Ubuntu Is Planning To Make The ZFS File-System A "Standard" Offering

[mjg59]

as long it doesn't include portions of Linux itself, it shall not be concerned about GPL.

Whether that's true is an undecided question. Direct inclusion is not necessarily the thing that determines whether a work is derived or not. There's certainly an argument that code that depends on implementation details of Linux is a derived work of Linux. It'll probably take a court case to make a precise determination on that point, so making unequivocal statements about it is a little irresponsible.

[mjg59]

That assertion is one that has been told to me by actual lawyers. I have yet to meet a single lawyer who disagrees. So far, everyone who I have met that disagreed lacked a bar number and therefore is not a lawyer. Such people are far more numerous than actual lawyers.

I have spoken to actual lawyers who have launched actual lawsuits in this field and who disagree with your position.

[p91paul]

This is an… interesting assertion. It's certainly not difficult to find lawyers who disagree with it.

I believe instead it would be difficult to find lawyers who disagree with it: again, not as a lawyer, but the only arguable point here is whether the package Linux+ZFS is to be distributed as GPL or not.

ZFS on Linux, as a single project, is derived from Solaris code, and is subject on the limitations of the license of the code it is derived from; but as long it doesn't include portions of Linux itself, it shall not be concerned about GPL. I believe this is what ryao meant with the statement you quoted. The concern arises if you unify Linux and ZFS on Linux in a single project; ryao is also suggesting you can do that, and I see his point, but it is a more delicated issue probably.

[p91paul]

My last post was a late night one (at least in the CEST timezone), and it was flawed and not polite. I apologize to all users of this forum.

I believe the misunderstanding between me and ryao is about two different interpretation of "derived works", which are both correct. His fist though was derived works are changes made to a work, and as such, he is right, they must be under a GPL-compatible license. My first thought is a derived work is the modified work, i.e. the work together with the changes, and as such it must be distributed under the GPL. The only thing I should have said was "Hey ryao, I believe the chart unixfan2001 is showing, when it reads 'must use GPL for everything' about the right to modify means what happens if you take a project, fork it, and redistribute the fork (not the patch alone); then the modified work must be under the GPL". Much shorter, more polite, more clear and more correct.

About ZFS on Linux: my tone was so bad and was being so much an ass you asked me if I was a lawyer. I know you probably knew the answer, and you made the question because you are more polite and me so you assume someone else might be right, but no, I'm not a lawyer, just a developer who pretends to be a GPL expert. The other flaw here is that I quoted GPL3, which does not apply at all to the topic, since Linux is GPL2. My point was: I believe this is a grey area, there are no technical reasons to prefer merging the codebases, so it's a good idea not to do that. Like a friendly advice given with totally unfriendly words, you know. I would also avoid doing that after only one legal advice, I would ask many of them. Real experts (unlike me) on licenses always surprise me with the number of interpretation they can give to a seemingly clear sentence.

About my rant on you don't like GPL, use Windows: that's the worst part. It's incorrect, as you pointed out, and I knew it was incorrect. I was like a bull seing a red flag, and the red flag for me is people claiming GPL does not offer freedom, or comparing them to EULAs for non free software. I advice those people to actually read an EULA, then read the GPL. They might be similar lenght, and both in legal language; but the GPL always insists on what you can do, instead of what you can't. It just asks to respect the original author's intent when it comes to redistribution. That might be a requirement someone is not willing to comply, and that is perfectly fine; but not a reason to attack people who like that requirement. Surely, said from the person who attacked nearly anyone on this thread, this must be lame.

I would add something: someone is asking why to maintain ZFS and not switch anybody to btrfs; well, I think the answer is because this is the spirit of free software, or at least my understanding of it. A person needs a piece of software, has some free time, creates it; or, a person dislikes an existing piece of software, and rewrites something similar from scratch. Why don't they commit to something existing? because maybe they dislike some of the foundations on which the existing software is built, and so the modification would be too broad and would be rejected. Or maybe just because they want to create something new, which is theirs. As for maintaining old software, you can see it as the protection of endangered species: you like it, you don't want it to die, so you keep it going. I don't know the feature comparison of ZFS vs btrfs, but from other people's posts here and elsewhere seems like there are enough differences to keep them both alive and well; and even if that is not true, I'm not willing to ask someone to stop contributing to ZFS and start contributing to btrfs, since it's probably their free time they are committing to the project. It would be like asking someone who goes running in the park to stay fit: stop running, start cycling instead, bicicles are faster and newer. Yes but the guy likes to run.

[ryao]

Because GRUB won't have it. FreeBSD folk went through a lot of pain in the ass to get it done on their non-GRUB end.

There are plenty of people using Ubuntu with /boot on ZFS. It requires a debootstrap install and using fewer feature flags unless GRUB2 HEAD is used, but it works fine.

[ryao]

This is an… interesting assertion. It's certainly not difficult to find lawyers who disagree with it.

That assertion is one that has been told to me by actual lawyers. I have yet to meet a single lawyer who disagrees. So far, everyone who I have met that disagreed lacked a bar number and therefore is not a lawyer. Such people are far more numerous than actual lawyers.

[GreatEmerald]

You'll still need a separate /boot partition with for example ext4 - don't know about UEFI though (never used it). It's a bit of a pain in the ass but I have ZFS as root FS on a LUKS encrypted device on most of my systems.

UEFI requires a FAT32 file system to boot from.

[cyrix]

This is great news for btrfs. As always, when Canonical makes some tech announcement for their OS, everybody starts working on the alternatives. They already boosted systemd and wayland development some years ago and now it's time to get a serious filesystem.

[nils_]

Because GRUB won't have it. FreeBSD folk went through a lot of pain in the ass to get it done on their non-GRUB end.

You'll still need a separate /boot partition with for example ext4 - don't know about UEFI though (never used it). It's a bit of a pain in the ass but I have ZFS as root FS on a LUKS encrypted device on most of my systems.

[sthalik]

Why not?

Because GRUB won't have it. FreeBSD folk went through a lot of pain in the ass to get it done on their non-GRUB end.