ZFS Targeted change against OpenZFS on Linux?

If I'm recalling correctly, "Linux" has done this before. Taken an exported function used by ZFS and bringing it back inside so it was no longer visible. Mailing lists (LKML) at the time were basically "eff it who cares ZFS sucks". So there is precedent for the hostility towards ZFS.
But do these changes that appear to be targeted at ZFS affect any other third party filesystem that someone wants to create? Or because of these changes, anything they do MUST be GPL?

For a long time it seems as if Linux has been antagonistic towards anything "not GPL"; moving towards "If it's not GPL you will not be able to work with the kernel".
 
First, don't ascribe to malice what can be explained by incompetence.

My goodness, NO. Who cares that ZFS becomes a "first-class" Linux filesystem? Quotes intended, of course. It's their problem, not ours.
There is no "their" versus "our" here. I use a variety of operating systems, including Linux and FreeBSD. It would be nice if Linux had a good ZFS implementation, since in my opinion, ZFS is the best freely available file system right now (and has been for a long time).
 
ralphbsz good point/perspective. I think the problem some of us have is we want to try other things (Linux, NetBSD, OpenBSD, DragonFlyBSD) but we have some level of investment in FreeBSD like having home directories or other critical infrastructure on ZFS. It then becomes "I want to try DragonFlyBSD but don't want to copy everything in my home directory, so can I do ZFS?"
I use DragonflyBSD as a hard example because of it's history with FreeBSD. I think there have been some good ideas there but last I checked (at least a year ago) you can't do ZFS on DragonFlyBSD (if this has changed PLEASE let me know). I would love to have ZFS available on everything (even BeOS).
 
There is no "their" versus "our" here. I use a variety of operating systems, including Linux and FreeBSD. It would be nice if Linux had a good ZFS implementation, since in my opinion, ZFS is the best freely available file system right now (and has been for a long time).
Alt Linux (at lest Sisyphus, that I use) have quite good zfs support, kernel module is in their default repo and is trivial to add (they even have GUI app for that - alterator / System management center). Although, this applies only to 6.12.x branch (6.12.47 is latest), but again it's same for nvidia and many other modules that are not in their 6.16/6.17 yet.
 
The Linux community has a long track record of being hostile towards their counterparts. There certainly is a "theirs" vs "ours". They reap what they sow.
Exactly my point. I moved away from Linux on my personal systems just because of this and on FreeBSD I use non-GPL software whenever I can. I could definitely do better in the DE-WM space but I like KDE too much so far...
 
Already worked around:


 
Not surprising at all. This wasn't a major change and the old struts were still available to be implemented and tested.

From what i see don't even expect penalties on the performance size but I'll leave it at when Michael will chime in.

Anyways, my point was to raise an heads_up and so far I'm proud of what could be achieved in such a short time by the ZFS community. Life can be good even in hospital bed.
 
Now, I've fear. I'm using zfs on all my systems, freeBSD and 3 Linux distros. I don't want change my filesystem because zfs is great
 
if that will be the future of zfs on Linux, I must dump. But I hope no. Actually I don't use freeBSD ad Daily driver only because on bsd I don't have a little bit of programs and driver
 
Rob N already worked around the new roadblock. No need to panic for now.
But it does make one (at least me) wonder "What next"? Is it possible the changes could affect other filesystems (even ones not yet created)? Are the basic filesystem needs public (what used to be known as VFS layer I think)?
I think (vague memories) adding ZFS in on FreeBSD caused some changes in UFS because of how deep UFS was in a few different kernel subsystems. End result was a bit of code cleanup/reorganization (some may say ObjectOriented thinking) on how filesystems should integrate.
As long as Linux folk don't wind up "privatizing" all the needed interfaces, it may wind up creating more portable openZFS code.
But I'm not sure I trust anyone anymore (except for my dogs)
 
Back
Top