Log from yesterdays (20070523) EPEL SIG meeting

[Patrice Dumas]

On Thu, May 24, 2007 at 07:52:38PM -0500, Michael Stahnke wrote:
> 
> It should be noted that these points of contention are not viewed the
> same among all members of the steering committee, and certainly not
> all members of EPEL.  

Indeed, but the whole impression I get is that in the steering commitee
there are people frankly in favor of 'all free software packages in
EPEL' while others don't seem to care. As shown in the fedora case what 
the members of EPEL that are in the minority think is not of much
relevance. Nobody will stop them to do what they want at their level
(including doing things contradictory with what comes from the steering
commitee), but their advice will only have a consequence of lengthy
threads.

> The writers of that RepositoryCollaboration
> statement were not all that involved in this discussion.   

At some point being there to argue a political point when in a minority 
is not worth the effort. Sometimes even for technical point it is
worthless and time consuming. Even if the people weren't there, their
arguments and positions were in everybody mind, so it doesn't make a
difference in my opinion. It could have delayed the meeting and add some
impolite statements in it, but I don't think it would have been
different.

> them, everyone should and would benefit.  Sadly, we are caught in an
> endless debate that morphs into flaming and hard-feelings.

I disagree. Those endless debates are there because there are different
models of development and every choice makes somebody unhappy. A choice
of one repo with all the packages against a variety of repo is going to
make somebody unhappy whatever is choosen.

> I don't have an answer right now to the problem (and the problem is
> well beyond repotags), but I do hope one emerges in the near future.

There is no problem as such, but a choice to be made. The Fedora example
shows that it is possible to almost replace the other repos, and as
somebody pointed out the Fedora model has some advantages over the
multi-repo situation. This is clearly an uncollaborative solution but it
may be the best one -- or not.

> Currently the userbase  is the most affected by this lack of
> cooperation, and they should be the most important.

No, they shouldn't. The choice ebetween one big repo or a multi-repo
world has to be done within EPEL. Whatever the choice is it may fail,
for example the one big repo could be choosen but history may be
different from what happened to fedora. In any case users are not
involved a lot in this issue except that they are the long-term 
target.

In any case the choice should be done, because it should impact the 
development path of EPEL. For example if the one big repo solution is
used there should certainly be a specific repo with the latest versions,
not only the default EPEL repo, the repotag issue is moot. If the 
multi-repo alternative is chosen, then there is a need to work with 
other repos, for example to have them split out non-free or illegal 
in the US part such that their repos may be in the default config, new 
processes and infrastructures are needed to play nice in a multi-repos 
world with some agreement on some standardizations.

--
Pat