OctConf 2012 report

OctConf 2012 was a success. Over the course of five days, we had 20 participants involved, and all but two were there for at least four days. Big thanks to the Centre de Recherches Mathématiques who helped us with all of the organising and logistics.

The first two days were informational and dedicated to getting everyone up to speed on development methodologies. The next three days were a more involved open discussion and development. The schedule was followed with almost no modification. We were sadly unable to record talks due to not being able to secure proper equipment, but we do have the slides.

The following talks took place:

What is Octave? by Jordi Gutiérrez Hermoso
Introductory talk. Raised a few points for future discussion.
Octave’s architecture by Jordi Gutiérrez Hermoso
A discussion about the source code diretory layout and how to find code. Prompted a wiki page about debugging.
Status of Octave by John W. Eaton
Guiding talk about principles and policies plus general state of Octave development given by lead dev.
Octave Speed and Cell Arrays by Daniel Sebald
Raised problems with execution speed related to cell arrays along with possible solutions. Prompted some ideas for how to proceed.
JIT Compiler by Max Brister
First GSoC talk. An explanation of JIT compiling and the important single static assignment (SSA) technique.
The Octave GUI by Jacob Dawid
Second GSoC talk. The upcoming GUI design and layout. Implementation details in Qt. Lots of discussion about how it should look and what Qt can do.
Agora Octave and Packaging by Carlo de Falco, David Pinto and Juan Pablo Carbajal.
An open discussion about how we need to reorganise Octave and code distribution. Resulted in moving Agora to the agora.octave.org domain, and a roadmap for how to proceed.
Bringing Least-Squares Spectral Analysis to Octave by Ben Lewis
Third GSoC talk. Discussion of the method and implementation details. Lots of healthy discussion.
BIM package by Carlo de Falco
An interesting discussion about how to solve an diffusion-advection reaction boundary value problem with Octave tools and other free software.

In addition to the talks, many other important ideas were discussed and decisions about the development of Octave took place. A few highlights:

  • OctConf will now take place annually, with the next one taking place in Europe. The details should be ready within six months from today.
  • Various suggestions on where to improve documentation.
  • Rearrangements of source tree layout
  • Distribution of Windows and Mac OS X binaries will occur in the future, but we will charge for them on a pay-what-you-want scale.
  • Paid support for Octave and specialised binaries (rpms, debs) for enterprise setting will be produced and promoted.
  • Max’s JIT compiling branch got merged back into Savannah. It will be released with Octave 3.8 as an experimental compile-time feature, disabled by default.
  • Have a public list of release goals, editable only by project admins.
  • Better advertisement for Octave. Who uses it, how, write a semiannual newsletter about new developments in Octave.

Lastly, there was also much informal socialising. Lunch every day at various locations, and on Tuesday we went for beer at L’Amère à Boire and that same night saw Greece’s fireworks in Montreal’s international fireworks competition. The final farewell on Saturday was a Mexican brunch at El Sombrero with a few people who stayed for the weekend.

Much fun was had by all.