[beryl-dev] Talks about a merge between Beryl and Compiz

[Kristian Lyngstøl]

The majority of the developers have allready gotten wind of this
discussion, and it's time we discussed this in the open. Everyone
deserves to know and be able to comment. Closed-door meetings do not
belong in the free software world. All our developers deserve to know
what's happening, not just a select few. New, aspiring developers that
may just be lurking deserve to know, and comment. I forsee a flamewar
as a result of this (like anything that involves both compiz and beryl
beeing put in the same sentence), but that can't be helped. It's
better to take the flamewar now, before the process has gone too far.
It is bound to happen anyway. Hopefully the discussion can be
constructive. Let's get ahead of the flames for once.



Robert (racarr) contacted David Reveman a few days ago in respons to
his post about "The beryl situation", and started a dialog about a
possible merge. He did this on his own initative, it was not in any
way an attempt to force a merge, but rather to investigate the
possibility. They have exchanged a few e-mails, and I belive Robert
really wants this to happen, and so does a few other people.

Beryl and Compiz are two programs that wish to accomplish more or less
the same thing. In the perfect world, two projects working for the
same goal, should be merged into one, however, the world is not
perfect.

Compiz was forked for a reason, and so far, I haven't seen any
evidence that those reasons are gone. So why would we merge now, when
we have a very real chance to make a big difference?

The difference between Compiz and Beryl is still quite small, neither
project has made any great architectural advances during the last few
months. And luckily for us, we have taken the effort of keeping
up-to-date with important compiz changes, the same is not true the
other way around. So how exactly would a merge happen, practically
speaking? It's not a matter of just porting a few patches or doing a
diff. Would we use Beryl as the base, or Compiz? Either way, it would
require quite some time.

I haven't seen much work on Compiz over the last few months that would
really suggest that it is necesarry for us to merge. While David
Reveman continues to work on Compiz, the real groundbreaking work he
does (And he does some work that will benefit Compiz and Beryl
greatly), is not with Compiz any more, but with X.

When the fork happened, the main reason was that the patch set between
compiz-quinn and compiz was too big to just be a patch-set. I was not
a part of the developer community for either projects at the time, but
I do remember going from vanilla compiz to compiz-quinn, and how it
just felt better. There weren't many differences, but the few
differences that existed really mattered. The work on the settings
system beeing a major factor. So far, I haven't seen any hint that
David Reveman takes those issues sufficiently serious. Let me quote:

 "The number of people who are interested in text-based configuration
support is very limited. Giving a few thousand SLED users a good user
experience has higher priority for me than implementing something that
a few users that hang out on the forum wants. I hope you can respect
that."

No, I'm sorry, but I can't respect that because I simply don't belive
it's true. The fact that clue-less people are willing to accept
anything that works doesn't make it a good solution, which Reveman so
often reminds us when he describes our work more or less as a set of
dirty hacks. GConf may be nice if you actually want something like
that, but a lot of people don't. Beryl has dealt with that, and has
never gotten any bad reviews due to the settings system, even when we
didn't have gconf support. Not from objective sources anyway.

But that's just a techincal detail, really. An example. People will
allways disagree, and I can accept that Reveman disagrees with me on
this. But that's not where it ends.

The reason the fork happened was the fact that the patches from Quinn
never really made it into Compiz. The fact that the compiz-quinn
patchset became well known even outside the compiz community should be
evidence of that. You don't need to go into techincal details as to
WHY that happened, simply that it did happen when it shouldn't have.

If you read between the lines of David Reveman's mails with Robert
(Which I will not quote here as they were not sent to me, merely
pasted), he wants each author to retain control over whatever he or
she wrote. Because that's how it's done in the Free Software Community
apparantly. Well, that's not acceptable to me at this point. That is
exactly why the fork happened in the first place. In the Beryl
Project, we work as a team. One person may hold a copyright on a piece
of code, but he does not have a veto on what patches goes in. Sure, we
discuss with the author (if he's on the team) before making
significant changes, but in the end, the original author doesn't hold
complete control. The council does. While the idea of each author
controling his code sounds "fair enough", it will effectivly give
Reveman full control over the code, since he is the original author.
He knows this, I wonder if Robert realise this, I hope he will now.
While it is important to respect the work of the original author, I
don't belive it is reasonable at this point that one person holds
complete control. We allready tried that, it didn't work. If Reveman
had the leadership abilities needed to pull that off, then there
wouldn't have been a fork in the first place, but he has shown that
he's unable to cooperate sufficiently.

Also, there is the matter of a license.

I am a firm beliver in the GPL.

Reveman states:

"I used the MIT license in compiz for two reasons. When moving to a
composited desktop, we're moving more and more of the X server (which
is MIT licensed) functionality into the compositing manager and I want
to make sure that this new composited desktop that we're creating can
be deployed anywhere a regular non-composited X desktop is used today.
"

I don't see how using the GPL would limit anything. Several window
managers are GPL. And to be honest, I don't want people to abuse the
freedom that free software grants you. The discussion between copyleft
and non-copyleft licenses doesn't belong here, there are plenty of
flame wars about that going on. But for me personally, using a
non-copyleft license is something I would rather avoid. The fact that
Reveman refuses to see that there are good reasons for the GPL, even
if he might not agree with them, is not a good sign. The MIT license
is one of the things he seem to insist upon in the e-mails he
exchanged with Robert.

Also, I question Reveman's competence in moral dilemmas, which a
license essentialy is. Recently, Ben Reeves, also known as
"Zootreeves",  the site administrator of the Compiz forums (at the
time), was caught red-handed breaking into a Beryl server and
performing sabotage. I feel it is crucially important to condem this
kind of behavior. It is unacceptable for community leaders to behave
this way. The Beryl Project was kind enough not to press charges
hoping for a public apology, and if you ask me, no such thing has
happened. And Reveman has not even commented on the situation which
was experienced as grave, considering the fragile relationship between
the Compiz and Beryl communities. Is a man who doesn't seem to care at
all when another high profile member of his project breaks the law in
an attempt to damage the oposition, really fit to comment on moral
dilemmas? On what's right and wrong? Or even lead a project? It seems
to me, Reveman only comments on right and wrong when it is in his
benefit.

And that's also just one case. Since the fork, there have been
numerous flame wars between the compiz and beryl communities, I have
never seen Reveman do anything at all to put the flames out, though I
have seem him throw some gasoline on the fire. This behavior doesn't
really encourage a merge if you ask me.

I understand perfectly well that Reveman doesn't use web forums, I
don't really like to use them much myself either to be honest. But
when there are a handfull of compiz community members who seem to
worship Reveman, and uses his statements and actions as ammunition for
their fire, I feel it would be the right thing to do for him to
intervene. He wouldn't have to agree with anyone at all, just stop the
flames. He is pretty much the only person that can do that at this
point, yet he either doesn't care at all about the vocal part of his
community, or he likes to see flame wars started by his followers.
Yes, followers, because that's what they are. They seem to treat him
as a god, and he lets them do this.

This complete lack of willingness to bring the Compiz and Beryl
communities closer, makes me think that his interest in a merge is not
as honest as it would have to be to suceed. Surely he wants Beryl to
go away, but he hasn't shown any sign that he actually wants to admit
that he is not flawless, that he can not lead a merged project, that
he does not consider the members of the Beryl team as a bunch of silly
kids. Even if we merge and someone else than Reveman is named project
leader, it would be on his terms, just as it was before the fork. The
project leader would be powerless to stop him because, after all, the
original author should decide what goes in and what doesn't go in.

I belive this attitude would be a real obstacle for the team. It would
make it impossible to do anything without Reveman agreeing. It would
put Reveman's opinions over the wishes of the majority. Just like it
was prior to the fork.

So fellow Beryl developers, read between the lines. Do not assume
Reveman wants the best for both projects. Do not assume this merge
will be a success. Do not accept sacrifices in our freedom as
developers to make this merge come through. Ask yourself how many
times Reveman has spoken about our work in a positive setting, or even
slightly diplomatic. Ask yourself how much effort he has made to bring
the projects closer together. Ask yourself if anything has changed
since the fork. Then make up your own mind on wether we really want
this merge to happen at this point in time. Don't just listen to me
and nod, think for yourself. And don't just decide I'm a lunatic and
disagree on principle. While that would be an understandable
conclusion, even lunatics have valid points from time to time.

-- 
Kristian