Canonical Looking For Community Feedback As Ubuntu 22.04 LTS Development Begins

[Siuoq]

No, read it *again*. I said Snaps make sense for some things. Like Firefox, so the user always has an up-to-date web browser and not surfing the web unknowingly with security holes. (e.g. think grandma who may not run the Software Update every other day)

I don't understand, security related updates update silently, not through the Software Update, with all the random and unnecessary things. (I do remember, it partially crashed my firefox once, that it updated in the background.)

[down1]

Yes it is snapped by default, and I agree, there are no speed issues for me either. All my systems use SSDs though.

It works but the confinement isn't quite correct. It was complaining about ptrace in my logs and it has blocked VAAPI I think. If these are fixed I don't mind firefox being snap for the benefit of faster updates.

[matsukan]

This is not correct. The CDDL was specifically written and chosen to be incompatible with the GPL - likely to avoid code being used for Linux.
This is why even Linus Torvalds recommends against using ZFS:

Oracle could in any time decide to add GPL license compatible clause with current license model. Until then, openZFS can't merge to mainline Linux kernel codebase.

And it annoying that Canonical created listen to you channel and when people started complain about Snap packaging they decided to close that channel. Hopefully they really heard what users started to say: Stop this annoying snapping thing which will lead death of Desktop Linux as we know that now.

[reba]

Ok, can you explain to me how to remove the chromium-browser snap and install the .deb version? Because that would be great.

I'm also waiting for an answer to this.

[Developer12]

Which spyware? The Search Lense was removed quite a while ago.

I am shocked to see such a nonsense bashing ´for snaps, when we have the option to package applications that people want and that are missing as developers fear the amount of support they had to do for different versions of libraries and systems.

If you don't like them, don't use them. A handfull of duplicated libs wont kill performance.

These halfwits are regurgitating RMS rants from 15 years ago. They don't check if anything they say is valid.

[Maurits]

Oracle could in any time decide to add GPL license compatible clause with current license model. Until then, openZFS can't merge to mainline Linux kernel codebase.

And it annoying that Canonical created listen to you channel and when people started complain about Snap packaging they decided to close that channel. Hopefully they really heard what users started to say: Stop this annoying snapping thing which will lead death of Desktop Linux as we know that now.

To be honest, there is not much Oracle could do here.

Oracle could possibly (I don’t know the contents of the contracts when Oracle bought Sun Microsystems) re-license the closed source OracleZFS but that is a relatively little used ZFS implementation and rumour has it that Oracle has lost interest in it (and closed the department working on it).

Oracle could not re-license the open source OpenZFS implementation as they don’t own or control the OpenZFS implementation and as a third-party they can’t force the owners of the OpenZFS code to re-license their code. The only thing that Oracle could do is re-license the 20-40% of current OpenZFS code that Sun developed and open sourced in 2005.

As the majority of OpenZFS code was written long after that split (or the purchase by Oracle) you would need to ask the copyright holders of that OpenZFS code if they are happy to re-license their code. Those copyright holders include the Lawrence Livermore National Laboratory, Delphix, iXsystems, Nexenta Systems, Intel, Datto, NetApp and a whole host of individual open source contributors.

That means that if you’d want to re-license the OpenZFS implementation into a BSD license (the most likely candidate) then you need to get the many dozens of companies and individuals that have contributed to the OpenZFS code to transfer copyright to an OpenZFS foundation (or some such) and approve the new license. I think that is an insurmountable task and barely worth it.

OpenZFS using the CDDL license works well (currently on FreeBSD and Linux as Tier 1, soon also on macOS as a Tier 1 platform. Making Windows a Tier 1 platform is in the works) and allowing it to be mainlined into the Linux kernel wouldn’t add that much. All the enterprise players that combine OpenZFS with the Linux kernel just load it as a kernel module.

[Danny3]

I wish they would at least upgrade / update everything to latest versions !
At least the most important things like the Linux Kernel, Mesa drivers, systemd and the gaming related libraries.
As a Kubuntu user I hope to see updated Qt and KDE stuff.

[bemerk]

I wish they would at least upgrade / update everything to latest versions !
At least the most important things like the Linux Kernel, Mesa drivers, systemd and the gaming related libraries.
As a Kubuntu user I hope to see updated Qt and KDE stuff.

I feel you. Where do you draw the line though to have enough time to properly test the combination of the various components before the release?
If it's not stable and has bugs you missed, people will be unhappy. So a slightly older but more tested version might save you a lot of trouble.

[F.Ultra]

This is an urban legend. It's simply not true.

Not only was the Licence not written with any kind of GPL compatibility (or incompatibility) in mind, but NOBODY working on ZFS/Solaris was thinking about hypothetical linux compatibility. It was written YEARS before OpenSolaris was a thing. The CDDL was likely just the best compromise that Sun's legal team was willing to make when asked to draft a non-proprietary licence. The incompatibilities stem from subtle interactions of only a couple of terms.

(And really, if sun were that paranoid about competitors and that willing to warp their licence, you would have expected them to disallow linking against BSD code too.)

People push this myth because they're trying to project Free Software hostility onto Sun and portray them as $generic_big_evil_megacorp. It's a tactic to make people avoid everything they ever touched.



You should read up on the context of that quote. A *lot* of people tore a strip off of Torvalds afterwards for not knowing WTF he was talking about (OpenZFS's popularity in enterprise in particular; people sure as hell don't use BTRFS for RAID). The best guess is that he was thinking of Oracle ZFS, which is quite far from OpenZFS on both technical and political levels.

He's right that Oracle would sue the pants off the kernel if he tried to merge it into the tree. They sue the pants off of anything that moves. Nobody these days is suggesting he should do that or that it's even legally possible.

Aside from the SFC's bullshit argument that "all kernel modules are automatically derived works and thus must be GPLv2" (see: EXPORT_SYMBOL and EXPORT_SYMBOL_GPL, and of course nvidia) there isn't any reason you can't ship a ZFS module separately as eg. Ubuntu do and have for years now. It's also worth noting that Oracle themselves ship a CDDL linux kernel module for tracing.

Danese Cooper said in 2006 ( http://chuangtzu.ftp.acc.umu.se/pub/...pez_Ortega.ogg )

Mozilla was selected partially because it is GPL incompatible. That was part of the design when they released OpenSolaris. ... the engineers who wrote Solaris ... had some biases about how it should be released, and you have to respect that

And while SUN originally acknowledged Cooper as the author of CDDL (quote here from Simon Phipps)

...we have got Danese Cooper in the room, and she is the one who actually wrote the CDDL...

They (Simon Phipps and Bryan Cantrill) have since tried to paint a different picture where they since 2015 now claim that they instead where shocked that no one in the Linux camp would touch their CDDL licensed code. Which considering both the words of the license author and the license text itself sounds quite hollow.

[Danny3]

I feel you. Where do you draw the line though to have enough time to properly test the combination of the various components before the release?
If it's not stable and has bugs you missed, people will be unhappy. So a slightly older but more tested version might save you a lot of trouble.

Start with the upgrades early, like for example now, an then do it again each month until the end of february.
Bugs should be less and less by then and then one an a half month is enough for the few remaining ones.
They could even delay the release for a week or two to make sure everything is good.
I don't seen any point into releasing on scheduled time with an older kernel that is missing support for some hardware.
People with that hardware would've preferred to wait for a week or two and have it working compared to not working at all.