David Abrahams wrote:
I suppose I could also wrap the return type in a "wrapper type":
template <class T> struct wrap_rvalue { wrap_rvalue(T const& x) : x(x) {} T x; };
template <class T> wrap_rvalue<T> rval(T const& x) { return wrap_rvalue<T>(x); }
namespace boost { namespace serialization { template <class T> struct is_wrapper<wrap_rvalue<T> > : mpl::true_ {}; }}
ar << rval( f() );
That seems a worthy idea to me. The net effect would be to turn off tracking for this particular object. So, I would I would like to see this have a more widely understood name like dont_track(..) and added to the "case studies" section of the documentation. I think that the idea of temporarily turning off tracking (or other serialization traits) would have a usage beyond rvalues.
If that isn't convenient, and one had nothing else to do, he could explore the implications of implementing your proposed specialization so that it traps at compile time if the object might be tracked. This would mean that something like int f() would work while something like my_object f() would likely trap unless my_object was specifically marked as no_tracking. Thinking about this just a tiny bit makes this seem attractive to me. Of course that doesn't mean much until one sets out to actually do it.
That's actually pretty easy to do;
LOL - well for you maybe. Note one thing. Its not so much the code change, its all the other stuff that this generates: a) code change => b) documentation change/ehancement. This takes some effort to explain why serialzation is trapped or not trapped as a "side effect" of setting the tracking attribute. c) user support - Even if one takes the requisite effort to explain it, one will have to continually re-explain it as users have what they consider surprising behavior c) new tests - not a huge deal but still a minor pain - UNLESS something doesn't work with this or that obscure compiler. I'll consider it in more detail when I have the time. Robert Ramey P.S. I quick peek at the code below make me think that the line (serialization::tracking_level<T>::value == track_never), might better be replaced with: (serialization::tracking_level<T>::value <= track_never), RR
Index: interface_oarchive.hpp =================================================================== --- interface_oarchive.hpp (revision 3212) +++ interface_oarchive.hpp (working copy) @@ -28,6 +28,7 @@ class shared_ptr; namespace serialization { class extended_type_info; + template <class T> struct tracking_level; } // namespace serialization namespace archive { namespace detail { @@ -64,6 +65,14 @@ this->This()->save_override(t, 0); return * this->This(); } + + template<class T> + typename enable_if_c< + (serialization::tracking_level<T>::value == track_never), Archive & + >::type operator<<(T const& t){ + this->This()->save_override(t, 0); + return * this->This(); + }
// the & operator template<class T>
-- Dave Abrahams BoostPro Computing http://www.boostpro.com
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost