Re: more and more broken packages?
On Wed, Jan 22, 2003 at 02:41:19PM +1300, Nick Phillips wrote:
Oh do behave. Testing is a perfectly useful distribution for a machine
which needs things which are not in stable, is being run by someone who
doesn't want the extra hassles of tracking unstable, and is either not
in a situation where security is a serious problem or could take
packages from unstable when security updates dictate.
And who doesn't particularly care if the software they're using works.
Security aside, the packages in testing right now have huge bugs that
have no reasonable possibility of being fixed anytime soon. Testing is
simply not a good idea.
Installing packages from unstable may work and may not. Depends on the
package--in some cases it will be simple and in some cases it will be a
hideous mess.
I get tired of people evangelizing testing. They aren't the ones who
have to deal with the complaints about why no security updates are
available. (Usually in the form of an email along the lines of "what do
you mean there's no security update. I'm running many servers on testing
because I was told it was the best combination of stability and
features.") They also aren't the ones who have to deal with the
repetitive bug reports *that are for things already fixed in unstable
but unaddressable in testing.* (I increasingly just want to toss bug
reports from testing--they're fairly well useless at this point in the
development cycle.) Most of all, it annoys me that people advertise
testing as the thing they wish or think it is, rather the thing that it
actually is.
Testing is an internally consistent set of packages, nothing more. It
isn't a set selected for any other attribute, such as security or
stability. It's just a collection of packages capable of being
simultaneously installed from a snapshot of unstable. Oh, right, with
the additional caveat that a package undergoing a lot of bug fixes won't
hit testing because of the propagation delay, so nasty bugs will stay
longer in testing than in unstable, as a rule.
Mike Stone
Reply to: