maybe even as a copr repo?