Google Summer of Code 2007 Report
From K-3D
Overview
This year, K-3D participated in the Google Summer of Code (GSoC) program for the first time. Two students were assigned to us, one of whom was successful in completing his project.
Projects
Quadrilateral Remeshing
- Student: Ian South-Dickinson
- Mentor: Tim Shead
- Goal: Create a quadrilateral remeshing plugin requiring minimal user input.
Ian's project was succesfully completed and the source code has been merged into K-3D trunk. There are some outstanding cross-platform issues with the output, which Ian has committed to address. See Ian's original proposal on his page: User:southdii.
Fluid Simulation
- Student: Sanjin Sijaric
- Mentor: Bart Janssens
- Goal: Recreate the motion of liquid in a cubic volume, with a preview in OpenGL.
The initial goal of the project was quite ambitious: to recreate a mesh of the surface of a volume of liquid bounded by polygonal K-3D meshes. After the midterm evaluation, it became clear this was probably too complex for a summer project. The goal was revised to simulation in a cubic volume, with a visualization using an OpenGL point cloud. Such a system would form a solid base for a complete simulation engine in K-3D. Unfortunately, the revised goal was not met, mostly due to Sanjin's obligations to his day job in combination with GSoC. Due to the unfinished nature of the code, it has not been committed to the trunk, though Sanjin has promised to continue to work on the project when he has more time.
Lessons Learned
As this was the first GSoC for K-3D, many lessons have been learned, both about the student application process and the nature of the suggested projects. Some of this we figured out on our own, other items came from other GSoC projects that shared their experience.
Student Selection
This year, student applications were evaluated mostly on the quality of the application itself. Additional factors should be taken into account:
- The summer agenda (job, vacations, ...) of the student should be taken into account during student selection.
- A small code sample should be requested as part of the application process. This allows us to measure both student commitment and technical quality. It also ensures a minimal familiarity with K-3D prior to the start of GSoC
- Regular chat and email communication before program start, to ensure student availability. Though this was no problem this year, other projects have reported experiences with dissapearing students or very poor communication.
- Community participation: Extra points for students who are active on the mailing list and wiki before program start!
Project Proposals
- Lower the complexity of the proposed projects. Having highly scientific, fairly large-scale projects may scare away many students
- Propose projects that consist of multiple plugins. This year's projects "black-or-white": it either worked or didn't work, and if it didn't we had nothing. Having more modular projects makes it easier to follow progress, and increases the likelyhood of getting useful code out of GSoC
- Suggest popular topics. Nothing motivates students like seeing that their code gets used. Having only niche topics for the projects discourages this
Mentor Summit
Conclusion
While only one of the projects was successful, GSoC 2007 has been a great experience for K-3D. It has allowed us to learn a lot from the other participating organizations. If we are chosen to participate again next year, our experience will certainly help a lot to achieve better results.
