[packagers] What's up with RPMforge?

[Dag Wieers]

On Tue, 27 Feb 2007, Morten Kjeldgaard wrote:

> Excuse me for raising this question here, but I am getting more and more of an
> impression that RPMforge is stagnating, and I find that extremely worrisome.
> The lack of traffic on this, the packagers list, the never-updated website,
> and the absence of the primary driving forces (Dag, Dries, Matthias) on
> various forums all appear to be signs that things are not developing as we all
> wish for.

Morten I completely agree with your assessement and I'm not dissing 
anything here. I have brought this up (again) at FOSDEM and I hope we can 
resolve this sooner rather than later.

FOSDEM might be a corner-point in RPMforge's future. There are 2 possible 
ways out. A CentOS community effort is pretty much what we are hoping to 
do, but at FOSDEM we discussed with Red Hat and Fedora people to see if 
a merging effort with Fedora is possible.

CentOS here needs to make a decision. Only then we can think about a 
future for RPMforge. If the same thing will happen to RPMforge on RHEL as 
what happened with RPMforge on Fedora I can honestly say packaging is not 
what I will be doing in the future.

 
> If this is indeed the case, and if the primary forces of RPMforge are getting
> bogged down with work, why not have a discussion here on this forum of what
> can be done about it? I am sure there are several readers who would like to
> help out with various chores, and lots of people, myself included, have many
> packages that could be maintained within the RPMforge framework. I believe
> that was part of the initial vision as well. We all have lots of things to do,
> I find myself tied up with other chores for months at a time, but once in a
> while I have some spare time that I would like to spend giving back to the
> Linux community.

My main concern with this is that RPMforge will not make it if we add more 
people per se. I don't want to end up like with Fedora where you have 
a majority of packagers only maintaining 1 or 2 packages, because that 
complexity is hard to manage with the little resources we have.

 
> I recently installed Ubuntu on a spare workstation. It was a big pleasure, the
> distro works very nicely and is beautifully designed. You download a CD, and
> when the machine is up, you can install whatever you want via apt-get. And
> particularly, everything is available from a single repository. You don't need
> to fish around in a dozen of more-or-less incompatible repos for the stuff you
> need. And perhaps even breaking software in the process, as many of the repos
> seem to favour packaging the same software slightly diffently and with
> inconsistent dependency schemes. So even if I don't have the slightest
> sympathy for ESR's pathetic whining last week (everybody else knows that you
> have to pick your repos with care) it does raise a point.

When I read ESR's complaint I said to myself. Why did he choose Fedora ?
If he wants a platform for the desktop, a bleeding edge distribution is 
not what one wants. And I do not understand/agree with his non-free 
arguments either.

 
> I am _not_ considering a switch to Ubuntu. Those of you who have tried to
> author .deb packages know that it is an extremely tedious and complicated
> process, compared to authoring an RPM package. Ironically, the easy of
> creating an RPM package has resulted in the proliferation of RPM 2nd and 3rd
> party distributions to the extent that they are no longer compatible. Recall
> the fftw/fftw3 debate.

Well, compatibility needs to come from all partners involved. And while we 
were open to fix our side, FE never considered our existence important. 
You cannot work together with an entity that doesn't acknowledge your 
existence.

 
> RPMforge has long been my hope for the way out of this mess. However, RPMforge
> has IMHO been hampered by the need to support several different distributions.
> In particular, Fedora is a moving target, and a lot of effort goes into the
> cerimonial semi-annual recompilation of all the packages. Perhaps this would
> be a good time to direct RPMforge towards a single, stable, and well
> maintained mother distribution, as for example CentOS (RHEL). There are other
> advantages to dropping Fedora support, but I don't think we should get deep
> into that discussion at this point.

Fedora support is important to prepare ourselves for the next RHEL 
release. But it should not be our main focus. CentOS 5 will change a lot 
and take away much of the reasons why people use Fedora.


> To ensure a uniform packaging and dependency policy, I would suggest keeping
> one eye on the Debian distributions. They have already worked out a a naming
> dependency scheme that _works_.

The problem is that the Fedora naming policy is dubious in a lot of cases 
and at FOSDEM I tried to raise that point with the wrong example so most 
of the point got lost on semantics. We can't invent yet another naming 
policy.

Regarding the future, I'd like to do a call for volunteers for RPMforge so 
we have a small but experienced group of people that can give input in 
changing RPMforge's ways. CentOS has to be involved as well.

The aim is to build a culture where people can contribute and commit to 
RPMforge, for that there are a few ideological hurdles and a few technical 
aspects we have to sort out.

Are you interested to be part of this ?

Kind regards,
--   dag wieers,  dag at wieers.com,  http://dag.wieers.com/   --
[all I want is a warm bed and a kind word and unlimited power]