I would be lying if I said I am not surprised with the lack of
response followed by Allin's response on this matter.
Producing well-written proposals on _several_ potential projects
sounds daunting, but confining ourselves to the data importer
issue that Jack mentioned (perhaps plus one other) seems more
Here is a well-written proposal from last year:
Title: Connecting R and PostgreSQL using DBI
The GSoC project involves creating an R add-on package that provides
the 'glue' between the DBI package and PostgreSQL. This will bridge
the existing gap for a DBI connection between the premier Open Source
data analysis system R and the object-relational database management
> 1)- Importing more statistical distributions to gretl using
> stats-library... I think we can have lots of new statistical
> distributions and their inverse functions if a student imported
> some of the code or implemented the algorithms from R or some
> other resource.
Agreed, this would be nice from our point of view, but I'm afraid
it sounds boring for the student (just re-coding).
I have seen quite a few projects which involve recoding something in C
to gain speed and efficiency. Recoding may not be boring for the
student if he is doing it for the first time. He will be learning a
great deal, plus he gets paid for it. Good opportunity for us to get
something done which would be really boring for a senior programmer.
I see the point regarding some of the projects that I have proposed
being not modular and therefore unfeasible. On the other hand I don't
see what is wrong with my 5th proposal (regarding confidence bands)
that you did not comment on. It is modular, relatively simple,
something small but nice to have etc. I think it is perfectly doable.
Here is another similar project from last year:
Title: Adding 3D plots support for Gnumeric/GOffice
Gnumeric is capable of working on large sets of data and has a variety
of tools for visualising and analysis of 2D data. However, sometimes
two spatial dimensions are not enough. Especially scientists often
need to visualize their data in 3D. My goal will be to make it
possible by implementing the most commonly used visualization method -
a 3D surface plot. It will be implemented as a plug-in for GOffice
library, like all the other plots.
GSoC has some difficulties and extra work but I think it is worth it.
One difficulty is we need Allin, Jack and maybe others to accept
mentoring some of the students on the projects. As I said, I can
mentor one (maybe the confidence band project). GSoC is very nice
because that way you get more people involved, studying and
understanding the code. This is useful and healthy to have in the
Together with the filters library project, here I see 3 doable
projects. Plus I was sure that some of you guys would come up with
some other, nice, modular and suitable proposals. That did'nt happen.
If you are set on doing the one joint project thing with R, there is
nothing more I can say.
“Remember not only to say the right thing in the right place, but far
more difficult still, to leave unsaid the wrong thing at the tempting
moment.” - Benjamin Franklin (1706-1790)