I've taken part in many open source projects. Many times I haven't done much to mention, but sometimes I have done my fair share or even maintained the projects. This document lists the projects in which I have put my biggest efforts. See also Uncurled.

FrexxEd - 1991

I and Kjell Ericson felt unhappy with the text editors we had on the Amiga back in 1991. We decided it couldn't be that hard to write a better one, one that would fit our needs a bit more. We wrote that text editor during the years 1991 and 1996. It was a highly customizable and programmable editor designed for programmers. We released it as shareware during the time we developed it, but when we later got bored and stopped developing it, we changed the license so that it became free and open.

FPL - 1992

I started once upon the time, around 1992, to write a scripting language that we planned to use in our text editor FrexxEd which we wrote at the time. Frexx Programming Language grew pretty much without control and we changed direction many times during the first years of development. Eventually it turned out to become a very C-like interpreted language that was designed to let a program be able to add functions to the interpreter so that the language could be easily embedded and customized to applications. We also wrote a compiler that converted the ASCII-based source code into a byte-coded format for increased speed. The project was a closed-source one for a very long time but was later turned into an entirely open source project.

Dancer - 1993

I played around with IRC and wrote a first basic IRC bot sometime during 1993-1994. Bjorn Reese approached me when he had written his first bot embryo and together we developed Dancer to become one of the better defense-bots in the IRC world. The source was free and open from the start. Dancer got a rich set of commands and configurable features. It could hold a channel from being taken or harassed by attack bots and scripts fairly well. I was the maintainer of the project during around 1997 and 1998 until it was passed on to Tero Jänkä. Today, Dancer is decaying somewhat due to the lack of interested developers but there is still people involved in the project although both I and Bjorn have stopped.

fcpp - 1995

I once found a public domain cpp source code floating around. I developed it further and turned it into a full-blown ANSI-compliant C preprocessor. I also added enough options to make it useful to preprocess web pages and still today, most web pages I create are preprocessed using this tool.

curl - 1996

In late 1996 when I was involved in Dancer development I started to think about getting currency rates off the web to collect them in a database to offer bot users to exchange currencies online. To get the web pages, I wanted a small utility that could fetch a web page and nothing more. I found httpget by Rafael Sagula. It did exactly what I wanted. I soon found a very cool currency site on a gopher server so I added gopher to httpget too. When up to speed I later added FTP too and renamed the project to urlget since that seemed to be more appropriate then. As time passed, more and more things got into urlget and when I added the capability of uploading too I changed the project name again, this time to curl. curl is now a stand-alone tool that gets or sends data using URL syntax from and to a whole range of different protocols. The core of curl is now used in the separate library named libcurl that is being by other programs as well.

Hypermail - 1997

When we (me and my friends at Frontec where we worked back then) first started to play with web serving for real, sometime around 1996, we quickly discovered a need for a tool that could convert a mailing list archive into nice-looking HTML pages. We found hypermail v1.02 pretty soon and it didn't take long until we noticed hypermail's limitations; the biggest one being that it didn't handle MIME at all. I added some simple patches to make it deal with 8-bit Swedish characters at first, and later when hypermail v2 development took speed I wrote a whole MIME decoder for hypermail that not only dealt with the 8bit characters but also saved attachments properly and more. I also did some major cleanups in the hypermail v2 sources to make it deal with dynamic strings all over instead of the previous silly habit of static length buffers with buffer overrun risks and my config file parser replaced the former one.

mail2sms - 1998

My brother Björn Stenberg wrote a TCP/IP-based server for sending messages to GSM phones already 1995 something. He also wrote a client that could submit messages to the server and that client could interpret MIME mails, although with some problems. I wrote a script pretty soon that would take a message in a mail and send it as an SMS to Björn's server to all telephone numbers specified in the subject. You could make groups that enabled easy sending to a whole range of telephones. After a while, the limits of Björn's MIME parsing code started to appear more and more often and I wrote mail2sms, based on some code I already wrote for the hypermail project. mail2sms offers not only full MIME decoding but also regex search/replace, conditional replacing, time/date dependent actions and more.

spam.pl - 1997

I've been active on the Internet since 1993 and I think that especially in the beginning it seemed as if many spammers picked email addresses from the Usenet newsgroups. I wrote a lot of posts there and the more time that passed, the more spam I got. In 1997 I decided that I'd finally write a script that could extract all relevant information by itself from offending mails and send a complaint to the correct receivers. So, I wrote it, using plain Bourne shell. It turned out to be rather slow and in 1998 I rewrote it using perl and not only did it get faster but I also improved the parsing a lot. The script has evolved since, but has been used for hundreds if not thousands of spam complaints, many of those have been heard and have been the cause of closed accounts.

Netracer - 1998

In 1998 I and friends started to think about writing a game. It should be a game in which you compete by writing agents/clients that then fight against the ones your friends have written. The best programmed client wins. We soon landed on the car-game idea and netracer was born. A server keeps the track and car information and all clients that connect get a selected amount of information and can tell the server how it wants to run. Ambitious as we were we even attempted to make it reality in the real world! We got an R/C car and we were supposed to add a CPU-board, PCMCIA-based code in java, sonars, laser-things for distance-measuring, live video from the car and more. However, it soon turned out that the happy bunch what were to make this reality was lacking something essential: hardware skill. We didn't manage to attract any of the hardware builders at our company to join our project and soon the project died again. We went back to netracer, we redesigned it. we wrote a new protocol, practically a new server and now in November 1999, we are on track again and hopefully we'll be able to compete within a few months!

Triacle - 1999

When Bjorn Reese and I had written Dancer for several years and made a huge program out of it, it suffered from a slightly bad design. We wanted to add new features to it, but the basic design of the bot was so that it became terribly hard to do so without breaking a lot of old stuff or making very ugly kludges necessary. We decided the only sane thing to do was to write a new bot, all from the start and do a lot of design correct from the start this time. We knew what writing a bot means after years of experience. We got as far as to a pretty good ground and something to build on, a core that did most of the networking layer when we gave up. We found out that we just didn't IRC much any more nor did we run any bots actively. If you don't run a bot, how could we develop a good one? We did some nice work for triacle that has later been possible to re-use in other projects, so it wasn't all in vain.

Trio/dprintf - 1999

Involved as I have been in several projects including portable Unix software, it is obvious that not even the printf() series of functions are truly portable to the extent you'd want them to be. Being aware of this and wanting this to improve in the projects I'm involved in, I and Bjorn Reese initially wrote an implementation of the printf() functions and I called the package dprintf() (the name came from that my original package was daniel-printf and when I merged with Bjorn's stuff I just kept my name). The package was not really complete and after the triacle work had been done, we stood with a set of useful string functions for C and a set of incomplete printf() functions. Bjorn did most of the job of incorporating all those into the single package trio which now has a complete set of printf() functions, scanf() functions and a few other string-functions meant for portable Unix programming.

smash - 2000

Several years ago my brother wrote the SMS server I wrote about above. Now, the second generation of that is getting developed. We've designed a neat ASCII-based HTTP-like protocol that enables us to write clients easily on any platform using the language you want. There should be a plug-in system to enable many different protocols and engines for different operators and we do use an entirely new and better file storage syntax. This project is GPL licensed.

Subversion - 2000

In the mid-year 2000 people on collab.net started hacking on a replacement for CVS. I joined the project in the early autumn and I've been a contributor to the project since. I'm not any main player in this project, I mainly lurk in the background and commit odd fixes every now and then. This project really has the odds of becoming the future number one version control system, at least for open source development...

RockBox - 2001

Archos makes hard-disk based MP3-players. Rockbox is the open source project initiated in late year 2001 by my brother Björn to write an open source firmware for their products. I'm an avid supporter, participant and contributor since before I bought my own Archos Recorder. We write everything from scratch. The Archos target hardware is a SH7034 at 12MHz, mp3 decoding is done by a MAS3507D or MAS3587F.

Rockbox expanded to cover lots of more music players, including brands named iriver, iAudio and iPod.

roffit - 2003

Getting man pages rendered as nice web pages turned out to be a problem, so when I took that upon myself for the curl website I ended up create my own dedicated tool for this purpose.

I decided to call it roffit as it works on the nroff source format (man pages) and outputs HTML.

c-ares - 2004

During the summer 2002 Bjorn Reese and I once again joined up and started planning and organizing project 'Denise'. We intended to write a fine, multi-platform asynchronous name resolver library with a liberal license.

My own participation in this project made me instead adopt and fork the ares project to produce c-ares. My own involvement in c-ares is primarily to provide a good and solid asynch resolve library for libcurl.

libssh2 - 2006

I accepted a contract job to add SCP and SFTP support to curl and to make that work, I needed a SSH library to use. After having contacted both the libssh and libssh2 projects, the better answers came from libssh2. I joined the project, work on improvements I thought were needed and eventually took over as maintainer.

I later have tried to take a less active role in the project. I still do releases but the day to day maintenance is managed by by awesome co-maintainers.

Firefox - 2014

I was employed by Mozilla from early 2014 to the end of 2018, where it was my full time job to work in the Firefox networking team. We were responsible for HTTP, DNS, cookies, sockets and more in the browser engine.

trurl - 2023

It can be a challenge to write shell scripts that have to manage, parse or manipulate URLs. To help solve this, we created the tool trurl in the spring of 2023. It uses the URL API provided by libcurl.