Re: Google Code-In kick start
Olly Betts <
olly@...>
2011-10-28 08:02:31 GMT
On Thu, Oct 27, 2011 at 10:58:12PM +0200, Stefano Zacchiroli wrote:
> I've seen no replies to this first mail on list and worries me a bit
> (OTOH, I'm not on the IRC channel, so I might have missed something; my
> apologies if this is the case).
>
> Just to be sure we're on the same page: we all love you volunteered and
> we're sort of expecting that you people take the lead in organizing
> Debian's participation in Google Code-In. This includes discovering what
> is we have to do and when, in order not to miss the program deadlines
> (the first of which is very close: next Monday!).
I went to a GCI-related session at the mentor summit this last weekend,
so I have some idea what should be going on at this point.
AIUI, the really urgent thing is to put together a list of tasks. Note
that the requirements here are significantly different to last year.
Check the rules for the details, but there are something like 8
categories and you need a minimum number of tasks in each. I don't
think this will be much of a struggle for Debian, except that there's
so little time until the deadline.
Also, this year you'll need to "release" at least N tasks at the start
and then at least another N in (IIRC) early December. You can't create
new tasks at any other point - apparently there were issues with some
orgs creating tasks specifically for particular students, so they're
trying to address that.
Also, you can't change the difficulty of a task once it is released
(currently at least - Carol said she'd be happy to allow tasks to be
made less difficult to handle the case where a student does part of a
task, so if the melange team can implement that in time that may be
possible).
One thing I picked up which might be useful - it's fine for a task to
be fairly generic, so you can create 100 "fix a difficult bug in the
Debian BTS" tasks and another 100 "fix an easy bug in the BTS" bugs,
and similarly for "moderate". Then if the student picks a bug you can
tell them if it is easy/moderate/hard and so which sort of task to
claim. You can obviously do this for a number of sorts of task, not
just bugs.
I felt last year that GCI didn't really work as well for Debian as
GSoC did - the biggest problem seemed to be that students can only have
one task claimed at a time, so they're eager to get completed tasks
reviewed so they can move on to another task.
Ana told me we picked up a new active contributor to Debian from GCI
last year. Also, I noticed one of the students I mentored took part
in GSoC this year (not with Debian, but that's still a benefit for
Free Software as a whole), so taking part seems worthwhile, but it
would be good to make it clear what review times students can expect,
and also highlight this issue to potential mentors.
One benefit of the "generic tasks" I described above for bugs is that
students can start work on one while waiting for a completed task in the
GCI system to get reviewed, since the BTS provides a natural mechanism
for indicating you are working on a bug. You just need to watch out
for students trying to work on a dozen things in parallel really.
Incidentally, I'd suggest steering clear of "make a new package" tasks
- creating a new package which is likely to become instantly orphaned
isn't helpful for Debian, so the only benefits of such tasks are that
they get students involved and familiar with our tools. But I think
there are more useful things they could work on - e.g. updating a
package to address lintian errors and warnings, or for a transition.
> I'm sure many people on this list, including myself, will be glad to
> help with subsequent steps. But we really need someone to take the lead,
> and _that_ was the purpose of my call for help. Without that, Debian's
> participation in Google Code-In simply won't happen.
I'm happy to try to help out, though I am on semi-VAC currently, and
in particular I'm going to be very busy this weekend, so I probably
can't help get the initial task list built.
> PS of course, anyone else willing to volunteer and/or take lead is still
> welcome!
I'll be away for 2 weeks over New Year, which is why I didn't offer to
act as an admin - not sure exactly how it aligns with the GCI dates, but
last year it meant I got back less than a day before the deadline.
So given that and my unavailability between now and the application
deadline, I don't think I can sanely offer to take the lead. If the
current volunteers think it is useful, I'm happy to act as a co-admin or
in an advisory capacity (I've extensive experience with GSoC, and have
some GCI experience from last year as well as what I picked up at the
mentor summit).
Cheers,
Olly
RSS Feed