Namespace eo::mpi: For serialization purposes, don't depend upon boost. It would be easy to use only eoserial and send strings via mpi.
Member eo::mpi::AssignmentAlgorithm::reinit (int runs)=0: Not really clean. Find a better way to do it.
Class eoCMABreed< FitT >: handle bounds
Class eoCMAInit< FitT >: handle bounds
Member eoEsChromInit< EOT >::create_self_adapt (eoEsStdev< FitT > &result): Should we scale sigmas to the corresponding object variable range?
Class eoFileSnapshot: The counter is handled internally, but this should be changed so that you can pass e.g. an evalcounter (minor)
Member eoFlOrKAtomBinOp< EOT >::operator() (EOT &_eo1, const EOT &_eo2): check that we don't do twice the same
Member eoFlOrKAtomQuadOp< EOT >::operator() (EOT &_eo1, const EOT &_eo2): check that we don't do twice the same
Member eoGenContinue< EOT >::totalGenerations (unsigned long _tg): replace this by an "init" method
Member eoNPtsBitXover< Chrom >::num_points: Document this data member
Member eoOStreamMonitor::operator() (void): old verbose formatting, do we still need it?
Member eoParser::getValue (eoParam &_param) const: check environment, just long names
Member eoParser::readFrom (std::istream &is): it should be the next string
Member eoSteadyFitContinue< EOT >::totalGenerations (unsigned long _mg, unsigned long _sg): replace thi by an init method ?
Class eoValueParam< ValueType >: This should be changed to std::stringstream when that class is available in g++.
Member eoVlUniformQuadOp< EOT >::operator() (EOT &_eo1, EOT &_eo2): FIXME bad hardcoded limit, should use an algorithm that guarantee a correct size in a finite number of tries
Member gp_parse_tree::parse_tree< T >::subtree::node_allocator: use the std::allocator interface