Guest post by On his blog, Mark Shuttleworth has outlined a proposal to change how Canonical handles Ubuntu releases. In the proposal, future LTS releases will receive new kernels and software (something we’ve already seen in 12.04.2 which had a backported Xorg stack from Quantal) and interim releases will only be supported for 7 months instead of 18. Of course, the current situation where you often have to upgrade the whole OS just to get new software and drivers isn’t great, so Canonical might be on to something here.
I know I’m preaching to the choir, but this sounds like what FreeBSD does, with separation of the base system from ports. The system itself is stable between point releases (sometimes drivers are backported from rolling -CURRENT), but ports can be as bleeding edge as possible (allowing a few weeks for Linux-isms to be worked out). The result is the ability to use recent software without breaking the entire system in an upgrade.
The problem doesn’t seem to be so much the kernel in this case, as Canonical. My experience has been that despite already being quite bleeding edge, Ubuntu’s practical hardware support has been horrible, even when compared against distro’s with older software.
When I tested Ubuntu on my AMD Fusion, graphics were scrambled (on Opensuse/fedora/everything else however, they were fine). On my Intel, the audio drops out, but with a workaround, and on every other distro, works fine. Whilst Ubuntu had hibernation issues on 2 systems I tried, hibernation worked fine on other distros. How is it that other distros, often with older kernels, don’t have the same problems as ubuntu?
Part of the problem is the kernel, and software, however, Canonical’s problem seems to be that they are burning all their resources on creating projects, which already have perfectly good alternatives (Bazaar vs Git, Unity vs everything else, Systemd vs whatever, wayland vs their plan). The end result seems to be that they are spreading their resources too thin, and don’t seem to have time perfecting and testing. Canonical needs to rethink what they are doing, start reusing the wheel more, and perform more testing.
I admire Shuttleworth a lot, and the Ubuntu developers, but, if they continue down the path they are going, they will degrade the public view of Linux further, and hinder development / innovation, because neither their projects nor competing projects will build the infrastructure to support innovations.
Ubuntu support.? You actually need support? Go back to Winblows..
Recent releases of Ubuntu have been very buggy. I wish that they would concentrate on getting the quality up (as amply described here and in many other forums) rather than fiddling around with this sort of thing.
However given Canonicals desire to be all things to all men (jack of all trades, master of none) I doubt that this will be the case which saddens me.
Ubuntu was once regarded as a breath of fresh air. Releases were solid and eagerly awaited. This is no longer the case and they don’t seem to care (from where I sit on the outside)
But this sort of thing (rolling release) is about trying to get quality up.
I don’t see how a rolling release would bring quality UP?!
My OpenSUSE 12.3 installation isn’t perfectly stable and if I wanted something more stable I wouldn’t pick something which moved even faster but rather something slower like Debian.
Upgrade what’s broken or need security fixes and keep what works. If you replace what works with the latest and greatest you’re risking stability.
“Slightly more rolling” doesn’t roll off the tongue very well. “Semi-rolling” sounds too passe.
Lets invent a new term. How does “lumbering-release” sound?
Leela: Wouldn’t it work better if the wheels were round?
Fry: It’s my invention, we do it my way!
Maybe so. a rolling release cycle has been discussed before at Canonical. But so far..??