On Sat, Feb 21, 2015 at 4:19 PM, MZMcBride <z(a)mzmcbride.com> wrote:
Erik seems to be pushing toward a model that favors
using OAuth and the
MediaWiki API over "deep integration" that comes with a MediaWiki
extension. He recently mentioned this here:
https://lists.wikimedia.org/pipermail/glamtools/2015-February/000343.html
He may be right that development for deployment to the Wikimedia
Foundation cluster may not be the best approach for every project, but I
think this view overlooks all the very real benefits that extension
deployment includes.
I don't think one size fits all -- every case needs to be judged on
its merits, though in the case of GLAMWikiToolset I am definitely
arguing for considering separation from the MediaWiki codebase because
it is so highly specialized. I also think we sometimes still have a
tendency to underestimate the value of non-MediaWiki tools and apps,
even though they've contributed millions of edits to Wikimedia wikis
already (though to be fair, without Magnus Manske the tally would not
be nearly as awesome).
Regarding the criteria for grantmaking, I think this initial blanket
prohibition against all MediaWiki extension development is indeed
something we ought to revisit. These grants can cover tens of
thousands of dollars of paid work, so we shouldn't treat the review
and integration burden lightly, and avoiding stalled projects that are
going nowhere was a reason I advocated for this restriction to begin
with. But as long as there is a good plan in place -- either not
significantly dependent on WMF or with clear commitments negotiated
upfront -- I do agree that the risks can be significantly mitigated.
Damon, Luis and members of their teams will need to weigh in on this,
and will want to think through the implications for their respective
areas, but it's a good conversation to have -- keeping in mind that
Luis is just starting in his new role, so please give him at least a
few days to get up to speed. ;-)
Erik
--
Erik Möller
VP of Product & Strategy, Wikimedia Foundation