Re: [boost] A possible GSOC 2011 proposal idea

Subject: Re: [boost] A possible GSOC 2011 proposal idea
From: Simonson, Lucanus J (lucanus.j.simonson_at_[hidden])
Date: 2011-03-21 12:58:56

Chad Seibert wrote:
> Right then. Perhaps the best alternative is to write a wrapper for
> common LP and LS solvers, based on the assumption that no combination
> of our efforts will ever out-match other open-source implementations.
> Does that seem a reasonable summer project? Write a wrapper around
> LAPACK, similar to LAPACK++?

A "modern" C++ DSEL for specifying linear constraints would be a very elegant piece of template metaprogramming, but special care would have to be taken for error handling.  If you SFINAE guard every interface such that nothing that shouldn't compile ever matches any function your errors would be "no function F at line #".  Since we have standard file formats for LP and ILP the first goal could be to just emit a standard format for consumption by an off the shelf solver.  Even that modest scope would be very challenging to complete in the GSOC time frame, but we expect a successful student project to continue after the summer and evolve into an active member of the boost community.

Regards,
Luke