On Thu, Mar 11, 2010 at 2:11 PM, Rob Lanphier robla@robla.net wrote:
On Thu, Mar 11, 2010 at 10:31 AM, Chad innocentkiller@gmail.com wrote:
I'm with Aryeh on this one. I can't really commit to full-time being able to mentor someone--I have work, school, et cetera--but I'm around often enough to be able to lend a hand or two if needed. Plus if we get our students on IRC, they'll have the benefit of real-time feedback from many developers, not just their mentor.
Absolutely! I think it's going to be exceptionally important to encourage students to be active on IRC and generally visible in the community.
And this is where we've seen a difference. Some people like Jeroen have really integrated themselves, hanging around, asking questions, becoming a part of the community. I'd like to see our developer community grow as a result of GSoC, not just have people show up for their few months and then disappear when it's over.
I'm also with Trevor in saying that I'd only like to work on projects that
have some tangible benefit to a larger group of people (ie: deployed on WMF sites, or a major new feature for MediaWiki users in general). GSoC has been really hit or miss with our community over the past few years. Whether it's lack of resources, or burnout, or who knows, but the ROI of developer time has been smaller than I think we'd like to see. We've had some great students in the past who've done some stellar work, and we've also had projects that went nowhere and ended up bitrotting somewhere. I think we'd all like to avoid the latter.
I put together a page of past projects here: http://www.mediawiki.org/wiki/Summer_of_Code_Past_Projects
http://www.mediawiki.org/wiki/Summer_of_Code_Past_ProjectsCould people that know what the true latest status of all of these projects go in and fill out what they know? That will help us figure out what the characteristics of a project that's likely to be successful will be.
I updated some of the ones I know about.
I suspect that projects that involve integrating large blobs of non-PHP code into MediaWiki are the most troublesome. They sound cool on the surface, but probably end up being a little bit like strapping a jetpack onto a cat ("what could possibly go wrong?"). Self-contained extensions without outside dependencies seem the most likely to have long-term benefits, as well as very carefully scoped tweaks to the core (e.g. improvements to lesser used special pages).
Rob _______________________________________________ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
Extensions are easy to write and review, sure. And I'm certainly fine with normal core code changes too, as long as they follow coding conventions and pass code review. But larger scope changes (typically anything involving the phrase 'rewrite X') are much harder to review, and I think should be avoided. We want projects that have visible results and trying to do large-scale changes sets people up for failure.
-Chad