[Rd] Google Summer of Code 2009

Friedrich Leisch Friedrich.Leisch at stat.uni-muenchen.de
Fri Feb 20 02:06:46 CET 2009

>>>>> On Thu, 19 Feb 2009 10:52:19 -0600,
>>>>> Dirk Eddelbuettel (DE) wrote:

  > [ Cool how nobody cared about Fritz' request not to post ideas yet :) ]

Well, I kind of expected that ;-)

See also below.
  > [ I broadly share Oleg's "wouldn't it be nice to have better plot devices"
  >   wish.  But I don't think it is a three-month summer target,

Yes, that's exactly what came to my mind first: As usual, please do
read docs before you post ... in this case the format of SOC (I
included the link in my original email, googling for "summer of code"
will also take you there): a student is paid to code three months for
us, the 3 months inlcude writing documentation. The student will not
be an expert in R internals, and no magic wizard. The student should
familiarize himself with the project before the actual coding period,
but there is only so much you can do in limited time. I think you can
expect a similar amount of code as in a master/diploma thesis (but
NOT a dissertation).

If you had waited for Manuels email you would also have learned about
another VERY IMPORTANT POINT: The collection of ideas for summer of
code is not like writing a list of wishes to Santa Claus (or the
Christkind or whatever your local variation may be): we only need
ideas which YOU ARE WILLING TO MENTOR, i.e., you write the specs for
the project, communicate with students interested in the project,
select the best applicant and supervise the student during the coding
period. I am not sure everyone on this thread is aware about this (if
all of you were I apologize). If you propose an idea, you
simultaneously agree to volunteer a considerable amount of your own
time. But that time can really be worth the effort (otherwise we
wouldn't be doing it).

  > and it's not on the side of things Fritz / Manuel prefer as it is
  >   infrastructure rather than pure statistics ... Then again, maybe
  >   we should put that up to a wider discussion.  I like
  >   'infrastructure' as R is a platform to me. ]

I have no "preference" for pure statistics: last year we had 75%
infrastructure ideas and 25% statistics. I simply want to shift the
percentages to a more even ratio, because we had many application on
the statistical side and I don't want to waste talent. It is also our
USP in the summer of code.


More information about the R-devel mailing list