Daniël de Kok

Memoirs of a Libranet developer

Oct 31, 2010

Tag: linux

Since my memory of working at Libranet is starting to wane, I feel I have to write down some memories of that time. It is a story about dedication to craft, and how things can turn sour. I will probably make some updates to this post as I find new things.

Prehistory

I jumped to the UNIX side of the fence in 1994. Being twelve years old, it was the only way to legally get a large amount of free (as in gratis) software. Regardless of my thoughts of GNU and the Free Software Foundation, this is a quality of free software not to skim over: lots of hackers found their calling, because they were allowed to tinker with their system. As many users of the day, I started out with Slackware Linux, but also tried many other distributions in the following years. In 1996 I bought my first FreeBSD CD set and worked with FreeBSD, later NetBSD, for years.

In 2004 I got fed of using BSD full-time: my studies required the use of software that ran only on Linux and Windows. Additionally, the time to keep my system in a running state was drastically shrinking. So, I was on the lookout of a easy-to-use UNIX system. OS X was out of my reach on a student budget. After careful searching, I settled on Libranet. I disliked non-Debian/Slackware distributions, and Xandros looked far too corporate. Libranet breathed a sense of craft and community. After trying the then free 2.7.0, I purchased Libranet 2.8.0. In the days, Libranet provided an ncurses installer that was friendlier than the old Debian Installer. Post-installation it also provided a user-friendly system administration tool, named Adminmenu.

After the release of Libranet 2.8.1 the team behind Libranet, John and Tal, started working on the next major release. Around this time I had become an active community member: I was a regular forum poster and beta tester, worked on a Libranet Guide, and sent in some small patches.

Working at Libranet

Suddenly, just before Christmas, I received an e-mail from John whether I was willing to work for Libranet. This was the greatest Christmas present I could wish for: getting paid to develop a Linux system. I enthusiastically reacted, and took the job.

At the time, development of Libranet 3.0 was in full progress. The framework for the new Gtk+2-based Adminmenu was done, there were some components for Adminmenu, a very good disk partitioner, and a working installer. In this phase, my work practically consisted of writing additional components for Adminmenu and adding documentation to each Adminmenu component and installation step. If I remember correctly, I wrote the following Adminmenu components: the help browser, CrossOver Office installation, Ndiswrapper, IPsec, firmware installation, services, disk usage, user settings restoration, cron times and numlock components. Some components were really fancy, such as the IPsec configuration and the kernel manager that Tal wrote, and are still unparalleled today in popular distributions.

After the initial spree of development, a QA phase started where we polished the whole system based on feedback from the private group of beta testers. This involved maintenance of upstream software as well. We maintained a set of kernel patches, and my first experience with the Red Hat bugtracker was by virtue of sending bugfixes to a hardware detection tool that we used, named Kudzu. It was a lot of fun to see our product come together, including things like packaging. Our beta testing group also seemed to be generally happy.

Ship it!

In April 2005 we shipped Libranet. Just having looked up that date, I am amazed about the amount of work we did in just three months: developing, QAing and releasing a Linux distribution. John sent out review copies to the usual suspects, and journalists and users generally gave it positive reviews.

The demise of Libranet

In the following months, the factors contributing to Libranet’s demise started to come into play. First, sales were not going so well. Libranet 3 did ok within our own community, but was unable to expand beyond that group. I believe one of the primary factors was Ubuntu. I believe that we were ahead with our installer and administration tools. But Ubuntu did provide a good out-of-the box experience, had great marketing, and was free (as in freedom and beer).

After this, an even huger blow to the community, but especially Tal, was the passing of John, Tal’s father. John was the heart of Libranet. And in a fully understandable turn of events, Tal decided that he needed some time off to reconsider life and work.

My own contribution to Libranet was also shrinking. After the release of Libranet 3, I was primarily handling user support, until we would start working on the next release. Since user support required regular attention, it did not allow for flexible time planning. I needed flexibility, since I was also studying for a university degree.

After a few months (I think near the end of the year), Tal pulled the plug. Libranet folded.

OpenLibranet

In the years after Libranet, I sometimes received questions from ex-users asking why we didn’t open source the Libranet installer and Adminmenu. What few know, since it was all happening behind the scenes, is that we tried. Tal didn’t just want to drop the sources on the net. Understandably, because Libranet was so stongly connected to his family, he did not like to see it dwindle. But he did make a very generous offer: if someone offered a good home to the Libranet software, he would be willing to opensource it.

So, we saw this as a hint, and built a team from community members. We had developers, translators, and user-interface experts. We wrote a plan for the future development of ‘OpenLibranet’. The first version, OpenLibranet 3.1, would polish Libranet 3.0 further, and would, if I recall correctly, switch to Xfce plus Thunar as its default desktop environment.

Tal was impressed with the plan, and would go for it, under one condition: we would develop OpenLibranet 3.1, and when it was done, he would give a final ‘yes’ or ‘no’. A ‘yes’ meant he approved, and Libranet could be opensourced, a ‘no’ was the end. This condition was understandable, but a deal-breaker to us: we could have worked for months, years in people time, potentially all for nothing. In hindsight, I think we should have taken the deal.

This marked the end of Libranet and potential revivals. I’d still like to ask Tal sometime if he is willing to put the Libranet source tree up for historical reference (even under a non-opensource license). I lost it, and am curious to have another look. But I admit not having the guts yet to ask ;).

After Libranet

After Libranet, I finished my university studies, and I am now employed as a PhD candidate in computational linguistics. On the desktop I switched to Mac OS X full-time. Sometimes with pain in my heart, but it is a polished system that helps me get to work done. Linux became too fragmented and rough to me.