Role of the Sponsorship Queue
persia at ubuntu.com
Tue Mar 2 18:32:30 GMT 2010
Daniel Holbach wrote:
> Emmet and I had a number of discussions about the nature of the
> Sponsorship Queue and we'd like to have your input on it.
For those who like archives, this last came up in December, and I
still feel the last three paragraphs of my last mail (1) are worth
> The major disagreement is if the sponsorship queue is
> - a general list of changes that need be reviewed (after having made
> sure they are actual code changes, etc.) - basically a TOREVIEW list
> for people who can upload
> - an exclusive list of work done by new contributors who want to join
> the Ubuntu Development team eventually - so a place where mentoring
I think that these are both importnat things to have. We need a
list of arbitrary stuff that needs review and treatment. Sometimes
that's testing, sometimes that's getting it upstream, sometimes it's
uploading, sometimes it's requesting modifications, sometimes it's
porting to newer or older versions. This is absolutely critical work.
That said, I see no reason one needs to have upload permissions for
this. The large majority of these patches are received from users and
really need to get forwarded upstream for review/application. Many of
them would just lead to more merge pain if uploaded in Ubuntu for
little clear value. Doing it this way not only increase the set of
people who are able to participate in helping to reduce the ~2000
outstanding patches we have, but also help improve the perception of
Ubuntu as sharing patches to our various upstreams (rather than us
"hoarding" them in LP due to lack of attention. Further, I don't
think that most patch submitters should have to learn *any* process
whatsoever to get their patches reviewed. Just attaching a patch
should be sufficient to get it reviewed and processed. I also think
we place an unfair burden on our users if we expect them to clean up
or rebase patches just because we didn't get around to looing at them.
That's clearly work Ubuntu can do.
I think that the process of working with those who want to be
Ubuntu Developers is completely different. These folk should be
finding and fixing bugs (perhaps by coding, perhaps from patch review,
perhaps by cherrypicking from upstream or other distributions, etc.),
and submitting good clean candidate packages that are suitable for
direct upload.. These candidates deserve close scrutiny to ensure
that they are correct. Small mistakes ought be identified so that the
nascent developer can learn. When these are in good shape, they
really just need to be uploaded, and the person preparing the "patch"
representing the updated candidate is acting fully as an Ubuntu
Developer (and expected to know processes, etc.), although they may
not yet have upload rights for the package in question.
> The distinction is important because "including contributed fixes" would
> require developers to check various lists for input and probably treat
> them differently ("can you update patch to do XYZ in a special way?" vs
> "I modified your patch and uploaded it."), etc.
I think the easiest way to do that is to promote *both* the Ubuntu
Sponsors and Ubuntu Reviewers teams. The Ubuntu Reviewers team (2)
fills the role of dealing with all the patches that need close review,
forwarding, etc. The Ubuntu Sponsors team exists solely to help those
developers wanting to uplaod stuff for which they don't yet have
upload rights, but with an entirely separate (and more training
oriented and critical) workflow.
> The second implication would mean a blocker for
> - the essence of the spec is to streamline the process from triaging
> bugs with patches and feeding them into the sponsorship queue.
This blocker is trivial to unblock: just discard the workitem.
There is already a script to hunt for bugs with patches and subscribe
the reviewers team. If the reviewers team is appropraiely promotoed,
and includes a bunch of folk who have upload rights, it doesn't matter
if the stuff gets into the sponsorship queue. In the case where a
member of Ubuntu Reviewers doesn't happen to have upload access for a
package that has an update that should be uploaded directly to Ubuntu,
they may use the Sponsorship Queue to get their new revision uploaded.
Havng restated my opinion, I'd also like to hear others thoughts.
Aside from minor comments here and there, and Martin's reply to the
last mail, this has largely been a discussion in various fora between
Daniel and myself, and we're not going to reach consensus without your
More information about the ubuntu-devel