Re: [eigen] Google Summer of Code

[ Thread Index | Date Index | More lists.tuxfamily.org/eigen Archives ]


2010/3/6 Stephen McCracken <stevemcc@xxxxxxxxxxx>:
>
> On Mar 6, 2010, at 12:08 PM, Benoit Jacob wrote:
>
> That's exactly the problem --- what we're really looking for is people
> who are likely to turn into long term contributors, to take care of
> boring stuff in the project, etc.
>
> How many of your core contributors use Eigen in the course of their paid
> employment?

Most of them, except me.

>  In my case, I'm unlikely to become a core contributor unless I
> can get a job at a FOSS-friendly company (Willow Garage, say) where I use
> Eigen regularly in my work.  If I did a GSOC project with Eigen, it would
> significantly improve my chances of getting hired at WG and ultimately
> becoming an Eigen contributor.

It's great when Eigen contributions help people find jobs. It just
happened to me. But look at it from our perspective: if you've
followed this list and the IRC channel for a while, you've seen that
we (few core devs) spend a lot of time already informally mentoring
new contributors. It doesn't sounds very interesting to us to enter
the GSOC to do yet more mentoring with an additional paperwork load,
when we already have a lot to traffic to handle. Eigen is really
special in this respect: it's very small and it naturally attracts a
lot of people.

> Now, it happens that WG also uses Python.  So if my primary goal is to go
> work for WG, then I might be just as happy doing a SciPy GSOC project.  In
> that case, I'd be more likely to get hired on a Python-heavy project, and
> perhaps become a Python core contributor.

Sounds like a SciPy GSOC would be a nice match for you :)

> So I think that mentoring a GSOC project does increase the probability that
> your student will become a core developer, even if the student is initially
> "agnostic".  Whether or not that uncertain benefit is worth the investment
> of your time, I can't say.

There are lots of reasons why this makes more sense for SciPy/NumPy
than it does for us. They're a far bigger project, they have an
academic mindset, and they probably can offer a lot more accessible
yet useful problems to solve for newcomers (one area where we're not
very good - the learning curve in eigen development is very steep).

Benoit



Mail converted by MHonArc 2.6.19+ http://listengine.tuxfamily.org/