Ubuntu moving to rolling release, how will this affect Kubuntu?

[woodsmoke]

dumpe2fs will provide a date for the original installation at least with ext 2,3,4
woodsmoke @woodsmoke-desktop:~$ dumpe2fs
dumpe2fs 1.42.5 (29-Jul-2012)
Usage: dumpe2fs [-bfhixV] [-o superblock=<num>] [-o blocksize=<num>] device
woodsmoke @woodsmoke- -desktop:~$

[Dutchman]

I just finished watching the UDS G+ Hangout with Jono Bacon and other Ubuntu developers going over how to switch to a rolling release. Very interesting. Here's a link if anyone wants to see it. The good news for now is that Jonathan Riddle (Kubuntu lead dev) did get a yes from them that 13.04 will be a regular release. It was debated and a lot of the devs wanted to start there, but they acknowledged there's too many other *buntu's and derivatives with too much invested in this release to change now. In Jono Bacon's words, "We don't want to pull the carpet out from under anybody." But I think we can safely conclude it will be the last regular six-month release.

It seemed from the questions Jonathan had (via IRC) that right now Kubuntu would rather stay as is, with six-month releases, and he asked about how the change to a rolling release would affect the repositories (like /backports and /beta). That, they didn't answer.

To me it would make no sense for Ubuntu to keep those types of repositories with a rolling release. It seems like many of the individually maintained ppas would also be brought "in house", no need for them if you're keeping software in the rolling release up-to-date. I think this is one of the bigger reasons Canonical is wanting to go rolling, they can give many of these ppa maintainers access to the main repos, freeing up there own packagers for things elsewhere. It's a win-win for Canonical.

I think all the other *buntus will have to follow the rolling release model, or somehow come up with the manpower to do their own thing. What will be really interesting is what the spins do, like LinuxMint, Netrunner, Voyager, etc.. Mint already has a rolling release with their LMDE, will Mint become rolling too? The next few months are going to be interesting indeed! If anything, Canonical sure knows how to shake up the Linux world.

[kubicle]

If anything, Canonical sure knows how to shake up the Linux world.

Yes..."shake up" as in "shake up the baby". :P

[PaulW2U]

I think all the other *buntus will have to follow the rolling release model, or somehow come up with the manpower to do their own thing.

Kubuntu's problem is maintaining and supporting multiple versions of KDE.

At the moment, well until April, KDE 4.7, 4.8, 4.9 and 4.10 are all being supported, maintained or being prepared for release (in addition to whatever was in 10.04, the previous LTS).

If there is only going to be an LTS and a rolling release then what version of KDE would the rolling release include?

The current LTS has KDE 4.8 with later versions back-ported. Would the rolling release, if it were already in existence, include the stable 4.9.5 or the most recent 4.10.0 (much of 4.10.1 is still in -proposed) which for me anyway has a number of minor bugs?

Jonathan Riddell has recently stated that Kubuntu versions, when released, include a stable KDE 4.x.2. Presumably this policy will have to change.

[Dutchman]

Yes..."shake up" as in "shake up the baby". :P

I think it will be worse for Canonical. They're potentially going to lose a lot of the commercial involvement they currently enjoy, System76 for example. A computer manufacturer doesn't want to be stuck on an LTS release with shipped products because it will be outdated within a month or two. And how are they suppose to offer support for a rolling release? Then there's proprietary partners like Steam. They're stated reason for choosing Ubuntu as the launch platform was it's brand recognition and user base. I see a large portion of that user base jumping ship when they go to a rolling release. What do you think their partners will do?

Canonical's reasoning for wanting to go to a rolling release is to free-up developer time so they can concentrate on making their dream of "convergence" happen. Having one OS for every platform. It's certainly possible to do, Plasma has shown that. But who are they going to get to put it on hardware when they've developed a reputation for cutting their partner's throat?

When I wrote "shake up" it was only for being polite on the forum: I thought it more appropriate that "f**k up", thought that would have been more accurate. As for being "interesting", I think it will be. It's going to be tough for the *buntus, but the one I'm going to watch closely is Mint. They're actually in the best position to benefit. They have the largest user base (it's a constant toss-up between them and Ubuntu) and have shown that they can innovate on their own. They're the most likely harbor for these partners that Canonical is going to lose. With LMDE they've proved they can work with and directly from the Debian builds, and if they suddenly find themselves wooed from companies with some money and muscle to invest, they could wean themselves from Ubuntu and do their own Debian spin. I feel Mint is going to be the one to watch in all this.

[Dutchman]

Kubuntu's problem is maintaining and supporting multiple versions of KDE.

At the moment, well until April, KDE 4.7, 4.8, 4.9 and 4.10 are all being supported, maintained or being prepared for release (in addition to whatever was in 10.04, the previous LTS).

If there is only going to be an LTS and a rolling release then what version of KDE would the rolling release include?

The current LTS has KDE 4.8 with later versions back-ported. Would the rolling release, if it were already in existence, include the stable 4.9.5 or the most recent 4.10.0 (much of 4.10.1 is still in -proposed) which for me anyway has a number of minor bugs?

Jonathan Riddell has recently stated that Kubuntu versions, when released, include a stable KDE 4.x.2. Presumably this policy will have to change.

That's what Riddell asked at the UDS Hangout, and didn't receive a real reply. I got the feeling that Ubuntu is going to just let the *buntus figure out how to deal with it on their own. Jono Bacon's comment was they would go ahead with 13.04 because they "didn't want to pull the rug out from under" the spins, but beyond that they had no answers for what the spins should do, which leaves me to believe they're adopting a "sink or swim" attitude towards them.

[PaulW2U]

http://summit.ubuntu.com/uds-1303/me...contingencies/ at 1815 today may reveal a little more about Kubuntu's plans for the future.

The Xubuntu project are similarly affected. I've not seen anything from the Lubuntu team yet.

[Simon]

Well just FYI Dutchman I did view a good portion of the video, I'm sorry but could that group get any more nerdy? I find System 76 has a very interesting point made. If they are shipping PC's with an out dated OS on it, they will end up losing potential customers. In fact, I wouldn't be surprised if they abandon ship and head for Gnome Red Hat or some other distro. While as I stated before, this doesn't bother me which path they take, it will matter to future developers, businesses, and some end users. I will be following this with interest all the same. However I disagree about the view that "LTS must die" on his blog. I like LTS that is the prime reason I dropped XP, Microsoft no longer supports it. Meaning eventually all my potential future software needs would end up requiring Vista or higher.

Thanks for the information Dutchman!

[oshunluvr]

Personally, I don't think Carl Richell's (System76) opinion is correct. Yes, 12.04 LTS was released in Dec 2012, but I still get updates weekly and if you download and install the newest iso you get 12.04.2 - clearly a newer version. They're worried about the nomenclature instead of the reality. Windows XP was released in 2001 and the last "Service Pack" in 2008, yet 1000's of companies still use it and that didn't stop it's popularity. What he wants is a name change along with every update. I seriously doubt his customers are so dumb that they would be fooled by that or need to be placated that way.

Besides, it wouldn't take that much effort to test each new release on their hardware and offer Newest version or LTS as an option when you buy. Seems like corporate laziness to me. Oh, and doesn't Ubuntu have an option to allow you to install the newest release via an upgrade? Duh...

At this point, I think most users who select linux as their OS of choice understand they're not buying Windows. I'm a little surprised that Simon changed to linux only because XP's support will stop next year. I assumed Simon was more "advanced" than that!

[unheeding]

I think a rolling release would be great. I'm already rolling (mostly) with raring right now. I just wish they would test some of the things before pushing them out (kscreenlocker_greet I'm looking at you!)

[eggbert]

I never really understood why we even need a 6 month or LTS release schedule for a Desktop OS. For a server it's a great thing, but for a desktop use case, linux package management/release schedule seems fundementally broken. Maybe I'm a dimwit, but why the heck can't I get fresh versions of desktop end-user software like LibreOffice, GIMP, etc? Dependency hell...

But whose problem should that be? The distro or the app developer? I would argue it's the app developer's problem. Why should cannonical, or any distro, have to "bless" certain packages like GIMP to be in the repo. Either they meet the core dependencies the system provides, or they don't get to be in the repo.

Why not have a "core system" with the kernel, libs, all the plumbing etc, and then 3rd party projects target THAT core base system. The way its currently done seems so counter productive with all the upstream/downstream coordination. Look, if someone provides a "sytem" and I want my app to run on that system, I'll make damn sure it works with whatever dependencies it provides. Cannonical should be setting the rules here, and 3rd party app developers should be ensuring their apps work with it, packaging their apps for it, and updating it themselves.

Why is this not workable in desktop linux land?

[GreyGeek]

When I installed 12.04, the first 5 year LTS, I decided to go with it until the next 5 year LTS was released, probably in 17.04. I have the KDE PPA, which is keeping my KDE updated, and I'm currently running 4.10.1 with no problems. Five years a LONG time, and a lot of things can happen between now and then. Kubuntu 12.04.x is working perfectly, and as the old saying goes, if it isn't broken don't fix it.

[NickStone]

If your the type of person who adds ppa's of most of the packages installed to your repos' then in effect you already have a (semi) rolling release ubuntu based system.

[Simon]

At this point, I think most users who select linux as their OS of choice understand they're not buying Windows. I'm a little surprised that Simon changed to linux only because XP's support will stop next year. I assumed Simon was more "advanced" than that!

Not the only reason I dropped XP altogether but was a large part. I would hate to be the guy out there holding onto the Windows 3.11 running under MS DOS with the delusion there will be a revival. I am a 3rd year medical nerd, shooting for a Phd. When I do get there I hope to see Kubuntu support with me, not because I would need it as much as the staff would. But no I am not an "advanced" computer nerd, I am trying. My field of study is going to demand it since genetics is basically programming humans. LoL Now if we can get a computer running on DNA I would kick some ass.

[Achaean]

I never really understood why we even need a 6 month or LTS release schedule for a Desktop OS. For a server it's a great thing, but for a desktop use case, linux package management/release schedule seems fundementally broken. Maybe I'm a dimwit, but why the heck can't I get fresh versions of desktop end-user software like LibreOffice, GIMP, etc? Dependency hell...

But whose problem should that be? The distro or the app developer? I would argue it's the app developer's problem. Why should cannonical, or any distro, have to "bless" certain packages like GIMP to be in the repo. Either they meet the core dependencies the system provides, or they don't get to be in the repo.

Why not have a "core system" with the kernel, libs, all the plumbing etc, and then 3rd party projects target THAT core base system. The way its currently done seems so counter productive with all the upstream/downstream coordination. Look, if someone provides a "sytem" and I want my app to run on that system, I'll make damn sure it works with whatever dependencies it provides. Cannonical should be setting the rules here, and 3rd party app developers should be ensuring their apps work with it, packaging their apps for it, and updating it themselves.

Why is this not workable in desktop linux land?

I agree too. This is one of the major problems at my opinion, which is preventing Linux from getting an everyone's desktop system ( and not only for experienced users).
This infamous "dependencies hell" problem, coming out from it's earliest days.
Still no real answer to this problem.

Being a little heretic to the Linux spirit, I like the classic-UNIX approach (that BSDs has), of package management (with PC-BSD being the most user friendly, at my personal opinion). One program, one package.
On the other hand, BSDs has other serious problems for desktop using, being outdated and most of all, lacking major GUI utilities (like live partitioning and image backup).
At my system, I'm using GParted for live partitioning and Clonezilla for image backup.
I doubt if they'll port to BSDs any time soon, due to EULA philosophy reasons.
Of course there are workarounds, but not for everyone and for sure, far away from being a " desktop solution".