This was expected. One of the most radical Debianer (Mj Ray) tried to contradict the arguments I gave in my previous post about the original meaning of point #8 of the DFSG. And clearly without historical background, the interpretation of that point can lead to confusion. (Hint for Matthew: That’s why arguing with him is worthless right now.)

Can we then try to correct the formulation of the DFSG to avoid problems like the one we have now ?

I don’t have a patch ready for proposal but I can explain quickly what I have in mind. And if I get some support I may even try to draft a general resolution…

Contents (possibly non-free) which would serve the purpose of a free work and which are freely redistributable could be accepted into main together with the free work in question.

Of course the limitations of those “contents” would be documented in the copyright file of the package.

This idea solves all our current problems :

• with trademarked names and logo : we can use them;
• with firmwares : clearly firmware are non-free contents that serve the purpose of free drivers;
• with GFDL document with invariant section : the invariant section is non-free content serving the purpose of the free software documented.

Would this exception be a disservice to our users ? I don’t think so.

For the pragmatic point of view it’s clearly a win, we can keep the names of well-branded software, we can have drivers working out of the box and we can keep the most valuable documentation we actually have.

From the philisophical point of view it may look like a step backwards, but it’s not. Non-free content is only accepted if it serves the purpose of a free work. And I used the term “content” on purpose, a “program” would not qualify as “content”. Otherwise a non-free software enhancing a free software could have been accepted. But clearly that’s a no-no. We’re about “free software” and that’s what matter for us, so we do our best to spread them and this may require to distribute them together with useful non-free contents.

We probably also need to write a special documentation describing in more length the spirit in which each point of the DFSG has been written and clarify with recent history what is commonly accepted behing each buzzword (i.e. “work” can be program, textual content, multimedia content. and other points like this one). But that’s for later I bet.

Eric Dorland (the maintainer of mozilla-firefox) is wondering if he can use the name Firefox for his package and still respect the DFSG.

His reasoning is that we can call it Mozilla Firefox only because we’re Debian and DFSG point #8 forbids the license to be Debian specific. So he wonders if the trademark license complies with the DFSG.

His reasoning has many flaws :

1. The DFSG has been written with software license in mind and not trademark license (argument defended by Wouter Verhelst).
2. His interpretation of the point #8 of the DFSG is too strict compared to the original purpose of that point (see my mail and this one from Matthew Garrett).
3. His logical conclusion is not coherent with our goal to serve our users (cf. opinion from Anthony Towns). It would be a disservice to our users to include Firefox with a different name.

Firefox included in Debian with its original name is still free software – DFSG compliant – (Eric recognizes that himself) and that has always been enough for us to accept to include a software. Nothing more is required.

Isn’t that enough to make it clear that we can and should include Firefox with its original name ? It’s not against our principles and we’re not breaking any of our rules.

The Mozilla Foundation has good reasons to try to protect his name just like we did when we asked the “TrustedDebian” project to rename itself in order to avoid confusion with us. Their trademark license is perfectly acceptable and we should accept it because we applied a similar one to our own trademark!

I think that we forgot this when we made the last set of “editorial changes” to the DFSG. We remembered a bit with the general resolution postponing the changes until after the release of sarge… now sarge is getting close and we’re going to face the problems that we created.

I hear more and more Debian developers who are disappointed by this move, some are leaving, others are considering to leave. I’m too attached to Debian to be able to leave, I’d rather make Debian change again because I also think we have made the wrong choice.

We have many derived distributions which enhances Debian in a way or in another, we shouldn’t make their life more complicated by making changes that they will have to undo just to have a working system.

Even forgetting about the derived distros (which are an important part of Debian IMO), our users deserve a working system right after the installation. We have two solutions: either change our policy so that apt-get points to non-free by default and let packages from main depend on packages from non-free so that the required bits of non-free can be automatically installed, or change again our rules so that we can keep those essential bits were it makes sense (in main directly).

The first solution makes no sense because of the nature of non-free: we have so many different licenses of packages in non-free that it’s difficult to know which packages can be distributed. So the only real solution is to stay in the old situation: apply DFSG to the software only and keep useful things in main until a free replacement is available.

Feedback is welcome… because the only way to make this change happen is to propose a new general resolution.

Two years ago I recommended to vote for Martin Michlmayr because I was much more involved in the internal politics of Debian and because I was a candidate 3 years ago … nowadays I’m less involved and I’m voting based on what I’ve read: the platforms, some mails on debian-vote, the interviews on Linux Weekly News and the logs of the IRC Debate. I must say, it’s a difficult exercise for a busy person. It took me many hours… but I’ve done it because I wanted to vote without randomly choosing.

It looks like not everyone did the same effort, otherwise I can’t understand how some people forgot to put Jonathan Walters below “None of the above” (thanks to the people who send their replies to debian-vote instead of the good mail alias). Jonathan is a good Debian developer but that’s all. As a leader, he wouldn’t give us a good image. Of course, it’s just my opinion. You’re free not to share it.

Coming back to the subject, who did I vote for ? Anthony Towns is placed first on my ballot. I think he’s the most pragmatic candidate this year. He knows how things work within Debian, he knows what doesn’t work. All the candidates agree that lists have become unpleasant places and that we should remember that we’re all in Debian because __working for Debian is fun__. Anthony proposes a drastic solution (rules for banning noisy people and so on), it’s possibly too much but it’s worth trying. A first step would be to write a [Code of Conduct](http://www.ubuntulinux.org/community/conduct/document_view) like Ubuntu has: it’s quite effective there, anytime someone starts a new troll someone responds by “please stop that and read the code of conduct”, and the thread dies there. This will be difficult for Debian because we have so many people used to react quickly… but temporary bans for people who can’t refrain would be incitative.

Branden Robinson is still not very well placed on my ballot even if he’s no more the same anti-french guy he used to be in the past :-). His way of behaving reminds me too much of real politicians: he discards hot subjects (no reaction to the rebuttal of Anthony Towns), he says he doesn’t know if he will represent himself unless 100 people support him, etc.

All the other candidates are much like clones… good Debian developers with good ideas, it’s difficult to choose between them. I put Matthew in second position on my ballot without many reasons. It’s just that I have a good feeling, maybe because of his involvement in documenting the work of teams like the ftpmasters.

BTW, I do not like the idea of the SCUD team. The leader must delegate what he can’t do, there’s no need for him to have co-leaders. Or maybe the SCUD team is good idea but it doesn’t need to be formalized… and advertised as such. The leader naturally has people who advise him.

Last remark, I’d like to know who Martin Michlmayr votes for this year.