Monthly Ubuntu Releases Proposed
An anonymous reader writes "Scott James Remnant, the former Ubuntu Developer Manager at Canonical and current Ubuntu Technical Board leader, has proposed a new monthly release process for Ubuntu Linux. He acknowledges that with the six month releases there are features that end up landing way too soon, leaving them in a sour state for users. With his monthly proposal, Remnant hopes to relieve this by handling alpha, beta, and normal releases concurrently. It's unknown whether Canonical will accept the policy at this time."
Rolling releases are great for devs because it lets you put your new feature into the release cycle when it's ready instead of locking it down in whatever state if you don't want to miss the 6 month cycle.
The trouble is that this is terrible for users. The 6 month cycle is already a little aggressive (but tolerable) on support forums. Monthly releases would cause so much confusion when you're searching for other people who have experienced your problem.
Also, how does the support cycle work? Are you going to provide parallel support for 24 releases for two years? If not, do I have to upgrade monthly? I support too many computers for that to be a realistic option.
i know what! let's go back to releasing "when it's ready"! that would be great! oh wait... that's what debian do.
Debian does have rolling releases, it's called Testing and Unstable.
Dilbert RSS feed
Because who is going to work on last month's version? "Oh, just upgrade you'll get all the new fixes." And all the new bugs.
Bleeding edge is fine for hobbyists, but grown ups? We need a version that's going to start solid and get steadily better.
If you were blocking sigs, you wouldn't have to read this.
It's better to be warned by the name up front than learning it the hard way as with Ubuntu.
The Tao of math: The numbers you can count are not the real numbers.
Those names hardly inspire confidence for use in a production environment..
Quite. If they were a respectable company they'd call their test rolling release something like "Windows Update with Genuine Advantage"
I might not have been clear. I use Ubuntu (LTS) as a fire-and-forget (for three years) installation for non-tech-savvy users (Read: my mom, my mother in law... You know *those* kind of people). Personally, I do manage to run a Debian installation and once you do set it up like you want it, you'll be fine. However, you can't use Debian as a fire-and-forget installation if you want things that people require from their modern day desktops. This is mainly due to the "free-at-all-costs" stance.
That's fine, I understand that it's a good thing, and I can live with it. Try to see it from a user perspective though. I trained them for years (even while still on Windows) that they should use Firefox for browsing and Thunderbird for email. The migration to Ubuntu was easier because of this knowledge. My support calls are next to nothing ever since I switched them to Ubuntu LTS. That's how I want it, and I know the distribution will keep itself healthy. Do I have the guarantee with Debian backports? Does it get as much "love" as the main branch? I sincerely doubt it. Going with official Mozilla downloads is a no-go, as I'd have to login remotely to their system to update it every time I hear of a greater security issue.
There are other things, like for example the only large support call I had this year. That was Ubuntu, so it probably wouldn't have been avoidable at all. My mother in law had a big issue with a PDF. Turned out it was a PDF with a form and the built in PDF reader (evince, I think) didn't handle that. At least, I could remotely login and install Adobe Reader from the repository. I know Ubuntu has it. Debian might in the non-free section, but I'm not sure.
Sometimes it's the small things. I happen to be multi-lingual. In Ubuntu there is a great tool in "System"-"Administration" called "Language Support". It's basically a hub for anything language related: Want the interface in Dutch? Only want the German spellchecker? It's there, it's a click or two and it's installed. Debian simply doesn't have an equivalent (or I didn't find it).
While I agree that Gnome2 is great and Unity and Gnome3 are definite steps backwards, the Debian themes do look a bit dated. I can live with it. It's fine, I found that the "Shiki" theme is great even though I prefer a light-theme. On Ubuntu Radiance is what I use and I love it. Still, for me, lacking compiz and a bit dated theme is okay. However, my users are used to the polish Ubuntu gives. I'm, pretty sure my users won't miss compiz if I'd take it away, but the polished themes are something else. We know it's just eyecandy and not important, but how would you feel if you're used the the Windows 7 interface (which I dislike, but that's not important) and gave you a Windows 2000 interface (Which I loved). You'd probably wouldn't be happy (I'd be, but put yourself in the shoes of a non-IT user).
It's lots of these little things. I'm certain it's completely because of my inability and incompetence. I'd be glad to read a how-to for achieving just that: have a modern multi-lingual, proprietary-software, friendly, Linux desktop that doesn't look like made in 2000 which I can install and forget for three years.
I actually wrote a bit about trying to achieve this. Feel free to read it: You don't realize how much polish Ubuntu provides... and Backports is the magical word.... and finally I have to give Ubuntu 11.04 some slack., which I need to include because it shows that the problems I had with Ubuntu aren't limited to Ubuntu itself.
Ahhh...the great dumpster continuum. Many a free computer will be found there. -- sowth (748135)
"What I fear, is that the proposed shorter release cycles are going to make Ubuntu break too often. That will turn off users, and they cannot afford to lose even more users after the 11.04 release."
That's not unreasonable or irrational.
If you folks will forgive a geezer, I was doing software release management, testing and version control long before most of you were born. I've watched with interest and occasional amusement while you kids have managed to relearn much of what we learned rather painfully in the 1960s. And I'll give you credit. PC software works better than I would have thought possible given the way you approach it. And by "you" I don't mean just Ubuntu, or just open source, Microsoft has quality problems also.
Nonetheless, I gave up on Ubuntu and its spawn years ago -- mostly because of quality issues. Apt-get is wonderful ... if the stuff that is apt-gotten works. Too often it didn't. It appears to me that the problem is that software gets captured, locked down, and released without adequate testing.
Anyway, three thoughts:
1. Rolling releases probably are not a good idea except for really critical fixes. My experience (which I agree may not apply to your world) is that you really, really need to consolidate a release, then test it thoroughly before inflicting it on users.
2. It is perfectly possible to do releases in parallel with several in different states of production. Developers don't like it. So what? What matters is user experience, not developer inconvenience. But there is a limit to how many parallel products you can keep straight. And it is not a large number.
3. In the world I worked in, there was a minimum time required to consolidate and test a release. For us, it was 8 weeks. We tried 6 weeks (once) and couldn't do it. Your world is quite different, but I'll bet you have a minimum time also, and it may well be longer than one month.
You can't see ANYTHING from a car, You've got to get out of the goddamned contraption and walk...Edward Abbey
THAT'S IT!!! I'm moving to Gentoo!
May the Maths Be with you!
I see two problems with that, both at the core of Linux.
- Closed source driver support.
- The Unix/X11 (and thus Linux) kernel model doesn't allow for a high enough level hardware abstraction. So the desktop environments and applications have to do a lot that should be in the core OS.
Than how does OSX do it? And Solaris? And HP-UX, AIX, and the others I'm too lazy to mention?
This is an open source problem, not a unix problem. Commercial UNIXes which also use X.org seem to have no problem what so ever.
It all comes down to who WANTS to do it. OSS will remain second class because no one wants to do the hard work, like testing and making things stable. Everyone wants to just do the new shiney feature.
Persistent Volume manager for Kubernetes - https://github.com/dwimsey/openshift-pvmanager