Re: [Boost-users] [Boost][Serialization] Possible Bug inOptional.hpp

From: Marc Viala (mviala_at_[hidden])
Date: 2007-08-22 14:03:59

Thanks for your quick answer Robert. 
 
Sure it doesn't affect the program behaviour but the debug with Visual
C++ IDE is little bit more painful ;-). I'm going to fix my own copy of
the file.

Marc Viala 
De la part de Robert Ramey

 
clearly its a bug but it shouldn't affect any programs, I'll fix it in
my working copy.  You can fix it by just initializing to 0.
 
Robert Ramey
 
        "Marc Viala" <mviala_at_[hidden]> wrote in message
news:D0D608FD34E14C4BA93A30AC162651E18BA519_at_SERVEUR2003.acticm.com...
        I've downloaded recently the version Boost 1.34.1 and performed
some tests on ours applications integrating Boost components. 
         
        During theses tests, I've tried to deserialize some XML archives
generated w/ Boost 1.33.1 including some boost::optional<> instances and
I've observed that the Visual C++ 7.1 IDE emits run-time checks in the
load<boost::optional<>>()  @ line 60 in file
boost/serialization/optional.hpp.
         
        template<class Archive, class T>
        void load(
            Archive & ar, 
            boost::optional<T> & t, 
            const unsigned int /*version*/
        ){
            bool tflag;
            ar >> boost::serialization::make_nvp("initialized", tflag);
            if (tflag){
                unsigned int v;
                if(3 < ar.get_library_version()){
                    ar >> make_nvp("item_version", v);
                }
                detail::stack_construct<Archive, T> aux(ar, v);

                       ^^^^^^^^^^^^^^^^^^^^^^^^
        Here Run-Time Check Failure #3 - The variable 'v' is being used
without being defined.
         
                ar >> boost::serialization::make_nvp("value",
aux.reference());
                t.reset(aux.reference());
            }
            else {
                t.reset();
            }
        }
         
        Is it possible to avoid this run-time check by a
preinitialisation of the local variable "v"? Is it a Bug? 
         
        Best regards,
         
        Used Configurations:
        Windows XP
              Boost 1.34.1 (through Boost Installer w/ DLL libraries)
         
        Marc Viala