ArchiveReorganisation and sponsoring

Emmet Hikory persia at
Mon Sep 1 04:49:39 BST 2008

Scott Kitterman wrote:
> In my experience in Debian working on getting sponsored for team related
> uploads versus general uploads via mentors, it quickly became apparent that
> my odds are subtantially better in a team environment.

    Indeed.  Except in rare cases, those working with a given set
packages are significantly more likely to also be interested in
integrating submitted changes or sponsoring improvements.

> I expect it will be similar here.  I believe that leaving UMS/UUS in place
> for general sponsorship and encouraging individual teams to develop their
> own approach will be sufficient.

    My concerns with this model are:

1) Unless "universe" is restored to the previous definition of
"unseeded" packages, a number of packages will fall into "universe"
that cannot be sponsored by the Masters of the Universe.
2) As many additional packages would be added to the UMS queue, it may
add to the load for those primarily responsible for triage of the

    Problem #1 ought be fairly sensible to solve: as the
archive-admins will already need to be reviewing any package that
moves into seeded management, it just becomes a greater definition of
"main".  The requirements for any given package to enter this
redefined "main" may well vary depending on the seed, the name may
well change, and the nature of the ogre model may need additional
consideration, but these are all factors that apply to
ArchiveReorganisation in a way not specific to sponsoring.

    Problem #2 could be resolved through greater proactivity about
sponsoring by the maintenance teams, and development (and publication)
of preferred workflows for packages of interest to these teams.

> Premature optimization is a sin in process design too.  I'd suggest some
> basic initial approach like this will be sufficient to get us started.  We
> can see how it goes and optimize from there.  I think many of the problems
> you're concerned about will sort themselves out as sponsorees figure out
> where the sponsors are.

    Part of the worry is about how that is to be communicated.  When
someone is very new to UbuntuDevelopment (although perhaps deeply
familiar with Debian-format packaging), it ought be simple for that
person to request sponsorship for an improvement.  The very simple
model you've proposed above (even with the changes to address problems
I've noted) meets that requirement.


More information about the ubuntu-devel mailing list