Re: Understanding the two-year release cycle as a desktop user (and a Debian newcomer)
On Tue, 25 Feb 2020 16:55:46 +0100
Sam <debian@samueldgv.com> wrote:
> Hello,
>
> I would like to hear opinions about the release cycle of the Stable
> Debian releases for a Desktop user.
>
> I love the Debian ideals and perks (its social contract, independence
> from big companies...) and understand to a certain extent the
> fundamentals on why keeping "old-ish" versions of packages with
> backports and the Shiny new stuff syndrome, but I fail to see how
> Debian can make a useful desktop distribution with the current
> release cycle.
>
> For example: My main PC is an already two years old ryzen-based
> system and a Vega graphics card from 2017, and the kernel used in
> Stable has regressions which cause complete, unrecoverable system
> hangups on Vega cards which were not alleviated until kernel 5.3
> onwards (and they still keep happening, though rarely!). This means
> that to ensure stability on a Debian installation I would need a
> backported kernel, or use Debian Testing or Sid, which IMO collides
> with the point of a Stable release.
>
> I also see everyday many announcements about performance (GNOME) and
> usability (KDE Plasma) improvements which are not exactly new
> features. This is obviously happening on more recent releases, which
> Debian may not see (unless these changes are also backported, which I
> would find extremely cumbersome?) until approximately two years have
> passed since that.
>
> All this makes me think that while Debian is a fantastic
> distribution, its Desktop, common user-facing side of things would
> greatly benefit from something like a separate yearly Stable release.
>
Software is always a trade-off between age and stability. If you want
new software, you must accept that it will have more bugs than older
software (providing it is widely used, and people report bugs). Some of
these bugs will have security implications.
After Debian Testing is frozen, it undergoes six months of concerted
efforts to fix the bugs. But six months out of every year would mean
the developers not doing all that much development. After release, only
security bugs (pretty much) get fixed. Until the freeze, Testing is
only about two weeks behind Unstable, so if you want the latest
software, you could choose either, after the freeze, Testing will
obviously fall further behind. Testing is generally a little less buggy
than Unstable.
I use both options: Stable for my server and netbook, Unstable for my
workstation. I don't use Gnome or KDE desktops anywhere (though I use a
few of their applications), I use Xfce on netbook and workstation,
command line on the server. I'm really not bothered about whizzy window
animations and other special effects. The applications run the same
whatever the desktop, in fact a little bit quicker if you don't waste
resources on visual gimmicks.
Finally, the unique selling point of Debian Stable is that you can
*always* do an in-place upgrade of one version to the next. That means
a lot to people running servers, though they also like that their
automation scripts don't break after routine upgrades. Windows hasn't
done major version upgrades since Win 2000, (except the trivial 8/8a/10
upgrades) and I don't think many other Linux distributions offer it.
It was originally hoped to release at one-year intervals, and the
current roughly two-year interval is not set in stone, it's just the way
things have worked out. It seems to be a good ratio of development time
to bug-fixing time. Given a large number of new (unpaid) developers,
the cycle could be faster.
--
Joe
Reply to: