So let’s catch up quickly. Daniel started Gentoo, he led it, Gentoo grew and attracted all sorts, including (and especially?) those who would rather see Daniel not in power. I have a theory: some of those people who lobbied for distributed power structures were actually after being in control for themselves. It was certainly true of Zach (he admitted as much). To be honest, it was sort of true with me, as well. There were times when I thought it wouldn’t be a bad idea for Daniel to step aside for a while and let me drive things. There, I admit it 🙂
Anyway, the momentum for distributed power started building (I believe largely due to back water conversations rather than any largely shared sense of unhappiness). After that, there was little choice left for Daniel but to relent. So Gentoo Technologies, Inc. gave way to Gentoo Foundation, and the Gentoo Managers gave way to the Gentoo Council. And of course, Daniel gave way to the Board of Trustees (13 in number). Various proposals happened in between, of course, to not let any one entity have too much power. So managers had to be neutered in order to be on the council (a note about this later).
And this brings us to the status quo for the last couple of years. Nobody allowed to “have” a vision for Gentoo. Actually, I think many people have visions for their projects, very few have a vision for the distribution itself. And of those, nobody has the power/blessing to see that vision through. There’s a big power vacuum, a complete state of anarchy. Democracy has brought to us that everyone can have a voice. Thus people believe that they should shout, no matter whether they make sense or not.
The new thinking is that the Council itself should have a head (told you we would come back to that point). Part of the thinking is that the anarchy is good. It hasn’t destroyed us, that’s for sure. Maybe the dictatorship did almost destroy us. I don’t know, to be honest. What I do know is that it’s a lot more agonising these days, with endless discussion on endless discussion because of endless discussions about endless tangential points to the original endless discussions, ad nauseum. At least when the buck stops, you can feel a little bitter but at least you know it’s done, and you can move on.
The point of bitterness for me is that two of those three who first advocated the idea of socialism/democracy to Daniel in the first place only succeeded in substituting one problem for another, and then not bothering with trying to fix either problem. I believe the best direction for Gentoo is to split itself into a core with overlays, as I’ve mentioned before in other posts.
The beauty about the old days that most people seem to miss for some reason was the size of the Gentoo development community. It was as efficient and friendly as it was precisely because of the kind of intimacy that is inherent in a small group. And in all honesty, that sort of happens already, because people tend to only care about the projects in which they have direct involvement — be it the amd64 architecture, science packages, clustering, gnome, whatever. Most of those devs lose nothing by having their own out-of-portage overlays. Hell most people with @gentoo.org could probably even keep that address (or a derivative like @contrib.gentoo.org maybe, who knows).
The core group of devs would be people who make packages and programmes that are crucial to a running Gentoo: baselayout, kernel and sys-*. Even portage itself should be in an overlay — with a party on the core team to determine the requirements for acceptance as an officially sanctioned package manager.
Things would work, because overlays could be given a lot more independence and a lot more visibility. Right now they are all overshadowed by what’s in the tree already. Instead, overlays would create competition, and thus increase the quality of submissions. The barriers to entry in our current system are high. Overlays reduces those. Anyone can enter — it’ll take a lot more quality and skill to be accepted as official (requirements to be determined).
Those are my thoughts. I’ve tried not to slander anyone, but feel free to take issue with me.
(my own opinions..)
When Daniel left, he did not replace his authority with any body of similar authority. Ok, it is arguable that he did with devrel, but without written proof nobody believed it and they could not exercise the authority. The managers and/or council has not taken on any greater authority than before either. Without that authority, Gentoo’s development has turned into an uncontrolled chaos. This results in misguided goals and the tolerance of bastard devs and users.
Does the chaos work? Somewhat, Gentoo has continued on for a couple of years this way. However, with that you see several offshoot projects within Gentoo that don’t have the support of the development community at large, and you have some of your best developers leaving to spend their time elsewhere (and in that sense, the anarchy you mentioned doesn’t work).
I mentioned this last year[1], but reading Five Dysfunctions of a Team is an eye opener to some of the improvements that can be made within Gentoo to make it 10x better than it is right now. What frustrated me the most is that it became obvious that the way Gentoo was heading, there really was no way to get everyone together and make the necessary changes to fix things. The authority to do so was gone.
So what am I getting at? A dictatorship isn’t the answer, but you need someone to have a final say along with a system of checks and balances. Trying to manage a group of overlays under the name of Gentoo is going to be even harder still if you want to maintain any kind of reputation in the brand. I don’t have an answer that the community at large would buy into 100%, and that is why it’s near impossible to fix the major problems in the organization.
With a management team that every group reports to, and focuses on a key goal together some cool things could be done. What is the common goal between the portage team? releng? infra? arch teams? You could say Gentoo is the goal. But are they all working together to meet the common goal, or do they work separately for their own ideas and agendas, with that main goal twisted in some way for each group? I saw (and participated in) too much of the latter in Gentoo. Groups worked together when it was to their benefit rather than to meet a common goal, and the lack of social guidelines meant that if you needed something from someone else you better wish that you never pissed them off. Yes, this was all a lot easier when Gentoo was small and intimate. But, I submit that it is also just as possible on a large scale given the appropriate structure and workflow organization.
Anyway, I read pleas for changes in the organization of Gentoo on an almost daily basis between blogs and mailing lists. I wish anyone luck who is busting their butt trying to find a solution. It’s tough.
Cheers!
[1] – http://staff.osuosl.org/~cshields/?p=36
I (as a user) want things to just “work” and not search around for it in various overlays, then check if they don’t conflict with other packages from other overlays and then check afterwards if they don’t do something “strange” e.g. download apache sources from hackapache.org instead of apache.org.
In the official gentoo tree there are developers responsible for packages, in overlays you’re often just left on your own if things work for you or not. Think “It works for me, so it’s your problem.”
Next thing, in gentoo when a maintainer leaves a new one will step in at least for important stuff. In overlays again you’re often left alone if someone steps away or has too little free time.
It would be ok for a chaotic gentoo but there are people relying and trusting in gentoo devs and the official portage tree. People who run their (secure) servers with gentoo. I bet they would step away from gentoo when they have to trust in various overlays.
Gentoo is a strong distribution and not just a base system for hacky things. Your idea would rip that apart..
I hope before this decision is made there will happen some kind of user survey/voting.
Maybe gentoo should enploy something like Quality Circles or some of the TQM measures. Quality circles could mean putting together _one_ from each team/party/subfraction of the gentoo project (portage, releng, arch, infra). These groups should meet once a month and present a unified (for this groups of up to 5 devs) vision once every three month. The vision should be posted in the forum or in the GWN. This may help provide a way to get a idea where gentoo is going and unify visions and the big picture.
There groups shouldnt be elected and there can be as many as long as there is at least one. The condititions for the groups are also simple: Only one member/representative from/for each subproject of gentoo.
http://en.wikipedia.org/wiki/Quality_Circles