Hi all, I've watched the throw some_error<e1,e2,>(...) discussion:
Could you please explain? How does using Boost Exception require more boilerplate code than the traditional approach?
I mean the throw site:
throw an_error(a,b,c)
vs.
throw an_error() << boost::error_info<a_tag>(a) << boost::error_info<b_tag>(b) << boost::error_info<c_tag>(c)
Why not have a class like this: // t1, t2, t3 - tag classes template<class t1, class t2, class t3...> struct tags { // not sure if "underlying_type" is the actual underlying type but you catch my drift typedef typename t1::underlying_type p1; typedef typename t2::underlying_type p2; // hold the arguments tags(p1 a1, p2 a2, ...) : m_p1(p1), m_p2(p2) ... {} p1 m_a1; p2 m_a2; .... }; template<class t1, class t2,...> boost::exception & operator<<(boost::exception & e, const tags<...> & t) { e << t1(t.m_a1) << t2(t.m_a2) << ...; return e; } This way, you can have your cake and eat it too: typedef tags<tag_file_name,tag_open_mode,tag_errno> file_err; throw my_error() << file_err(name,mode,errno); Best, John -- http://John.Torjo.com -- C++ expert ... call me only if you want things done right