list migration due to alioth shutdown

[Russ Allbery]

Peter Schober <peter.schober at univie.ac.at> writes:
> * Russ Allbery <rra at debian.org> [2018-01-30 05:13]:

>> My understanding is that lists.debian.org is not set up well for
>> creating dozens or hundreds of lists, so listmaster has said that they
>> can't handle package team mailing lists.  There are just too many of
>> them.

> So where does Debian expect all those package team mailing lists to go?
> I.e., whose problem does Debian intend to make this, making sure
> coordination works between Debian Developers, Debian Maintainers and
> volunteers trying to help out the Debian project?

Yeah, this has been the most controversial part of this whole migration.

The root of the problem is that the Alioth framework as a whole is not
maintainable, and is being replaced, but the folks who volunteered to work
on the replacement were primarily interested in the VCS hosting problem.
So that's what they set out to replace, rather than the mailing list
hosting.

The mailing list hosting was basically working, but it was part of the
same overall framework, so it's hard to keep it running independently (and
the folks who were running it don't want to keep running it).

Meanwhile, the Debian listmasters have always been reluctant to create a
ton of @lists.debian.org lists, which is why everyone started using the
Alioth lists in the first place.  There are a lot of issues around that,
including some software issues, but my understanding is that their root
concern is that the @lists.debian.org lists aren't designed for
decentralized management.  There isn't a way to delegate list ownership to
other people or distribute the work, so it would mean taking on a ton more
work for central listmasters to move all the lists there.

Mailman is basically the correct model, hence alioth-lists.debian.net,
although I'm not sure why it's not becoming an official project service.

> Clearly alioth-lists.debian.net is not the answer to this, at least not
> if one isn't hell-bent on repeating this excercise in 1-2 years time.

Yeah, I have concerns about that too, but it seems to be what we've got
for right now.  My guess is that the level of demand and popularity will
mean that it will stick and we'll find some long-term solution.

-- 
Russ Allbery (rra at debian.org)               <http://www.eyrie.org/~eagle/>