LCA04 Tux logo
LCA04 Main Logo
Call for Papers
Media Centre
Partner Programme
Works in Progress
Event Record
Dymocks Books
Inetd LiveCD
Speaker Coaching
About SA
Linux Australia







Sun Microsystems


The University of Adelaide


NetCraft Australia










Australian Developer Magazine

Other LCAs

2005 Canberra
2003 Perth
2002 Brisbane
2001 Sydney
1999 Melbourne


LCA 2004 started at Wed Jan 14 08:00:00 2004.

HP FIXITs (Formerly BoF Sessions)

Run one, or be at one! (Last updated 16/01/2004)

Program and Submission Requirements (for the prize) AVAILABLE HERE

What is a FIXIT?

BoF sessions are interactive peer sessions where you can discuss a given topic with a group, usually in an informal context. Most Unix-related conferences have BoFs, but for we thought we'd improve on the concept a bit. A BoF is for sharing and absorbing information, but so is just about every other minute of the conference.

So, a FIXIT is a BoF with an outcome. The sorts of outcomes we are looking for are ones which contribute, even if only a tiny little bit, to open source. See below for a list of examples. Think of a FIXIT as a chance to capture in electrons the sorts of valuable conversations, arguments, seat-of-the-pants debugging sessions or wild guesses that are happening all the time at

In other words, if there's something in Open Source that you think could do with a bit of fixing, it might just be a really ideal FIXIT session to hold at LCA!

We'll be recording outcomes on this webpage; possibly elsewhere. And YES, there are great PRIZES for the best one!

What does FIXIT stand for?

Nothing. It has all the kudos associated with an important-looking acronym without the bother of having to remember what it means. A bit like the HOWTOs.

I heard prizes! What are they?

Thanks to HP, we've got some great prizes lined up! They are as follows:

1st Prizes
Leader: HP iPaq with WLAN card
Top 4 contributing members: HP iPaq

2nd & 3rd Prizes
Leader: HP iPaq
Top 4 contributing members: WLAN card

More information about the HP iPaq, and Familiar Linux, is available here:

When are the FIXITs?

The third day of the main conference (16th January); running from 1615-1800 (1 hour, 45 minutes). See the timetable for more information.

Many conferences have BoFs running at any lunchtime or other break during the day. has decided that it would be better to have a specific slot.

Some sessions will run for 45 minutes, others may run for the whole duration; it all depends on how popular the session seems to be, judged on pre-conf discussions (see further down).

The day of the FIXITs, a timetable with each session, it's duration, and which room it's located in will be published both here and displayed on posters around LCA - information about the way you need to submit the summary of the session will also be made available. It's important you follow this if you want to be in the running for the prize :-)

How can I run one?

Email the FIXIT organiser (Ryan Verner) including:
  1. A paragraph (or more, if you like) about the topic
  2. The outcome you hope for
  3. (optional) A brief introduction about you and your background
  4. NEW: Your preferred discussion method pre/post-LCA (forum/mailing list)

After the FIXIT session, how do I submit my summary?

At the conclusion of the FIXIT, you can use either the discussion forum, or mailing list (see below) to post your outcome details. To be in the running for the great prizes, you'll need to email it to the FIXIT organiser directly (email above) that same day for judgement. There will be certain questions you need to answer (made available on the day of the FIXITs) - nothing very difficult, but it's to make things far more consistent so judging is fairer. Watch this space! Remember, the FIXIT's are all about outcomes and making a difference to Open Source.

Can you give me examples of outcomes?

Easy! They are the sort of thing that someone can usually type up in ten minutes (but require some creative thinking beforehand), but which move forward some aspect of open source a bit.

Things such as:

  • A written-down list of problems with a particular piece of software, with details of how to reproduce them or some clues about where the problem might be.
  • User Mode Linux, why its useful for ordinary LCA attendees
  • vi vs emacs: 5 things that each can do simply that the other finds hard
  • Reasons why some Linux distro/BSD's installer is bad
  • The most nasty C code in an open source server. Can you top file locking in fetchmail?
  • The Kernel That Will Replace Linux

What topics are acceptable?

Any subject relating to Linux or Open Source is preferred, but we're open on anything suitable that may be of interest. You can see from the example outcomes listed in the previous answer what the FIXIT topics probably were that generated them.

And when the FIXIT is accepted?

You'll get an email notification and your session will be listed on this place together with the location we have allocated.

I've got this wonderful topic, but I'm nervous...

FIXITs are very informal, laid back events, but if you need a hand structuring your session, just let us know - we can help you out. "She'll be right, mate!".

When can I submit my FIXIT topic?

Now! We'll be accepting topics right up until the first day of the main conference, but the sooner, the better. Each session is assigned it's own area/room, so the earlier you apply, the better we can organise it.

Where can I discuss these FIXIT's, pre and post-LCA?

There's two types of geeks out there; those who prefer online forums, and those who prefer mailing lists. We've provided both for you to discuss current and possible FIXIT's pre and post-LCA. Don't just discuss at LCA what you want to fix; make good use of the list/forum, and actually make it happen :-)

The Discussion Forums are here, kindly hosted by linmagau.

The Mailing List Signup Page is here, kindly hosted by Linux Australia.

Each FIXIT will have a preferred discussion method listed along with it; it's recommended you use that. The forums have a 'poll' feature, where you can vote for your favourite FIXIT - this will help with allocation of rooms and session length (depending on popularity). Making use of the discussion mailing list will also indicate to us how popular that session is.

Why would I run one?

Plenty of reasons, and besides, they're fun :-)
  • Get the chance to introduce that project you've been working on for a while.
  • Discuss/solve problems that you and others perhaps have been sharing.
  • Demonstrate that cool new program that you discovered!
  • Get reactions back from people about some brilliant new idea you've had.
  • Chew the fat about the nasty problem that exists in foobar
  • & plenty more...

What FIXITs have been planned so far?

DotGNU (list) run by Rhys Weatherley
DotGNU ( is a project to bring .NET technologies to GNU/Linux.

A number of DotGNU members will be attending LCA2004, and this FIXIT's aim is to get feedback from other open source community members as to what they'd like us to investigate that would help them.

The desired outcome would be a better understanding of our goals and a list of tasks for DotGNU to tackle during the upcoming year.

Rhys Weatherley is a member of the DotGNU Steering Committee, and the author of Portable.NET.

Rhys graduated from The University of Queensland, Brisbane, Australia, with an honours degree in Computer Science in 1990. Since then, he has worked in a number of positions at Australian universities and US companies.

He returned to Australia in late 1999 to pursue his own interests, including the founding of his Free Software company, Southern Storm Software, Pty Ltd.

His computer interests include programming language design and implementation, network computing, and user interfaces.

LCA 2005 Organisation run by Linux Australia
Discussion for LCA 2005; for those interested in assisting, or organising subsquent ones. More information to come.

Community Testing of Linux (list) run by Richard Sharpe

A harried project manager needs to know what collection of versions on the Linux systems he will deploy he needs to upgrade to or downgrade to to avoid the latest problem that the development team found.

An experienced software developer would like to contribute to Linux or some project surrounding it but doesn't know where to begin.

Each of these people are looking for information, and on a daily basis, many people working with Linux would like to know what bugs are they potentially facing or to have a higher level of confidence that the software they are using has been tested in some way. The problem as I see it is that there is not enough testing being done for Linux, and what testing is done is fragmented among many different groups of people with different tests and different databases.

I propose that the Open Software Community embark on a distributed testing project that is designed to use the large number of people who would like to contribute anything, even if it is only some spare CPU cycles on their collection of machines and a little bit of their time to start the tests.


This project will develop a testing framework and a data collection facility that allows many people around the world to develop and contribute tests. In addition, it will encourage people to run tests on all sorts of systems to maximize the information return.

The goal must be to gather testing data about as many collections of software and versions of that software as possible, so that we can answer questions about what set of versions of software pass or fail various tests.

In addition, all the tests must be scriptable so that they can be run in an automated fashion on any machine or set of machines.

Further, the testing framework must allow for tests:
  • To be run from and against the same system
  • To be run from one system against another system
  • To be run from one system while involving muyltiple other client and server systems
  • Possibly even to allow for multiple test drivers.


The vision I have here is that at some time in the future it will be possible for people to interact with the growing collection of test definitions, tests and results to pose and answer questions like:
  • What problems exist out there that I could work on to fix?
  • What tests exist in this area so I can think about devising a new test and then implement it.
  • Why did this particular problem not get picked up by existing testing and can I define a new test that is needed.
  • What tests have been defined but have not yet been implemented so I can find one to create.
  • What set of versions does this problem manifest itself on so I can pin it down more?
  • What confidence should I have that this set of software (Kernel, utilities and some applications) has been tested in my type of environment?
  • How much more testing should we as a medium or large organization do to assure ourselves that the version of Linux and these apps we want to use will actually work in our environment.


There are many problems with this project and one of the goals of the FIXIT would be to come up with solutions to them. Some of the problems are:
  • What testing framework to use
  • How to stop people from poisoning the database
  • What info do we want to collect
  • How to get enough info into the database
  • What to test? (Everything) (Kernel, APIs, X, protocol implementations, applications, Databases, web servers, etc, but start with the basics.)
  • How to structure the tests
  • How to manage the tests (including how to prevent invalid or broken tests from entering the system
  • The reports it should be possible to pull from the database
  • Where the database should go
  • What Applications are needed to interact with the database ... ie, something that can answer what components should I upgrade to get to a collection that overcomes the problem I currently have.
  • Integration with bugzilla ...
  • ... and so on


Collect ideas and work through some of the issues.

Assemble an initial team to start the technical work.

Richard Sharpe has worked in the computer industry for over 20 years and has contributed to two Open Source projects: Samba and Ethereal. He currently works for a storage startup called Panasas ( where he works on Samba, FreeBSD, and Linux.

Computerbank (list) run by Dan Treacy

The Computerbank Project is an Australia-wide grassroots volunteer effort to bridge the digital divide and keep perfectly usable machines out of our landfills. Working with machines kindly donated by both corporate entities and individuals this community based effort uses Linux and other Open Source technologies in it's aim to get computers into the hands of people and groups who might not ordinarily be able to afford them. This also enables us to give people access to technology that is now becoming almost a requirement of society whilst having a beneficial enviromental outcome as well.


Any or all of the following:
  • An increase in awareness of Computerbank and it's aim and objectives within the Linux/Open Source Community. Spreading the word of the work we do.
  • Seeking interest in starting Computerbank units in areas not already covered by an existing Computerbank operation.
  • Increase in membership/volunteers for exisitng Computerbank units.
  • Solving any pressing technical problems being currently encountered by any of the Computerbank units.
  • A gathering and fostering of closer ties between existing members and volunteers doing Computerbank around the country.
  • Discussions with regard to the restructing of the National body Computerbank Australia to best help both the operational units and Computerbank as a whole to achieve their goals.

Dan Treacy currently wears the many hats of President Of Computerbank Sydney, Vice president of Computerbank NSW (and NSW Co-ordinator) and a member of the Computerbank Australia Inc. Committee. He is currently self employed in the IT industry as has been using both computers and Linux for the best part of 10 years.

AutoFS (list) run by Tim Hockin
I'm the lead on a project that Sun has put together to examine autofs in Linux. Linux's autofs does not interoperate well with Solaris (maps don't parse, unsupported features, etc) and it does not work well for large installations (various reasons). We've been tasked with making it play more nicely with Solaris.

In so doing, we've been working on a couple of papers covering the multitude of problems and proposing an architecture for a next-generation autofs. Some of the things we're trying to tackle: expiry, map compatibility, direct maps, browsing, namespaces.

I'd like to get together with anyone who is interested and discuss the problems, the ideas, and potential solutions. The outcome I'd like to see would be either a shootdown of our current thoughts, a rework of those thoughts, or some nodding heads.

Tim Hockin has been using Linux since 1.2.3 was fresh. I've helped out here and there on various projects, including the Linux kernel and various daemons. I was a software engineer at Cobalt Networks, where I worked on the Linux kernel, the Cobalt ROM, drivers, apps, and whatever needed to be done. When Cobalt was acquired by Sun Microsystems, I found myself in Sun's new Linux Software Engineering group. I am currently a staff engineer at Sun, working on various Linux kernel and other issues. Though I work at Sun, don't ask me about Solaris, I just can't help!

Mandrake (list) run by Leon Brooks
Mandrake users, including Cooker denizens, will be attending the greatest conference in the southern hemisphere. This FIXIT will give us a chance to tag some names with faces, and the end result should be two lists.

One of questions and suggestions which will be put to Gael Duval and Frederic Lepied (Fred is Mandrake's Engineering Manager) and one of thanks and commendations which will be widely published.

The questions will emphasise procedural and structural considerations, but not entirely at the expense of technical issues.

The purpose of all of this is to make one of the world's most widely loved Linux distributions even better by identifying and distilling down the best and worst features for emphasis and elimination. This distilled list will be more valuable to decision-makers at MandrakeSoft than hundreds of random emails, and will hopefully expose opportunities for Mandrake contributors to be more effective.

If you plan to attend, please drop me a line so I know who to expect.

Leon Brooks is a Linux consultant in Perth, Western Australia. He has written numerous small FOSS tools and contibuted to larger projects, is a Committee member of Linux Australia, The Society of Linux Professionals (WA) and Perth Linux User Group, writes articles for Australian Developer Magazine and is a pro-am loudmouth on assorted 'blogs.

MySQL Database Design & Optimisation (list) run by Arjen Lentz
A guided discussion about methods and best practices for optimising MySQL installations and applications, from database design to server and query optimisation. "How to make the MySQL dolphin fly" ;-)

Intended outcomes include:
a) Increased awareness of methods and possibilities for making MySQL database applications run as efficiently as possible, by exchanging knowledge and experience.
b) Getting to know people in Australia who can help eachother with such matters.
c) Take the first step in connecting these people.

Arjen Lentz works for MySQL AB as lead technical writer of the MySQL documentation team. He also teaches MySQL training courses in Australia.

Kernel mppe support (list) run by James Cameron
Many distributions lack support for microsoft point to point encryption (mppe) despite a kernel module being available in the ppp 2.4.2 sources. mppe support is not in kernel 2.4.x or 2.6.x.

This is causing extra support work for the pptp projects, as pptp depends on mppe.

An example of how to get the mppe module into a debian system:

License is bsd without advertising clause.


Code review, modifications, testing, and peer support to increase the probability of acceptance of the ppp 2.4.2 mppe module into stable and development kernel sources.

James Cameron is the release engineer for pptp client. He has never contributed to the kernel before.

Qbalt (list) run by Ryan Verner
Qbalt's original aim was to release a more capable, alternative OS for the nifty Cobalt series (Qube, RaQ), now EOL'ed by Sun, which will be updated and added to regulary by volunteers.

Even though this is still true, and initial releases will be primarily for Cobalt units, the main aim now is to develop a Cobalt-like OS for white boxes (x86, initially, with MIPS/PPC following). As a result, development's taken rather longer than initially intended and I'm keen to start getting others involved with the project.

Qbalt is based on Debian Stable, with backports maintained by the Qbalt team; it takes many lessons learnt from the success/popularity of the Cobalt interface; the methodologies of an 'appliance'. In short, it makes the project very different from other existing offerings such as Webmin, which aren't really designed for endusers, and very different from other 'panels' such as Cpanel, Plesk, and Ensim, which although are quite capable, seem to have completely ignored the KISS method and don't make very effective appliances.

What also makes Qbalt different from these others is the fact that almost everything is manipulated through abstraction layers; long story short, it results in flexibility never seen before - use the MTA of your choice, for example (given a definition exists), yet the UI stays exactly the same.

I'm keen to both explain the mentalities behind Qbalt, get input from current users and developers of Cobalt units, and possibly get some developers onboard. It may also be a chance to show off what's already been worked on, and plan out what's to be done in the coming year.

I'm disqualifying this session from the FIXIT prize, as I'm the organiser of the sessions and it just wouldn't seem fair to deem myself a winner, as nice as the prizes are. ;-)

Ryan Verner is a web applications developer from Adelaide, Australia. He's worked in many subindustries of IT, such as multimedia, consultation, development, and sysadmin, and is an organiser of LCA2004.

SPAM (list) run by Rusty Russell
There are multiple approaches to SPAM prevention. All have different problems: some require much infrastructure, many provide very limited protection. These include Sender Permitted From, distributed clearing houses, duplicate filtering, statistical content analysis, signatures, whitelisting, and others. A coherent plan involves more than one of these technologies, and their deployment.

Intended outcome: a plan, and a trial group of people prepared to test different spam protection proof-of-concept implementations.

Rusty Russell is a kernel hacker who knows very little about spam.

Networking The North wireless project (forum) run by Ian Loxton
Ian has been a passionate supporter / advocate of open source and community software for many years, more recently seen in work within LinuxSA and the Community Software Distribution Network as well as here at LCA2004. He has also spent many years supporting a range of groups in Salisbury and the northern area including Scouting, Schools, Community Houses and others. He is the Chair of Salisbury North Neighborhood Network Inc (SNNN) which is a project in its 3rd year of providing free computer & Internet access to the local community. The "Networking The North" project has grown out of the principles behind SNNN - building up the social fabric of the community by breaking down the digital divide. Married with 4 children he is 50-something and lives in Salisbury North, South Australia.

The Northern region encompassing the City of Salisbury and the City of Playford is a region with large pockets of economically depressed suburbs, large unemployment, high youth crime and high teen pregnancies. Community groups struggle to stay in touch and keep ahead of developments affecting their services. The opportunity for collaboration and mutual support only happens rarely. The region is noted for large portions being connected through pair-gain telephone exchanges thus Internet access is spotty at best. Young people especially have little IT skills which limits their chances of jobs in a job market increasingly relying on some form of IT knowledge.

Connect these community services through an Intranet allowing full collaboration, new sharing, peer support and problem solving. Provide them with a better quality Internet access at a reasonable price.
Provide the community with opportunities and training to build the social capital of the community and provide accreditation for this training and development.
Develop a sustainable resource to enhance the lives of the region's citizens.
Build a wireless network infrastructure to act as the tool to achieve these outcomes.

Preliminary development shows the infrastructure to comprise 9 - 11 radio towers distributing Internet/Intranet through an 802.11 network to around 3 - 400 client sites.

The primary outcome of the FIXIT is to develop technical specifications that would be appropriate to operate this network. This specification will be a guide to;
the types of antenna that would give the results required
the probable hardware need to operate and main the network
the kinds of open source software that could be utilised within the project.

Ian has been a passionate supporter / advocate of open source and community software for many years, more recently seen in work within LinuxSA and the Community Software Distribution Network as well as here at LCA2004. He has also spent many years supporting a range of groups in Salisbury and the northern area including Scouting, Schools, Community Houses and others. He is the Chair of Salisbury North Neighborhood Network Inc (SNNN) which is a project in its 3rd year of providing free computer & Internet access to the local community. The "Networking The North" project has grown out of the principles behind SNNN - building up the social fabric of the community by breaking down the digital divide. Married with 4 children he is 50-something and lives in Salisbury North, South Australia.

Working with arch/tla (list) run by Robert Collins
tla is a recent revision control system, with some very very cool and useful features (such as real distribution, offline work, smart merging, gpg signing....). However, it appears to be hard to learn without some interaction - a FIXIT would allow a bunch of folk to learn at once.

Outcome: At least one person more who understands what tla can do for them, and how to get it to do it for them.

I'm happy to be approached about tla anytime during LCA, and afterwards via irc ( #tla) or on the gnu arch mailing list I'm happy for non-public stuff to come to me at

I earn my crust as an IT consultant, doing various things from system admin through to linux appliance creation. I hack heavily on squid and tla, amongst other things.

Moving Linux Hand Held Devices Into The Mass Market (list) run by Tim Ansell
Linux is a fast, stable, compact system which seems ideal to be used in hand held devices. A lot of research has been done into Linux on hand held devices. As well a few commercial devices have been produced including PDAs Phones and WebPads. However there has been little penetration past the Geek market, Palm and PocketPC still maintain 100% of the market.

Most of the citied reasons include the lack of applications and poor core (such as PIM) applications.

The idea of this FIXIT is to discover ways to make life easier for the developers and users when targeting Linux devices. This should help to improve core applications and application support.

Related Links:

Desktop Environments:
  1. Qtopia (
  2. Opie (
  3. GPE (
Distribution Projects:
  1. OpenZaurus (
  2. (
  1. Linux Devices (
  2. The Sharp Zaurus PDA (
  3. The Samsung Yopi PDA (
  4. The Agenda PDA

I'm a student at the University of Adelaide (studying B.Eng (IT&T) + B.Arts). I work at AirNet, a wireless ISP in Adelaide, doing program development and Linux System Administration.

I am highly interested in Linux on PDA's and currently I own a Sharp Zaurus C700. I have been doing minor work for both the OpenZaurus and Opie projects for about a year. I also maintain the oz-compat package which allows many commercial applications to be run on OpenZaurus. On a side note I also offer rewards to people who develop solutions which I think are important to the advancement of the Zaurus (and Linux on hand helds) platform (

Debian Apache (list) run by Daniel Stone
The debian apache fixit aims at fixing a few problems in the debian packages apache, specifically
  1. vhost-base -- a set of scripts to add virtual hosts to any number of
  2. packages (postfix, apache, etc) in one command ,as well as to remove,
  3. change them etc.
  1. create a debian package for php4 in apache2 (which copiles diffeent
  2. compile option
  1. Closing apache2 bugs including the APXS bug
  2. General reducing the bug count on apache

Daniel Stone is a developer on many FLOSS projects, including Debian and He currently works on the X libraries, Debian's XFree86 and apache2 packaging, and other coding projectgs. He has been working on open source since 1999, and spoke at the Debian mini-conference, on maintaining large packages.

Annodex codefest (list) run by Silvia Pfeiffer
Having taken much video footage at the conference, we'd really like to set up a Website where we can search for the cool scenes of the conference and browse the footage in the Web way. The Annodex technology as presented in the Tutorial on Wednesday allows just that, in a way integrated with existing Web infrastructure. The core developers of Annodex are at the conference and keen to code and develop with you. So, let's start making Annodex video footage and install the tools, fix the bugs, improve the browsers, hack on the Apache modules, hack on new language bindings etc. ... just have an Annodex codefest!

Check out

Silvia Pfeiffer, scientist at CSIRO Sydney, research background from audio and video content analysis and signal processing, author of the Ogg RFC, contributions to MPEG-7 standards, author of MPEG maaate, for Annodex: author of libcmml, mod_cmml.

Conrad Parker, senior software engineer at CSIRO Sydney, crazy audio hacker, author of sweep, AUBE, speedmine, boids, and tractorgen, for Annodex: author of libannodex, liboggz, xine Annodex support, mod_annodex.

Andre Pang, software engineer at CSIRO Sydney, enjoys mixing music, loves Macs, zsh and haskell, author of vimacs, for Annodex: author of the OS X AnnodexBrowser, video-lan Annodex support.
Linux® is a registered trademark of Linus Torvalds.
The original Tux penguin is Copyright © by Larry Ewing.
All material on this page is Copyright © "Linux Australia" unless otherwise noted.