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 is there a possibility that Fedora could add support/packages as well?
Regards, Zach #aikidouke
On 01/14/2016 03:26 PM, Zach Villers wrote:
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 don't know where you got this information. If a package is in NEW, it is not yet part of Debian. In fact, it means that ftpmaster review (which includes both licensing and technical reviews) is still pending. This query confirms that there is no zfs-linux source package in Debian:
Any Debian developer can upload packages to the NEW queue. This does not mean they end up in the distribution.
Florian
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?
On Thu, Jan 14, 2016 at 10:04 AM, Florian Weimer fweimer@redhat.com wrote:
On 01/14/2016 03:26 PM, Zach Villers wrote:
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 don't know where you got this information. If a package is in NEW, it is not yet part of Debian. In fact, it means that ftpmaster review (which includes both licensing and technical reviews) is still pending. This query confirms that there is no zfs-linux source package in Debian:
< https://packages.debian.org/search?suite=default%C2%A7ion=all&arch=any&a...
Any Debian developer can upload packages to the NEW queue. This does not mean they end up in the distribution.
Florian
devel mailing list devel@lists.fedoraproject.org http://lists.fedoraproject.org/admin/lists/devel@lists.fedoraproject.org
On 14 January 2016 at 08:58, Zach Villers zachvatwork@gmail.com wrote:
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?
Here is a simple if then for figuring out how ZFS support may ever get into Fedora:
1. Is ZFS now licensed under a GPL 2.0/BSD or MIT license? [These seem to be ones acceptable to be included into anything kernel related.]
2. If yes, please email legal@redhat.com to see if a further review is needed for patents or other problems.
3. If no, then "Not going to happen."
-- devel mailing list devel@lists.fedoraproject.org http://lists.fedoraproject.org/admin/lists/devel@lists.fedoraproject.org
Thanks Smooge/Florian. I respect your opinions, thoughts, and explanations.
Regards.
On Thu, Jan 14, 2016 at 12:25 PM, Stephen John Smoogen smooge@gmail.com wrote:
On 14 January 2016 at 08:58, Zach Villers zachvatwork@gmail.com wrote:
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?
Here is a simple if then for figuring out how ZFS support may ever get into Fedora:
- Is ZFS now licensed under a GPL 2.0/BSD or MIT license? [These
seem to be ones acceptable to be included into anything kernel related.]
- If yes, please email legal@redhat.com to see if a further review is
needed for patents or other problems.
- If no, then "Not going to happen."
-- devel mailing list devel@lists.fedoraproject.org http://lists.fedoraproject.org/admin/lists/devel@lists.fedoraproject.org
-- Stephen J Smoogen. -- devel mailing list devel@lists.fedoraproject.org http://lists.fedoraproject.org/admin/lists/devel@lists.fedoraproject.org
On Thu, Jan 14, 2016 at 9:25 AM, Stephen John Smoogen smooge@gmail.com wrote:
Here is a simple if then for figuring out how ZFS support may ever get into Fedora:
I originally believed it was simply a licensing issue that was preventing the inclusion in Fedora, but apparently that isn't true: http://warpmech.com/?news=myth-busting-series-zfs-on-linux-has-license-probl...
Gerald B. Cox (gbcox@bzb.us) said:
On Thu, Jan 14, 2016 at 9:25 AM, Stephen John Smoogen smooge@gmail.com wrote:
Here is a simple if then for figuring out how ZFS support may ever get into Fedora:
I originally believed it was simply a licensing issue that was preventing the inclusion in Fedora, but apparently that isn't true: http://warpmech.com/?news=myth-busting-series-zfs-on-linux-has-license-probl...
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.
Bill
On Thu, Jan 14, 2016 at 10:45 AM, Bill Nottingham 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.
Am 14.01.2016 um 19:57 schrieb Gerald B. Cox:
On Thu, Jan 14, 2016 at 10:45 AM, Bill Nottingham <notting@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"?
is Canonical located in the US? no! is Redhat located in the US? yes! you see the difference?
https://en.wikipedia.org/wiki/GNU_General_Public_License
ZFS cannot be included in the GPL-licensed Linux kernel, because it is licensed under the GPL-incompatible CDDL
PERIOD
On Thu, Jan 14, 2016 at 2:14 PM, Reindl Harald h.reindl@thelounge.net wrote:
Am 14.01.2016 um 19:57 schrieb Gerald B. Cox:
On Thu, Jan 14, 2016 at 10:45 AM, Bill Nottingham <notting@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.
[0]: https://raw.githubusercontent.com/zfsonlinux/zfs/master/DISCLAIMER
Am 14.01.2016 um 20:20 schrieb Neal Gompa:
On Thu, Jan 14, 2016 at 2:14 PM, Reindl Harald h.reindl@thelounge.net wrote:
Am 14.01.2016 um 19:57 schrieb Gerald B. Cox:
On Thu, Jan 14, 2016 at 10:45 AM, Bill Nottingham <notting@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.
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
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
Am 14.01.2016 um 20:35 schrieb Michael Catanzaro:
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.
i don't care what you 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.
and Redhat legal decided against now you!
On Thu, Jan 14, 2016 at 2:35 PM, Michael Catanzaro mcatanzaro@gnome.org 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).
On 14 Jan 2016, at 11:39, Neal Gompa ngompa13@gmail.com wrote:
On Thu, Jan 14, 2016 at 2:35 PM, Michael Catanzaro mcatanzaro@gnome.org 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/
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...
On Thu, Jan 14, 2016 at 3:37 PM, Simon Farnsworth simon@farnz.org.uk wrote:
On 14 Jan 2016, at 11:39, Neal Gompa ngompa13@gmail.com wrote:
On Thu, Jan 14, 2016 at 2:35 PM, Michael Catanzaro mcatanzaro@gnome.org
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@lists.fedoraproject.org http://lists.fedoraproject.org/admin/lists/devel@lists.fedoraproject.org
On Thu, Jan 14, 2016 at 5:07 PM, Zach Villers zachvatwork@gmail.com wrote:
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?
There are two councils that matter: the Fedora Council[0] and the Fedora Engineering Steering Committee[1]. Both of these will likely have a say in this matter.
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.
This should not be necessary, as our kernel is very close to mainline, making it quite easy for kmod authors to target.
B - Be prepared to support a filesystem that needs modules build by DKMS
This shouldn't be too terrible, as most of the userspace would treat it like any other filesystem.
C - Find maintainers ( I would volunteer - I'd have to learn packaging)
I'd certainly be willing to assist if it were allowed.
D - Plan what release/testing, etc....
Again, definitely willing to help out here.
E - Decide if it when/if it would be supported in the installer and make those changes as well
I would hope that Anaconda would grow support for it if we did have it.
F - do other stuff...
What else do you think would need to happen?
Should this be presented on another list?
I think all the relevant stakeholders are present on this list. *If* this goes anywhere, then we would make a proposal to the Fedora Packaging Committee[2] for a policy change.
Good Discussion all...
[0]: https://fedoraproject.org/wiki/Council [1]: https://fedoraproject.org/wiki/Fedora_Engineering_Steering_Committee [2]: https://fedoraproject.org/wiki/Packaging_Committee
On Thu, Jan 14, 2016 at 5:29 PM, Neal Gompa ngompa13@gmail.com wrote:
On Thu, Jan 14, 2016 at 5:07 PM, Zach Villers zachvatwork@gmail.com wrote:
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?
You can contact legal@lists.fedoraproject.org with legal questions independently of any FESCo/Council items.
There are two councils that matter: the Fedora Council[0] and the Fedora Engineering Steering Committee[1]. Both of these will likely have a say in this matter.
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.
This should not be necessary, as our kernel is very close to mainline, making it quite easy for kmod authors to target.
B - Be prepared to support a filesystem that needs modules build by DKMS
This shouldn't be too terrible, as most of the userspace would treat it like any other filesystem.
Except for that whole other thread you started where I laid out why we said we aren't going to do kmods as a deliverable from the Fedora project. If one wants to start this in a separate repo, that's up to them.
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 :).
D - Plan what release/testing, etc....
Again, definitely willing to help out here.
E - Decide if it when/if it would be supported in the installer and make those changes as well
I would hope that Anaconda would grow support for it if we did have it.
F - do other stuff...
What else do you think would need to happen?
Should this be presented on another list?
I think all the relevant stakeholders are present on this list. *If* this goes anywhere,
Being transparent, I don't see this making it past legal. On the very off chance it did, I don't see it making it past some of the other policy and technical hurdles we have.
then we would make a proposal to the Fedora Packaging Committee[2] for a policy change.
A change for what? Allowing kmods in Fedora? DKMS itself is already provided, but we do not provide binary kmods themselves. I've already explained why in the other thread. Or do you somehow mean shipping module source that is then built on user's systems? That seems... fragile at best.
josh
On 01/14/2016 07:29 PM, Josh Boyer wrote:
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:).
Being a maintainer of a file system is a very, very large task. Specifically, file systems (especially ZFS and btrfs) have a lot of hooks/dependencies into the core IO/block layer and memory management system.
It definitely is not a "packaging" job, you need serious kernel skills here.
Putting the license aside, I think that it would be a really bad idea to pull in any kernel component that was not upstream in the mainline kernel. The maintainers of the ZFS code (who I know) can share the pain of chasing the rest of the kernel. Building a file system as a kernel module or having it built in is just a packaging issue, we should not use packaging as a way to circumvent legal and community norms (in Fedora or the upstream kernel community).
Coming back to the issue of the license, I think that there is no chance that it will ever land in the upstream kernel as long as it has this license.
My personal opinion is that the effort is better spent on working on btrfs or the host of technologies in device mapper that provide very similar features/functions to what people are looking for in ZFS.
Regards,
Ric
On Thu, Jan 14, 2016 at 3:07 PM, Zach Villers zachvatwork@gmail.com wrote:
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?
I think there are parallel "deal killers" including legal. It makes sense to pursue them in an order that combines least effort with greatest chance of failure. i.e. asking legal to do a bunch of work doesn't make much sense (to me) if there isn't a critical mass of people willing to do the work to create and maintain the packaging. And it'd be a good idea to go through exactly what that work is, and what Fedora resources will be needed, and ask those resources (kernel team, Release Engineering, Fedora Packaging Council) what will be needed and if they have the resources and interest available to help, etc.
And again, what deficiency there is with the existing ZoL package for Fedora, and how Fedora benefits having it offered within Fedora? I
Reindl Harald h.reindl@thelounge.net writes:
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?
Doubtless it doesn't count, but an old version of the LLNL HPC-oriented GNU/Linux distribution which incorporates ZFS is at ftp://gdo-lc.ucllnl.org/pub/projects/chaos/5.1/SRPMS/. (I don't know if the RHEL7-derived one is published.) There's another from a US lab that you may have heard of http://ftp2.scientificlinux.org/linux/scientific/6x/.
There is a distribution of Linux by a US corporation with CDDL modules http://yum.oracle.com/repo/OracleLinux/OL7/UEKR3/x86_64/index_src.html for their general purpose GNU/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
At least knowing about LLNL when sounding off about ZFS could stop someone looking foolish.
On 14/01/16 02:28 PM, Dave Love wrote:
Reindl Harald h.reindl@thelounge.net writes:
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?
Doubtless it doesn't count, but an old version of the LLNL HPC-oriented GNU/Linux distribution which incorporates ZFS is at ftp://gdo-lc.ucllnl.org/pub/projects/chaos/5.1/SRPMS/. (I don't know if the RHEL7-derived one is published.) There's another from a US lab that you may have heard of http://ftp2.scientificlinux.org/linux/scientific/6x/.
There is a distribution of Linux by a US corporation with CDDL modules http://yum.oracle.com/repo/OracleLinux/OL7/UEKR3/x86_64/index_src.html for their general purpose GNU/Linux distribution.
Does Oracle include ZFS in their ISO by default? If so, it is asking for trouble by violating GPL license mixing with an incompatible CDDL. Note that Oracle owns ZFS and has the liberty to sue any large commercial distribution. For that reason, it is not worth a legal risk.
On Sat, Jan 16, 2016 at 12:39 PM, Luya Tshimbalanga luya@fedoraproject.org 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.
On Sat, Jan 16, 2016 at 3:00 PM, Chris Murphy lists@colorremedies.com wrote:
On Sat, Jan 16, 2016 at 12:39 PM, Luya Tshimbalanga luya@fedoraproject.org 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.
Am 16.01.2016 um 22:07 schrieb Neal Gompa:
On Sat, Jan 16, 2016 at 3:00 PM, Chris Murphy lists@colorremedies.com wrote:
On Sat, Jan 16, 2016 at 12:39 PM, Luya Tshimbalanga luya@fedoraproject.org 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
On Sat, Jan 16, 2016 at 4:18 PM, Reindl Harald h.reindl@thelounge.net wrote:
Am 16.01.2016 um 22:07 schrieb Neal Gompa:
On Sat, Jan 16, 2016 at 3:00 PM, Chris Murphy lists@colorremedies.com wrote:
On Sat, Jan 16, 2016 at 12:39 PM, Luya Tshimbalanga luya@fedoraproject.org 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.
Am 16.01.2016 um 22:19 schrieb Neal Gompa:
On Sat, Jan 16, 2016 at 4:18 PM, Reindl Harald h.reindl@thelounge.net wrote:
Am 16.01.2016 um 22:07 schrieb Neal Gompa:
On Sat, Jan 16, 2016 at 3:00 PM, Chris Murphy lists@colorremedies.com wrote:
On Sat, Jan 16, 2016 at 12:39 PM, Luya Tshimbalanga luya@fedoraproject.org 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
come on get a laywer and dicuss that topic somewhere else
first they laywer needs to find out *which* of both licenses maybe violated - if it's the CDDL then it's no problem for Oracle, if it's the GPL, well it needs a lawsuit in doubt
IT DOES NOT MATTER what Oracle does IT DOES NOT MATTER what anybody else does ABOVE DOES NOT MATTER for Fedora
licensing is a minefield and hence Fedora / Redhat legal stays on the SAVE SIDE - so WHAT needs to be discussed again and again
and since you still did not realize it: even without the legal questions a out-of-tree module WON'T make it into Fedora and so the whole topic is done
On 14 January 2016 at 12:20, Neal Gompa ngompa13@gmail.com wrote:
On Thu, Jan 14, 2016 at 2:14 PM, Reindl Harald h.reindl@thelounge.net wrote:
Am 14.01.2016 um 19:57 schrieb Gerald B. Cox:
On Thu, Jan 14, 2016 at 10:45 AM, Bill Nottingham <notting@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.
On Thu, Jan 14, 2016 at 6:54 PM, Stephen John Smoogen smooge@gmail.com wrote:
On 14 January 2016 at 12:20, Neal Gompa ngompa13@gmail.com wrote:
On Thu, Jan 14, 2016 at 2:14 PM, Reindl Harald h.reindl@thelounge.net wrote:
Am 14.01.2016 um 19:57 schrieb Gerald B. Cox:
On Thu, Jan 14, 2016 at 10:45 AM, Bill Nottingham <notting@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.
Am 15.01.2016 um 01:07 schrieb Neal Gompa:
On Thu, Jan 14, 2016 at 6:54 PM, Stephen John Smoogen smooge@gmail.com wrote:
On 14 January 2016 at 12:20, Neal Gompa ngompa13@gmail.com wrote:
On Thu, Jan 14, 2016 at 2:14 PM, Reindl Harald h.reindl@thelounge.net wrote:
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
that don't make them to a general purpose Linux distibution which as the whole point of "who is "Lawrence Livermore?"
On Thu, Jan 14, 2016 at 4:07 PM, Neal Gompa ngompa13@gmail.com wrote:
LLNL is still actively involved in the ZFS on Linux project, so they are still doing something with it.
Correct, and that can be discovered with a Google search - which found this: *https://events.linuxfoundation.org/sites/events/files/slides/OpenZFS%20-%20L... https://events.linuxfoundation.org/sites/events/files/slides/OpenZFS%20-%20LinuxCon_0.pdf*
On the Canonical side there was this article with some comments on the licensing issue: https://www.phoronix.com/scan.php?page=news_item&px=Ubuntu-ZFS-Standard-...
Here is another link from the people who should know: http://zfsonlinux.org/faq.html#WhatAboutTheLicensingIssue
If the FSF believed the above statement from zfsonlinux was not correct, one would think they would quickly and clearly rebuke it; especially considering Canonical's plans.
I haven't found any such statement, perhaps someone else has the link.
Again, there can be a multitude of valid reasons that Fedora does not include ZFS. Personally, I think BTRFS is a much better choice than ZFS. I just don't believe people should hang their hat on GPL infringement.
ZFS on Linux: Copyright and Licensing Issues (pdf created april 2013) This was written by a lawyer, but the disclaimer is this is a cursory evaluation, not legal advice. http://www.rtt-law.com/public/files/docs/Williams%20-%20Software%20copyright...
But I still predict bigger obstacles. I just don't see how it'd ever be OK to have a tainted kernel being used on installer media, and that'd have to be the case in order to get install time support, either for root fs or even as just /data or /srv storage. I'd eat my hat (and a billy goat) if any of either the kernel, QA, or installer teams were OK with this. It's too much work and risk for very little upside. So it'd be a post-install package at best no matter what, in which case what's the advantage of it being offered by Fedora rather than from the ZoL org folks?
--- Chris Murphy
Stephen John Smoogen smooge@gmail.com writes:
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.
To be clear for people not familiar with this, there are two levels of (then) Sun filesystems involved. They use many ZFS filesystem (pool) instances to back the Lustre parallel filesystem. ZFS replaces the original Lustre ext-ish stuff. See the history in http://zfsonlinux.org/docs/LUG11_ZFS_on_Linux_for_Lustre.pdf. [I'm not sure the rms quote there is in context.]
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.
Yes, but this specifically concerns the Sequoia 50 PB, 800 GB/s parallel filesystem, e.g. http://zfsonlinux.org/docs/Morrone_Sequoia_LUG2013.pdf and the above. I assume it's served off EL6-ish systems, by the way.
On Thu, Jan 14, 2016 at 11:14 AM, Reindl Harald 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.
Am 14.01.2016 um 20:34 schrieb Gerald B. Cox:
On Thu, Jan 14, 2016 at 11:14 AM, Reindl Harald <h.reindl@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
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...
but you don't understand "derived work" which affects kernel modules
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
a standalone software with CDDL is a different topic
Just because something is incompatible with the GPL doesn't in and of itself blacklist it from being included in the distribution
no, but when you link that incompatible code with the kernel which is GPLv2 code it's a complete different topic
for "out of tree kernel modules" just read the other thread....
Reindl Harald wrote:
a standalone software with CDDL is a different topic
[snip]
no, but when you link that incompatible code with the kernel which is GPLv2 code it's a complete different topic
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.
for "out of tree kernel modules" just read the other thread....
It would not make any difference anyway: http://www.gnu.org/licenses/gpl-faq.html#GPLStaticVsDynamic
Kevin Kofler
On Fri, Jan 15, 2016 at 7:38 PM, Kevin Kofler kevin.kofler@chello.at wrote:
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.
Kevin, I don't believe that is the case in this instance. No one is talking about mixing code. If you do have something however specifically regarding the FSF stance on ZFS, I'd like to read it - I've searched and haven't been able to find anything. As I previously mentioned, one would think if the FSF had issue with it they would come out and rebuke all the statements out there that say it isn't an issue; especially since Canonical is actively working on it.
Hi
On Fri, Jan 15, 2016 at 11:42 PM, Gerald B. Cox wrote:
Kevin, I don't believe that is the case in this instance. No one is
talking about mixing code. If you do have something however specifically regarding the FSF stance on ZFS, I'd like to read it - I've searched and haven't been able to find anything. As I previously mentioned, one would think if the FSF had issue with it they would come out and rebuke all the statements out there that say it isn't an issue; especially since Canonical is actively working 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.
Rahul
On Fri, Jan 15, 2016 at 9:16 PM, Rahul Sundaram metherid@gmail.com wrote:
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.
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.
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.
Kevin Kofler
On Sat, Jan 16, 2016 at 10:13 AM, Kevin Kofler kevin.kofler@chello.at 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.
I don't know if it's possible to talk to the Debian folks about it, because they've clearly discussed it for quite a while[0][1][2]. In terms of legal things, Debian appears to use the SFLC for legal advice about ZFS, and I don't know (nor wish to speculate) whether Red Hat Legal talks to those folks.
[0]: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=686447 [1]: https://lists.debian.org/debian-devel-announce/2015/04/msg00006.html [2]: https://lists.debian.org/debian-devel-announce/2015/05/msg00004.html
On Sat, Jan 16, 2016 at 7:38 AM, Neal Gompa ngompa13@gmail.com wrote:
On Sat, Jan 16, 2016 at 10:13 AM, Kevin Kofler kevin.kofler@chello.at 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.
As a technical matter, Fedora could ship ZFS source only. I don't know whether that would help the legal issues.
On Sat, 16 Jan 2016 10:38:42 -0500 Neal Gompa ngompa13@gmail.com wrote:
On Sat, Jan 16, 2016 at 10:13 AM, Kevin Kofler kevin.kofler@chello.at 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.
I'm confused by the two above sentences. :) The NVida binary only non free driver's source is only available to AMD/nvidia. The shim that connects it to the kernel has source available, but the driver itself does not.
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.
Nope. The benchmark if it's legal to include something in Fedora is what Fedora Legal says. Thats it. We have (and likely will continue) to disagree with Debian on some things (like mp3).
IMHO, someone needs to ask Legal if it can be included. If for some reason things have changed and they say yes, then it would hit the out of tree kmods issues and likely be impractical there.
So, no, I don't think it's going to be possible or practical to include ZFS.
kevin
On Sat, Jan 16, 2016 at 9:55 AM, Kevin Fenzi 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; and 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.
Am 16.01.2016 um 19:43 schrieb Gerald B. Cox:
On Sat, Jan 16, 2016 at 9:55 AM, Kevin Fenzi <kevin@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
nonsense
if it comes to law topics in many cases 5 lawyers are coming up with 6 opinions and every decision made has *nothing* to do with any other party
Fedora cannot make that determination
it don't make it for others it just makes it for Fedora
if Redhat legal says "no" to be on the safe side that has *nothing* to do with any otehr distribution
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
see above
and now please *stop* this topic ZFS and external kmods don't make it to Fedora suck it!
On Sat, 16 Jan 2016 10:43:12 -0800 "Gerald B. Cox" gbcox@bzb.us wrote:
On Sat, Jan 16, 2016 at 9:55 AM, Kevin Fenzi 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; and 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.
kevin
On Sat, Jan 16, 2016 at 11:35 AM, Kevin Fenzi kevin@scrye.com wrote:
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.
ROFL... you know the internet Kevin... and any conspiracy theory can and will be propagated... but I didn't believe you intended to imply that...
Am 16.01.2016 um 20:51 schrieb Gerald B. Cox:
On Sat, Jan 16, 2016 at 11:35 AM, Kevin Fenzi <kevin@scrye.com mailto:kevin@scrye.com> wrote:
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.
ROFL... you know the internet Kevin... and any conspiracy theory can and will be propagated... but I didn't believe you intended to imply that...
hence the next time as Fedora legal directly instead repeat the same discussion again (it's not the first time this topic appeared) where people in the thread even point to "Oracle Linux" with "but they do too" while Orcale holds all copryrights and patents in ZFS after buyed Sun long ago
I think someone needs to agree to become the maintainer for such a package, correct? Who's willing to do this?
I haven't tested ZoL on my UEFI+Secure Boot NUC yet, but my expectation is that this kernel module isn't going to be signed by anything trusted, so it'd fail to load. If that's true, there's an advantage for building it within Fedora infrastructure if it's going to get built with a Fedora signing key. But the kernel module private key used for signing isn't kept around for building other modules (it goes back to the ether).
And if that can't or won't happen, is there a problem with the way it's package by the zfsonlinux.org folks now? They have a Fedora specific repo.
Anyway, it all sounds non-trivial. It's going to take a plan.
Chris Murphy
On Thu, Jan 14, 2016 at 10:57 AM, Gerald B. Cox gbcox@bzb.us wrote:
On Thu, Jan 14, 2016 at 10:45 AM, Bill Nottingham 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.
Am 14.01.2016 um 20:15 schrieb Gerald B. Cox:
On Thu, Jan 14, 2016 at 10:57 AM, Gerald B. Cox <gbcox@bzb.us mailto:gbcox@bzb.us> wrote:
On Thu, Jan 14, 2016 at 10:45 AM, Bill Nottingham <notting@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
oh yeah - bring a DKMS package out of tree *for your filesystems* in the mix and wait for bugreports of users in trouble....
do it at your own - but don't expect it a good idea for a general purpose distribution and *no* "but others do" is no valid justification
Am 14.01.2016 um 19:45 schrieb Bill Nottingham:
Bill, your maildomain is burned and the only reason that your mails appear here is there the mailing list is whitelisted based on SPF
1.5 URIBL_SBL_A Contains URL's A record listed in the SBL blocklist [URIs: splat.cc]
1.5 URIBL_SBL Contains an URL's NS IP listed in the SBL blocklist [URIs: splat.cc]
7.0 URIBL_BLACK Contains an URL listed in the URIBL blacklist [URIs: splat.cc]
-100 USER_IN_SPF_WHITELIST From: address is in the user's SPF whitelist
Zach Villers zachvatwork@gmail.com writes:
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?
What's wrong with the zfsonlinux repo? I don't run Fedora, but it seems OK on RHEL6.