Le Thu, 08 Apr 2010 19:03:20 +0200, Jorge G. Mare <koki@xxxxxxxxxxxxx> a écrit:
Looking at how GSoC has played out in the past, I highly doubt that encouraging students to work on what they feel most motivated works well for Haiku. On the one hand, the resulting code is not likely to help the project move forward along its development roadmap (in many cases, it just ends up gathering dust); on the other, and perhaps more importantly, you have to consider the fact that mentoring sucks up developer attention and time, so you are in fact also diverting resources from the R1 goal. A much better strategy would be to try to match the priority of the GSoC projects to that of the various milestones in the development roadmap. In other words, list up what Haiku needs first as top priority ideas, and then "the nice to have but not so important for now" projects after that. You can then tell students they are more likely to be accepted if they choose a high priority project, so that they have an incentive to choose a project that will actually help Haiku move along its development roadmap towards a release. Not that one should totally rule out original proposals from students; there is always a chance that a student could come up with a really good idea. But at least the project should make it clear what it's priorities are, so that the students also know what areas are more important to work on.
No, no, no.Due to the way GSoC works, when a student picks a task, no one can help working on it for the whole summer (besides the mentor, to some extent). This means having a student working on a task locks any other dev from contributing to it. If the code the student writes gets lost, then it's 3 month waiting for nothing. That's why the proposals this year are more disconnected from R1. Also, I don't think the project being forgotten in some obscure branch are the most numerous : see the locale kit and the webkit port as examples. The locale kit proposal at least was my own idea and not part of the list. I think it worked pretty well, and I don't think you see it as a bad thing ?
-- Adrien Destugues / PulkoMandy http://pulkomandy.ath.cx
