> 1) I don't like the non-intrusive way of specifying the > version/save/load operation. The rationale for having the > serialization<> template in the global namespace is quite weak. The > alternative of using free functions is much better IMO. The library > should use unqualified calls to those free functions and > Koenig look-up > would do the rest. This solution has also the benefit that > specialization is replaced by overloading, a feature that is > more easily > supported by compilers.
I happend to disagree with here. serialization<> is in a sence a traits class (that should probably reflected in the name - I will remark it in upcomming review) and I believe it is properly designed. Gennadiy. _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost