[fab] 2006-06-20 agenda

Tue Jul 4 11:47:06 UTC 2006

On Mon, 19 Jun 2006 13:21:04 -0400 (EDT), Max Spevack wrote:

> * Did FESCO talk about the sponsorship questions that were raised in our 
> last meeting?  Did they make any decisions?  What were those decisions?

So I arrived at this thread and tried to collect some pieces:

| http://fedoraproject.org/wiki/Board/Meetings/2006-06-06

| Extras Sponsorship
|
| First off, we all agreed that we're talking about this at the Board level,
| but the actual action on the issue of Extras Sponsorship needs to happen
| at the Fedora Extras Steering Committee (FESCO) level -- they are the ones
| with the decision making accountability in this case, to either act or not
| act.
|
| We heard at the Red Hat Summit that the sponsorship process isn't as
| smooth or efficient as possible.

Again and again, complaints like this are totally inappropriate. Who is
"we", and who talked to you? Define "smooth" and "efficient". How "smooth
or efficient" could the process be?

| When we started talking about it, these
| points were made, for consideration by anyone who enjoys the topic:
|
| * Some people don't want to be sponsors, and as in all parts of Fedora,
| the number of people working on something is always a limiting factor.

Anyone, who may have observed FESCO for the past months, should know how
new sponsors have been proposed/selected.

Initially, we assumed that very active reviewers would enjoy being able to
also sponsor new contributors, so they could become even more productive.
Suggested reviewers were collected and discussed briefly. On agreement,
somebody then contacted the persons privately to ask whether the role of
being a sponsor was accepted or rejected.

A problem with this is not that somebody might reject the offer to become
a sponsor. It is that hardly anybody does enough monitoring of "potential
sponsors" in public Fedora places of activity (like bugzilla, CVS commits,
mailing-lists) to suggest new sponsors _every week_. The amount of
bugzilla traffic caused by package reviews is big. Automated metrics,
collecting numbers on "approved packages", "packages in FE", "open
reviews" and so on, only help a little bit, since there is a big
difference between approving packages of existing contributors, who create
clean packages, and taking a look at new contributors and their packages,
which sometimes need a lot of work to even build and run. For potential
sponsors (in general, very active reviewers) the primary question is
whether and how they approach new contributors, who need a lot of help and
guidance, and their packages, which may contain a lot of problems and
pitfalls.

Hence we started asking for self-nominations. To learn about active
reviewers who are interested in becoming sponsors. Interested people then
could give good examples of their reviewing activity.

Now to the sponsor's obligations. When you approve the account request of
a new contributor, you're not done. It may turn out that the sponsored
person needs much more hand-holding (with Using CVS, maintaining the
packages, build problems, upstream changes) and monitoring (incoming PRs
and responses, package changes). It would not be the first time. And it
has been criticised a couple of times that there is no post-review QA for
FE. Actually, packagers are able to reintroduce packaging bugs and
pitfalls, throwing away the good help they've got during the single review
of their package. Some of these bugs are show-stoppers for everyone at
Fedora Extras, such as the unfiltered SONAME "Provides", which have hit us
at least twice, causing the buildsys to fail for many contributors.

Sponsors need some time to get to know other _new_ contributors before
they decide whether to sponsor them. If new contributors are not patient
enough, two very good recommendations for them are

 - to offer working packages, at least make the src.rpm build -- it is a
   real pain to see how some reviewers spend much more time on incoming
   packages than their soon to be maintainers,

 - don't argue with the reviewer about the PackagingGuidelines -- wrong
   person, wrong place, spend your energy on contacting the packaging-list
   or the Packaging Group if you feel it's important,

 - ask questions (if you are unsure about something) and show your
   interest.

The sponsorship process guarantees the availability of minimal resources
for help/guidance (either if asked directly or if they discover something
while monitoring sponsored contributors). If the number of contributors,
who need help regularly, increased too fast, you can be sure that the
number of problems in FE would increase, too. Plus the number of people,
who drop off silently (in frustration or because they cannot handle the
requirements), leaving orphaned packages until they are discovered.

| * Are there major problems, or were there just a few folks grumbling?

Occasionally, there are "a few folks grumbling" _in the wrong places_
about the time it takes for some packages to enter FE. Maybe these are
the same folks who want the package reviews be removed?

Everyone, who has complained in unknown/secret places before, talk to
FESCO directly in case of serious complaints. Don't try to introduce some
kind of pressure by talking to Red Hat employees or other people you
assume have "something to say". This is damaging the community side of
the project.

There is also wrong terminology/jargon being used. Some folks keep saying
"packages are sponsored", but it is: _persons_ are sponsored, _packages_
are reviewed and approved.