Re: GNOME 2.2 summary 22/06/2003
On Wed, Jun 25, 2003 at 03:09:34PM +0900, Miles Bader wrote:
> ??? I don't think that's true, in general. testing's not perfect,
> but it's not anywhere near `utterly broken.'
On the contrary, it's even more true in the general case. You can
*not* guarantee that a random package in testing is free of bugs, free
of security problems or even in working condition. That is not what
testing is, that I understand. But that's what people _perceive_ that
testing is. When talking to Joe User I ask why they are using testing
instead of unstable, and many times the answer is "because it's more
stable".
Like I said, the working assumption is that problems among packages are
going to be expressed as dependencies and conflicts. People have
provided plenty of examples where this assumption broke.
> For the most part testing seems like what it was intended to be -- a
> slightly old verion of unstable without unstable's occasional nasty
> bugs.
Ah...
Packages are installed into the `testing' directory after they have
undergone some degree of testing in unstable. They must be in sync
on all architectures where they have been built and mustn't have
dependencies that make them uninstallable; they also have to have
fewer release-critical bugs than the versions currently in testing.
This way, we hope that `testing' is always close to being a release
candidate.
"close to being a release candidate". That doesn't say anything about
it's relative age with respect to unstable or stable. It's just a set
of packages that can be used as an starting point for making a release.
In that sense, and to answer my own question, yes, testing has
fulfilled its purpose: right now you just need to iron out a couple
hundred bugs and you have a release. Or something like that.
I'd buy that if we'd make an honest collective effort not to let
unstable rot for years. Maybe for sarge+1 or sarge+2. You need to
start up with something that doesn't have the ridiculous ammount of RC
bugs that testing has.
> Because the problem cases (e.g. Gnome) seem to be the exception,
> presumably they could be handled with a few extra knobs somewhere.
http://ftp-master.debian.org/testing/testing_probs.html
"Exception" is one or two, not one or two hundred.
-m.
Reply to: