The Google Summer of Code (SoC) is now an annual event, in which Google pays computer science and software engineering students to work full-time in open source projects over their summer vocation on predefined tasks.
The students get hands-on experience with large-scale software development in the real world and the projects get concentrated work from developers-in-training.
The list of open source projects submitting tasks (confusingly called “projects”) to the 2007 SoC is significantly larger than last year and very diverse. Everything from operating systems (Ubuntu and FreeBSD) and office suites (OpenOffice.org and AbiSource) to media tools (Blender and Audacious) and content management systems (Drupal and Joomla!). Google is not playing favourites either, funding projects with histories of vigorous competition, such as KDE and GNOME, and PostgreSQL project and MySQL.
From an educational point of view Moodle stands out immediately. There are a number of student projects such as “Developing new question types for the quiz” and “Automated grading for Computer Programming Assignments” which seem to be very useful and have wide applicability.
Some of the more geeky tasks make me wish I had the summer off to spend on something like this. GCC is has some interesting C++ tasks and JikesRVM has very attractive looking Java-in-Java and garbage collection tasks.
I suspect that the SoC requires a different sort of organisation from open source projects than they’re used to, but I can also imagine that there are many small open source projects which could usefully be laying the groundwork for SoC 2008: identifying discrete independent tasks, finding individuals willing to mentor young developers, and thinking about the requirements from Google.
It’s worth pointing out that this year OSS Watch has one of their own offering to mentor students. I have in fact been involved with the GSoC programme every year since its inception as a mentor and as The Apache Software Foundations internal administrator for the programme.
There are a few things that Stuart mentions that are worth highlighting:
They are not tasks. They are projects in their own right. That is, the students themselves are expected to take on this sub-project within the mentoring project and to manage the whole thing from beginning to end, they are in control, not the mentor. A GSoC project allows for a great deal of freedom on the part of the student, far more than a “task” typically would. This is open source remember, there are no task masters and there is a whole community of resources to help, contribute and advise on the GSoC project.
This is one of the fundamental problems with GSoC. Students usually come to the project with the intention of earning the money. However, the real goal should be to learn about open source development. It can be difficult for projects to manage a GSoC student in the same way they would manage a traditional community member. The best projects will insist on the GSoC student going through the normal community processes during thier time on GSoC. The cash is just a nice little bonus.
For me, and for most of the projects involved, the goal is for the GSoC student to continue to remain as a community member long after the money has been paid up.
I heartily endorse this. Although as discussed about “identifying discrete independent tasks” is not really what it is about. You should be thinking about what sub-projects can you allow a GSoC student to lead on.
I take your point that “tasks” is perhaps not an ideal term because of the implied fine granuality.
Personally, I find using “project” confusing, since it is already widely used to refer the mentoring organizations.