My percussionist's heart really jumped high here in Switzerland when I saw this (Flash video; no alternate formats, sorry). NP: Porcupine Tree / Lightbulb Sun

Colour me na ve, but I disagree that Debian is facing a crisis caused by dunc-tank.org. First of all, stuff like this has been going on for years, and I've been personally involved in cases where money was given to developers to work on specific aspects of Debian, though it's always been up to the receiver whether to announce it or not. I hope this does not come as a surprise to you. Second, there are two issues: funding developers, and our DPL engaged in the process in a way that's not obviously independent from Debian to everyone. I think the heated debate around the latter is a mere function of the emotional issues related to the former, so I'll only concentrate on the funding aspect. When people like Joey loudly protest about where the fun has gone, I start to wonder (again) about motivation. Everyone of us who contributes to Debian does so because of some kind of motivation, whatever it may be. In my case, that motivation is technical excellence: Debian is in a position to avoid hacks and do things right, and it's a joy to be part of a community that seems to think alike, that likes to be perfectionist, pedantic, clean, academic, advanced, you name it. I am also mighty proud of Debian, of what we've achieved, of what we are. Lastly, Debian powers every single computer I administer (and the number would be far lower if it wasn't for Debian), and our operating system is how I earn my living. To me, the issue of funding developers is tangential to the development of Debian. Why should it matter that s/he gets paid by the project and I don't? I'm still doing the same work as before, and I am still doing it for the same reasons. Note that we're talking about well-defined, limited use of funds, not employment contracts. When people say that "money corrupts", I think they're taking the easy way out. Money doesn't corrupt, people corrupt in the presence of money. I'll go out on a limb and claim that a person affected by an allergy will probably rather cure the allergy than to avoid the things to which s/he's allergic. Go figure. Even though I've never really cared much for stable releases, I can understand many of our users when they ask us to get along and release "etch", and I really doubt we can do it in time just chugging along like we've been doing it in the past. Of course, Debian is all about "releasing when it's ready", and we don't know yet whether paid release managers are going to enable us to release on time and when it's ready, but isn't it worth a try? I was also opposed to the idea of paying our release managers when it first came up. However, I've managed to change perspective and am now supportive of it: we are not hiring them, they will not be Debian employees; we are raising funds for them to be able to dedicate their full time to the final, tiresome stretch just before a release, without existential worries. We are helping them to move our project along. Especially since our release team has been one of the more open teams in Debian, being quick to respond to questions and training others in the Art of Releasing Debian, I feel rather good about this particular, well-defined and limited use of our funds. All other things aside, here are my top three reasons why (I think) we really want to release "etch" on 4 December 2006, in increasing order:

• we can be even more proud of our project than before because we've bashed one of the main, negative reputations of our project.
• we'll have a first instance of proof that a huge and chaotic project like Debian can meet a deadline. I think this alone would be a major achievement.
• Lars will have to tattoo his arm with our release names.

If it weren't for the last one... :) NP: Kammerflimmer Kollektief / Cicadidae

Yesterday, the Debian package tracking system notified me of a new patch Ubuntu applied to mdadm:

diff -pruN 2.4.1-6/debian/mdadm.init 2.4.1-6ubuntu4/debian/mdadm.init
--- 2.4.1-6/debian/mdadm.init 2006-09-21 21:13:04.000000000 +0100
+++ 2.4.1-6ubuntu4/debian/mdadm.init  2006-09-21 21:10:08.000000000 +0100
@@ -1,4 +1,4 @@
-#!/bin/sh -eu
+#!/bin/sh -e
#
# Start the RAID monitor daemon for all active md arrays if desired.
#

The changelog entry goes out of its way to reason the change:

Do not invoke init scripts with -u. Fixes dist upgrade from dapper to edgy mostlikely caused by /bin/sh pointing to dash.

Of course, it's debateable whether set -u is applicable to an init.d script it causes the shell to barf when uninitialised variables are used, just like Perl's use strict;. I think set -u is good practice and I use it everywhere. My /bin/sh also points to /bin/dash, and I cannot say that I've never run into problems. However, the fixes are usually quite trivial. I consider Ubuntu's change a hack fixing symptoms, not causes. Due to the immense time pressure under which the core team is maintaining the main parts of the distribution, I can imagine such hacks to be quite common. I'd hate for that to be my job as a maintainer. NP: Explosions in the Sky / How Strange, Innocence

Dear lazyweb: I love mutt, my faithful mail client for almost a decade now, but I think I've reached its limits. Not that mutt isn't doing fine, it's more my way of organising mail that's no longer working for me. To give a short overview: I receive all mail in a single account and filter messages to folders in a hierarchy. My filter set spans several thousand lines and has been growing for seven years now, the last "refactoring" was in 1999. When browsing mail, I either delete messages (for list mails mostly) or save them to the read folder when I am done with them. The hierarchy is mostly arranged by field of endeavour, and then subdivided into specific topics. For instance, I have =private/dudes/..., =debian/pkg/mdadm, =ul/phd, and =work/clients/.... I currently face two major problems with this approach:

1. Folders just don't cut it anymore. For instance: with my job, research, and hobby all somewhat connected to Debian, most messages I receive don't belong into one folder any more than they should be in another. What I want is effectively what gmail and several of the GUI clients do: any given message can have a number of tags, and virtual folders act as filters to display only messages with a certain tag.
2. Having a single sent and read folder for all worked for a while, but it's long been a major pain. What I really want is for sent and received mail to sit right next to each other in all cases, ideally based on tags as well.

I found mutt's X-Label patch (the Debian mutt package includes it), which seems to go in the right direction, but it's not exactly what I want (yet). For my purpose, it adds a selector ~y to use in searches and when limiting the display of a mailbox. Thus, I could implement virtual folders by saving all mail to a single folder (thanks to the header_cache patch), then executing l~ydebian to see all messages tagged debian. The problem with this approach is that it uses a functionality that I already use extensively: limiting. I like the thought of being able to use tags as filter criteria, but I don't want to depend on them. For instance, if looking at all messages with the debian tag, and then filtering those with mdadm in the subject, I currently have to hit l<space>~s mdadm (taking care to preserve the label limit), and I cannot use the simple_search functionality anymore. In the end, what I'd want is to be able to use the folder syntax for tags. Specifically, I think it would be cool to have folders as they are currently, but if you try to change to a folder that does not exist, or prefix the folder with vfolder (=vfolder/debian/mdadm), it would instead display the main folder with the different path components used as label filters. A syntax for logical OR would also be nice. All this should happen without affecting the limit functionality of mutt. I don't think this is currently possible, but before I start to pretend to plan my intent of eventually thinking about what it would take to have a go at coding a patch, I'd welcome your feedback. I can't imagine that I am the only one wanting mutt with tags... NP: Log / Every Time a Bell Rings, an Angel Gets His Wings Update: people have pointed me to Gnus and Mairix. I'll have to check out Gnus, but I wonder whether it can be used the way I do mail: currently, I use offlineimap (my favourite software award 2004-2006) to synchronise my Maildirs across a total of eight machines, so that it does not matter where I read my mail -- it's totally synchronised. I'll give Gnus a whirl soon though. Manoj already showed it (off) to me, and several other Debian developers seem to swear by it, so it can't be all wrong. With respect to Mairix, I know that tool; it basically populates folders with symlinks according to search criteria. One could say it instantiates virtual folders. The problem I have not been able to solve yet is quite simply that operations such as moving messages to other folders then operates on the symlink, which is not what I want, for obvious reasons. Even just using mutt to edit them results in a copy made, so the symlink association is lost. I'd be happy to be shown how to use the tool to make it do what I want. NP: Led Zeppelin / Led Zeppelin I

I was very concerned by Martin Krafft's appeal to voters in the Debian Project Leader election to shut up about their own "biased" opinions on the race. He argued that the candidates should campaign and that anyone who wants to spread their point of view should be running themselves or keeping their opinions to themselves and getting real work done. Perhaps I just buy into the whole deliberative democracy concept but I think this attitude is dangerous. In fact, I think it's essential that Debian publicly weigh the benefits of possible decisions and discuss, argue, and debate as a group. I think that every instance of public discourse (and no, flaming is not discourse) on project policy or leadership is a sign of a healthy and involved electorate and I'd like to see more of it. The leadership of the Debian project is at stake in this election. Our organization is more complex than a wheel with the DPL at the center. Our decision should be made as a project with a complex organization. That means conversations need to work the way the project does. Obviously, we vote as individuals. But that's precisely why conversations and discussions, through which we can make decisions as a community, play such an important role in informing our votes. I'll leave talking about enfranchisement in Debian, and the lack thereof, for another day.

Scott, the first thing I want to say in response to your post is that you are not alone. I did not know someone was attacked for wearing an Ubuntu shirt in Helsinki, but I did witness at least one "Fuck Ubuntu" incident, and it seemed that Ubuntu jokes and snaps at the new distro were very common. I find this utterly childish and stupid. I am sorry for everyone who was offended. But please don't blame or condemn everyone in Debian or the whole project for the actions of some of its members or followers. You've already linked to my thoughts on the Ubuntu-Debian situation, so I need not cover any of that anymore. I wouldn't want to either, because believe it or not, this article got me flamed by Ubuntu developers just way too many times, in personal mails as well as in forums. Instead, let's look at the now and then this time. I've heard a few developers express their unhappiness with Debian in the past year, much in the same way that you did. Not always was it related to Ubuntu, but quite often it was. There are newcomers to Debian who shake their head in disbelief how anyone can work in a project such as Debian, and there are the veterans among us who seem to concede that Debian has slipped away from their ideals and it's just not fun anymore. It's not hard for me to understand each of them. I've followed Debian for nearly a decade now and thus tracked how it changed over time. We are one of the largest free software projects, and size never comes without problems. So while the Debian project used to be an association of geeks enjoying their focus on technical excellence, the sheer growth of our project called for many other fields to be managed, and we did not always handle this increase in complexity very gracefully. After all, we are all just geeks who want to have fun, aren't we? As a result, several aspects of our project have been said to "stagnate", and parts (and parties) have diverged. As a volunteer developer, it's not always fun to run up against walls and find that your innovativeness is met with stubborness or simple silence. I say "not always" because I think it's really just a matter of perspective. Over the years, Debian has seriously affected my life, the way I look at things, the way I interact with people, and the way I do stuff. I am eternally grateful for much of that. But the learning process has not stopped. What Debian needs is a stronger grasp of the big picture. In joining all the other developers who are already doing excellent work in that direction, what better learning opportunity could there be to approach large-scale project management, away from academia and the corporate world. For me, this is a very interesting perspective of a fascinating project, one that directly relates to my current research endeavour. I was asked recently how we "control the anarchy", and the first thing I said was that "control" was not the right word. Debian is not about control, it's about the flow. If it's not going your way, you can swim against it, you can pull out, or you can try to find ways to change that flow, bottom-up rather than top-down. I think David's idea of a steering committee is an excellent one, and I'd be happy to dedicate time to such an effort. I wouldn't call it "steering" though, as that seems too active. Rather, I envision a group of people dedicated to keeping track of the big picture (obviously in close contact with the release team) to which developers can turn if they need advice. In the last few years, I've often asked myself why the heck I don't just give up, instead of continuously running up against problems that seemed hopeless. Even now, I sometimes think that I should have just pulled the plug, but my answer has always been no, and it looks like it'll remain that way. Why? Because we've come an incredibly long way, and that by itself provide for a lot of motivation. Mostly, however, I decided to hang in there because I tend to like challenges, even if at first I have no idea how to conquer them. When I then look at what some of the people are doing for the project, I come to realise that there are solutions, even if the road is a long one. You (Scott) strike me as a very innovative person, and not only because the problems you and I are looking at tend to overlap to a large degree. I strongly respect you for your work, and Canonical certainly deserves a tip of the hat for providing you with the environment to do so. Sometimes, I even envy you a little, simply because I haven't been able to get down to coding solutions as much as I would have liked to and you've beaten me to the punch on things like HCT or upstart -- even though my work in that area has been mostly vapourware, so there was no real punch to which you've beaten me... I think you'll know what I mean. Yet, scanning some of your older blog posts, I got the impression that you weren't always happy with the way you have to do things at Canonical. And still, you're enthusiasm for working on your projects is astounding; it wasn't hard to come to that conclusion just a week ago, even though we've only talked briefly in Cambridge. Have you asked yourself why that is? I'll go out on a limb and postulate it's because of your innovativeness, because of your urge to push technology further, to implement cool things, and to break free of the cruft that comes with 13 years of Debian and several decades of standards la POSIX and SysV. Debian is too conservative for you, and all the cruft weighs heavy. Ubuntu, on the other handy, is "fluffy" and not afraid to forge ahead, and you've been an active part of it from day one. As much as I hate to see you "with them" rather than with us, I think you've found your place. And so have I. I'm less about coding than I am about integration, and there's work to do. Debian is a massive project with an amazing track record of achievements to look at. And it has problems. Well, great! Let's attack them. Never give up. Another reason why I've stayed with Debian is because there are people who think alike. Three years ago, I would have been able to give you a long bullet lists of hopeless problems in Debian. If you ask me today, I would have problems arguing their hopelessness, for I've seen plenty of developers taking it up with these problems, and in the meritocratic way. Our release team is doing a fantastic job and I've never felt so positive about the promise we made with the next release. People like J rg, Zobel, Rapha l, the two Alex, Meike, Christoph, and the many others I'll leave out only in the interest of brevity (the list is longer and doesn't only span Germany and France) are doing great work, and that is truly motivating. And I ought to mention explicitly our leader, Anthony, who's been doing an excellent job so far. I hope he can keep it up. Also, take a look at our IRC channels. I would have to be lying if I'd say nothing has changed. And the same applies to mailing lists. We still have a reputation to deal with, but if you stop listening to what people say and actually take a look, I think we've come a long way. I'd even say it's become more fun to work on Debian again! I think we owe a lot of that to Debian Women. I am definitely proud to be a member of this community (with which I mean Debian as a whole, not just Debian Women). I think I could continue this blog post endlessly. Let's not. You (Scott) may not be a fellow DD any longer, but you remain a fellow D, and I hope we'll continue to let our cooperation prosper. Now if only you had told me that "getting upstart installable required changes to twelve different packages, including sysvinit itself" when you approached me with what sounded like a simple sponsorship request... I've started to have a go at the "herculian effort", step by step. Seems like an interesting experiment. NP: Eels / Daisies of the Galaxy (it's really hard to find a better band for a grey Sunday morning) Update: Maybe our reputation isn't all that bad, or Slashdot posts suddenly became all useful. A lot of insightful and interesting posts in the first batch of comments...

I thought this mail about the current state and future of NetBSD (via /.) was an interesting read, and one relevant to Debian not only because of our Debian NetBSD port. In particular, it felt good to be able to say "yeah, we've got that" to some of the points Charles raised. However, I also had to concede that we're short in some of the other points. I won't put numbers to these claims because it's not as black and white as this. What would be really great is if one of the Debian long-timers would sit down and reply to Charles with his/her experience! Having worked a bunch with the NetBSD crowd in the past, I really would hate to see this project go down. The NetBSD kernel, while not up to the task to provide the desktop experience, still makes for rock-solid servers, and some parts of it sparkle with technical excellence (like the virtual memory management, for instance). Maybe, just maybe a possible future for NetBSD would be to produce Debian NetBSD in a joint venture? NP: Shpongle / Tales of the Inexpressible