On 08/07/2010 10:42, Tim Vandermeersch wrote: > On Thu, Jul 8, 2010 at 11:41 AM, Tim Vandermeersch > <[email protected]> wrote: >> On Tue, Jul 6, 2010 at 10:12 AM, Noel O'Boyle<[email protected]> wrote: >>> I hope you can get this to work. I tried using boost::erf and got >>> weird boost errors - presumably something I was doing wrong. >> >> I got it working here but I had to comment out everything in >> boost/math/constants/constants.hpp since there were errors in the file >> (are these your weird boost errors?). It still compiles since we don't >> use any of the constants though. > > Btw, this is boost 1.43.0 > >>> On 5 July 2010 23:06, Tim Vandermeersch<[email protected]> wrote: >>>> Hi, >>>> >>>> While checking if the spectrophore code builds on windows I noticed >>>> the charge methods (qeq and qtie) require the error function erf. This >>>> is not standard and MSVC doesn't have it but boost does. I'm changing >>>> the build system to find boost and use it when available. Otherwise, >>>> the charge models are disabled (similar to when Eigen2 is not found).
Would it worthwhile to use a local, MSVC-only, implementation of erf? This would remove the dependence on Boost (which is not used otherwise on our supported windows compiler: VC9 express and later), avoid having to edit its code, and simplify the CMake files. There seem to be several implementations that would be adequate, since I don't think the requirement in the qeq and qtpie methods are very stringent. I have one (apparently good to better than 1 in 10^11 for the range I tested over) ready to upload (in a .h file), if this is considered appropriate. Chris ------------------------------------------------------------------------------ This SF.net email is sponsored by Sprint What will you do first with EVO, the first 4G phone? Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first _______________________________________________ OpenBabel-Devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/openbabel-devel
