Dear Fesco: Orphan package process needs work

[Michael Schwendt]

On Sat, 07 Oct 2006 11:37:44 -0400, Christopher Aillon wrote:

> > But you and him are used to deadlines like this.
> 
> Are we going to expect this of every single contributor?  Please ignore 
> the fact that he worked at Red Hat here, I'm talking in a more broad sense.

Yes, when there's a deadline of some sort and announced properly, it will
be very difficult to lower the hurdle for some packagers. Either there is
the goal to be ready with the repository at a given point of time or there
is no such goal. And if so short before the deadline unmaintained packages
are discovered, the only rescue would be to decide quickly and who may
step in and help with the needed action. That, however, must be in
accordance with the "no orphaned packages" rule as to avoid that somebody
contributes only the missing build jobs without interest in taking over
package maintenance, too (and for instance, also examine the package
status prior to just rebuilding it).

Fedora Extras are still learning. We've tried out different roads before
and perhaps will try out even other approaches next time. What has been
determined as bad, however, is to find out *after* the official release of
the distribution (i.e. Core+Extras) that there are orphaned rpms in what
we offer in the repository, possibly with open bug reports and security
vulnerabilities and that there are no volunteers to fix them, and
regardless of how important or popular a package may be according to some
voices. What's needed is packagers or full package maintainers. It may be
the individual packager, who has failed, but this falls back on Fedora
Extras as a whole, too.

> Penalizing users of the software because the maintainer makes 
> empty promises is just plain wrong.  Action was indeed needed on 
> someone's part for the package.  We should have found someone to perform 
> the action if the maintainer was making empty promises, rather than just 
> nix the package.

You're exaggerating with regard to the impact of removing rpms from a
"Development" type of repository. There's still time to bring them back.
Still a chance to name one or more co-maintainers.

> Great.  That process is nice.  It works.  What I'm talking about and 
> what doesn't work is that when a package maintainer is pinged and given 
> notice that "Hi.  You need to do this.  Will you?  If not, I will find 
> someone else" and then proceeds to agree to do work, stifling the 
> correct process of finding a new owner.  He stopped that process from 
> happening by promising to do work.  And then failed to do work.
> 
> I'm arguing that *in all cases* there should be an attempt to find a new 
> owner for the package the instant it is deemed the maintainer isn't 
> doing their job for the package.  There was none in this case because 
> the owner stifled that.

This has been commented on before.