Re: Sarge: GCC 3.4 in Sid
* Martin Michlmayr [Thu, Aug 05 2004, 03:47:34PM]:
> > Our current approach to release management came to be precisely
> > because the process of freezing unstable and trying to release it
> > was such a hideous mess, even when Debian was much smaller in scale.
> I don't think Eduard suggested freezing unstable and doing away with
> testing, but keeping the testing model and freeze testing plus
Yup. I did never say "freeze for release", just freeze. Do not add new
packages to Unstable. Instead, create a small crew that decides which
package are allowed to enter Unstable (with which level of changes) and
which are not. "central" maintainers (eg. libtiff/gnutls11 maintainers)
can pass a list of "allowed" changes in dependent packages. My imaginary
task force (say, 9 people) would decide which changes should be allowed
and which not. When the package sets in Testing and FrozenSid look close
enough, Testing can be released and Sid "defrosted" again.
Why all this? I still see the current system as pure chaos and doing
Sysiphus work. And the technical committee needs to much time to find
a decission, so it would be more practical to decide less important
<Zugschlus> fIPS: Wenn Du mich f�bl��st wie ich aussehe dann
hast du genau den richtigen vor Dir.