Gentoo: New release, "new" leadership

Last week, lots of Gentoo news came out, so it's a good time to look at what happened and what it means. Gentoo's 2008.0 release marked its first since more than a year ago, despite its attempts to release twice a year. Fortunately, Gentoo releases don't mean much because it's already a live distribution rather than a snapshot in time with occasional updates. A release provides a new kernel with the accompanying driver support, occasionally a flashy new bootsplash, and the usual bugfixes to the GUI installer, which is not universally loved. But what happened to make this release come so long after the last one? First, 2007.1 was canceled, largely because so many security vulnerabilities came out that it was impossible to keep up with release rebuilds. 2008.0 was scheduled to come out in March, so it slipped 4 months.

Tobias Klausmann described the problems well. Here are a couple of them:

In addition, Tobias cited three other problems:

• Release work is unpopular. The release engineering team is perpetually undermanned, basically because the work is boring and otherwise unrewarding.
• Bike shedding creates secrecy. Everyone's trying to chip in their own ideas of how things should work without having any experience or clue of what their ideas mean.
• Reproducing installation bugs is hard. This is much like the Linux kernel because the release engineers just don't have the hardware. In some ways, it's worse, because the people who file distribution bugs about problems installing are often inexperienced Gentoo users who don't know how to file a good bug. Often, bugs that make it to the upstream project have already been filtered by the distribution, but that of course hasn't happened here.

The main problem delaying 2008.0 was real life interfering with a critical developer. This is being addressed by creating new processes and backup people who can take over when others aren't around. As for the other problems, it's unclear how to fix them. Suggestions would be appreciated.

The other major news in Gentoo is the election of a new council. The council is a group of 7 people who lead Gentoo by making decisions on global issues. Two things make this election interesting:

• It was a forced election that resulted indirectly from a controversy over expelling developers from the project. It happened because of a technicality in the Gentoo Linux Enhancement Proposal (GLEP) that gave the council its authority. The GLEP requires monthly meetings and forces an election if a majority of council members don't show up to a meeting. The controversy came about because this was an additional meeting beyond the usual one, specifically to discuss the appeals of 3 developers who were fired. It was poorly announced (only mentioned in the meeting minutes). It's unclear whether a majority of council members even agreed on the time.
• The election involved people who think the social side of development matters versus people who think only the technical side matters. In Gentoo, the silent majority of developers rarely post to mailing lists, preferring to simply do development. Votes like this are often the only way they choose to express their opinions. In the past year, 50% of the traffic on the main development list came from 20 people, yet nearly 150 people voted in the council election and more than 250 are listed as active.

The 145 voters approach the highest number ever in a council election—here's how it compares with previous years:

This is the highest turnout since the first year the council existed, showing a significant increase in interest by the developer community in who their leadership was compared to the intervening years. To understand exactly who they voted for, these histograms show how highly each candidate was ranked, in order of result. The left side indicates that a candidate was highly ranked, and the right side shows that a candidate was poorly ranked.

Of particular interest is the position of "astinus," a developer who retired during the election but was still voted above three other people. Since these three people all favor ignorance of any social issues from someone with good technical contributions, this really shows how strongly the Gentoo development community supports the the creation of a friendlier environment.

Notably, of the previous council, every single one of the five members who ran for the new council was re-elected. This shows that the community didn't care about the mistakes that resulted in the new election. It also shows that the community supported the existing council's actions and believed in what its members were saying about the need for social change within Gentoo.

With its new release and its accompanying publicity, Gentoo has renewed interest from many users and has shown that it remains a distribution under active development. Having a new council in place for the next year puts Gentoo in position to rebuild its development community and keep development thriving so the publicity and new users gained by the release don't fade away.

Index entries for this article
GuestArticles	Berkholz, Donnie

---

to post comments

> Of particular interest is the position of "astinus," a developer who retired during the
election but was still voted above three other people. Since these three people all favor
ignorance of any social issues from someone with good technical contributions, this really
shows how strongly the Gentoo development community supports the the creation of a friendlier
environment.

That's an interesting observation. Donnie's conclusion also happens to be true, for my vote at
least. I was quite happy with the election results; this should be a good year for Gentoo :)

The image shows up now, but what caught my attention was the "suppressed 
zero".  The difference looks pretty dramatic until you note that the range 
is 115 to 150, so the variance in actual number of votes cast per year 
wasn't anything like it looks.  I initially wondered what had caused what 
appears at first glance to be a falloff (and reappearance this year) of 
perhaps an order of magnitude in interest, until I noted the range didn't 
extend to zero, so the graph was effectively of the delta, not of the 
absolute numbers.

Donnie, you could have at /least/ mentioned the suppressed zero, warning 
people that the magnitude as graphed therefore was not of the absolute 
number of votes each year, but rather, of the relative change between 
years.

Duncan