[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: testing/unstable reccomendation



Joey Harrison <netcrusher88@yahoo.com> writes:

> I understand there are no security updates for testing/unstable,
> because they are so fluid, but what exactly is the difference
> between them

Developers upload packages to unstable.  Those are available to users
who follow unstable within 24 hours.  A package can move to testing,
in general, after it's been in unstable for 10 days, it has no
release-critical bugs, all of its dependencies are in testing, and
moving it to testing won't break any dependencies.  It's still
possible for a package to pick up RC bugs after it moves to testing,
though, and the ftpmasters have deliberately broken some dependencies
to allow some things to move into testing.

> and, between the two, which is more reliable and better for someone
> who wants bleeding-edge packages that have at least a safe level of
> critical bugs?

How do you define "safe", how willing are you to fix (potentially
serious) problems that come up, how much do you follow development
email, and how connected to the Internet are you?  Testing is less
bleeding, both in the "edge" and "potentially fatal wound" senses, but
gets no security updates at all.  Security fixes in unstable come from
normal maintainer updates, so they'll take some time (which can
actually be less than the week and a half but is still non-zero, and
could actually be much longer) to dribble through into testing.
Unstable is more likely to break in a way that requires a lot of work
to fix, but will also get the fixes into the distribution much
sooner.  Also, a lot of people have complained about the testing
dependency problems; I don't know if they've really been fixed or not,
but it's another thing that is likely to go better in unstable.

(If you feel competent to ride the bleeding edge, BTW, and want to
help fix things, you may very well want to track unstable.  But learn
to use the bug-tracking system [http://bugs.debian.org/], definitely
subscribe to debian-devel-announce and also consider subscribing to
debian-devel and debian-devel-changes, and don't complain too loudly
on debian-user when it breaks.  :-)

-- 
David Maze         dmaze@debian.org      http://people.debian.org/~dmaze/
"Theoretical politics is interesting.  Politicking should be illegal."
	-- Abra Mitchell



Reply to: