Re: Bits (Nybbles?) from the Vancouver release team meeting
On Tue, Mar 15, 2005 at 07:13:39PM +0100, Marc Haber wrote:
> On Tue, 15 Mar 2005 11:17:25 -0500, David Nusinow
> <david_nusinow@verizon.net> wrote:
> >On Tue, Mar 15, 2005 at 08:04:31AM +0100, Marc Haber wrote:
> >> From what is public visible, the security team has lost at least one
> >> of the active members to ubuntu with no replacement up to today.
> >
> >We didn't "lose" him to Ubuntu. The man got a job and is busy. It would have
> >been the same with any other job that keeps one busy. There's no grand
> >Canonical Conspiracy [tm] to keep him from working on Debian.
>
> If his job is keeping him from working on Debian, he should step down
> from his post.
His position isn't preventing someone else from stepping up and joining the
security team. It's not like some package that he's claiming as his own and
letting no one else upload.
> >> And
> >> d-i's effort is practically wasted since we decided for d-i because we
> >> needed to support more architectures. If we had thrown out the
> >> doorstops years ago, we could simply have taken one of the readily
> >> available installers without rolling our own.
> >
> >No one is actually dropping those arches. They still have every option to
> >release a stable release using d-i.
>
> Do they? The announcement looks noticeably different.
My interpretation of the announcement, and this also comes from talking with
some of the people involved, is that this affords ports with the flexibility to
do as they please without slowing down the rest of the project. For *years*,
I've heard porters complain about ftpmaster and such. Well, now every port has
the full ability to take matters in to their own hands. They still upload to
unstable, just like always. Autobuilders for those arches still run, just like
always. These arches still have a host or number of hosts with sufficient drive
space to manage their port, just like always (although the url will be changed
to something as of yet undetermined).
The differences? Port packages don't go in to Debian mainline testing. However,
this does not preclude them from setting up a separate testing if they like.
The people involved with the Vancouver document know what they're doing, and
they've said (more than once when I've heard) that the unstable snapshot method
is better than setting up a separate testing, and I believe them. Also, RC bugs
that are specific to a port have a non-RC severity, although bugs are bugs and
should be fixed by the maintainer even if it's for a port. NMU's are still very
fair game in these circumstances. And when the time comes to release, it's
simply not the release managers' jobs to make sure the port releases alongside
the rest of Debian.
This does not preclude porters from making a stable release. In fact, all the
talk I've heard assumes that they will (via the snapshot method). Their ability
to work is no longer hampered by overloaded RM's/ftpmasters/whatever. I think
that when a port makes a stable release it'll be thought of as an actual Debian
release, especially when it sync's up with the mainline stable release. But
ultimately, it's not going to be the RM's jobs anymore to make sure that these
ports do release. Think of it as freedom (which it is) rather than exile (which
it's most definitely not) and the whole thing sounds much better.
- David Nusinow
Reply to: