Google has announced their Summer of Code, 2008 edition. The mentor organization application process begins on Monday, March 3 and I will be right there, ready with our group’s app. My FFmpeg cohorts and myself have been busily assembling a new Wiki page detailing what prospective students might do for the project, should FFmpeg be accepted as a mentoring organization for a third year. Like last year, we will be enforcing the requirement that students must successfully complete a qualification task in order to be considered for a project. I feel like I live in sort of a bubble these days, but I am becoming increasingly aware that straight-up, performance-minded C programming seems to be a dying art and we can’t necessarily count of students knowing the language already when they apply for the project.
Of course, you are free to add to the list, either in the 2nd tier proposals section or the qualification tasks. But don’t bother adding anything under the 1st tier proposals unless you are willing to mentor the project. Anyone can toss out any idea. But we need project ideas that can be plausibly completed by a talented student over the course of a summer, and we need mentors who will commit. As for qualification tasks, remember that these are bite-sized pieces of work that would ideally take a seasoned FFmpeg developer a few hours at most to complete. If you create a new qualification task, please put some detail into it. Look at it from the perspective of a new student who may not be up to speed on FFmpeg. A qualification task of “fix XYZ thing” is quite bewildering. Expound just a little bit. Remember to link to other pages within the Wiki.
I logged into my SoC mentor dashboard for the first time in a long time. It claimed that I had not completed my final program survey from last year, even though I’m quite certain that I did (I wonder if that’s why I never saw my mentor money?). Anyway, one of the questions:
What advice would you give to future would-be Summer of Code mentoring organizations? (required)
I don’t remember my answer last time, but this is the most honest answer that came to mind this time: “Don’t try to compete with us for prize students. FFmpeg is a sexier project — an alpha project, if you will — and you won’t beat us.”
From the student point of view, I agree heartily that the qualification tasks should be geared toward those who have possibly not had much experience with FFmpeg in particular.
My experience last year attempting to finish one of the qualification tasks (a decoder for an obscure game video format, BFI) left me pretty frustrated with the lack of documentation and guidance for developers new to FFmpeg. I consider myself to be a pretty decent C hacker and not too dim when it comes to learning new things, but getting my mind wrapped around FFmpeg in the (short) qualification period was a really large hurdle. One thing that caused a lot of consternation was jargon (abbreviated names for variables and functions, mainly) in the documentation and comments that was never defined. I did figure out enough to get basic audio decoding working, but I got stumped trying to figure out what exactly I needed to provide to get any kind of video decoding working (the fact that the video format used palettes did not help either – it seemed like there was not much infrastructure or documentation for dealing with palette changes). I eventually decided FFmpeg was not for me that year and moved on to look at different projects.
This is not to say that FFmpeg is doing something outright wrong here – perhaps I was just not prepared for the sort of work I’d be doing as a project eventually anyway; in that case, it was good that the qualification task scared me off. I’m not convinced that this was the case, but it’s certainly possible.
On the other hand, I’ll certainly give it another try this year – FFmpeg is indeed a very “sexy” project, and it’s one of the few projects that seemed to me to have that “straight-up, performance-minded C programming” mindset that you speak of, which is what I was looking for.
If you would like to try your hand this year, great! Just remember to get on the mailing list sooner rather than later and indicate your intention (this assumes that FFmpeg is accepted; I just submitted our group’s app; final notices should be in 2 weeks). The BFI playback task is still outstanding, so try again. You can email me privately, as well, if you want advice or help.
I am available for questions as well for anyone who needs some hints for a qualification task, though in general I think the mailing list is the preferred way. Maybe we can/should make a list of people who are ready to answer private mails on case someone has a question they do not want to ask on the list? Not entirely sure it is a good idea though…