Re: [eigen] Contributing to Eigen - best place to start?

[ Thread Index | Date Index | More Archives ]

Am 03.03.2015 um 12:40 schrieb Gareth Owen:
Former physicist by training, so I used to do a lot of stuff with matrix
decomposition, and things such as least-squares solvers. Nowadays I write a
lot of template-based vector math routines (hence the interest in Eigen),
but I figure if I want to contribute it might be best if I just go on a bit
of a deeper tour of Eigen than present and see how some of the bits I've
not looked into yet work, and if they aren't documented or have TODO:doc by
them maybe write a few comments - are you interested in doc patches to
start with?

Yes! Documentation, especially of the internal parts, can definitely be improved a lot. What is also a bit annoying at the moment is the amount of warnings Doxygen generates. Some of them are valid and shall be fixed, some are because our inheritance is too complex for Doxygen and should be worked-around (we already have som #ifdef EIGEN_PARSED_BY_DOXYGEN hacks inside the code). If you are into preprocessor-hacking, you may also have a look at Bugs 833 and 402 to reduce the amount of boiler-plate code.

But, as said before, start with what you feel most comfortable with or what you are most interested in. And don't hesitate to ask, if you have trouble understanding something.



On Mon, Mar 2, 2015 at 9:22 PM, Gael Guennebaud <gael.guennebaud@xxxxxxxxx>

On Mon, Mar 2, 2015 at 4:37 PM, Christoph Hertzberg <
chtz@xxxxxxxxxxxxxxxxxxxxxxxx> wrote:

You can start with what is (or may become) most important for you (and
you feel capable of solving).

This is the most important, especially to get started! Since Eigen
involves many very different skills, you could give us a few keywords
defining what you would find fun to work on, and then we could make some
more precise suggestions. Some keyword examples: multi-threading, dense
matrix decompositions, sparse-decomposition, SIMD, numerical accuracy,
performance, tuning for 2D/3D apps, template meta programming, ARM, doc,
cmake, c++11, STL, etc.


Dipl.-Inf., Dipl.-Math. Christoph Hertzberg
Cartesium 0.049
Universität Bremen
Enrique-Schmidt-Straße 5
28359 Bremen

Tel: +49 (421) 218-64252

Mail converted by MHonArc 2.6.19+