Re: [Boost-users] 1.56.0 beta release candidate 2 available for testing

16 Jul 2014

      Please ignore the following part of my previous email:

On Wed, Jul 16, 2014 at 9:14 PM, Klaim - Joël Lamotte 
wrote:
...
Then I hit the same link error I reported previously, from optional code
apparently:
3>typevaluemap.obj : error LNK2019: unresolved external symbol "class
std::basic_ostream & __cdecl
boost::operator<<(class
std::basic_ostream &,class
boost::optional<int> const &)" (??$?6DU?$char_traits@D@std@@H@boost
@@YAAAV?$basic_ostream@DU?$char_traits@D@std@@@std@@AAV12@ABV?$optional@H
@0@@Z) referenced in function "void __cdecl
testing_internal::DefaultPrintNonContainerTo
...
(class boost::optional<int> const &,class std::basic_ostream *)" (??$DefaultPrintNonContainerTo@V?$optional@H
@boost@@@testing_internal@@YAXABV?$optional@H@boost@@PAV?$basic_ostream@DU
?$char_traits@D@std@@@std@@@Z)
I investigated further and found that this error only occur in when this
function have been compiled:
template< class ValueType >
    inline boost::optional<ValueType> TypeValueMap::read() const
    {
        auto* slot = m_slot_index.find<ValueType>();
        if( slot && slot->value )
        {
            return slot->value.get();
        }
        return {};
    }
With slot's type  looking like this:
template< class ValueType >
        struct SlotOf : public Slot
        {
            boost::optional<ValueType> value;
            //...
Maybe I did something wrong here?
However, the link error report suggest that some boost optional's testing
function is called when it shouldn't (this appear when I'm compiling my own
project).
I found that it's a link issue related to GTest macros which apparently
generate this error when I try to test the optional result implicitely:

#include 
#include 

TEST( blah, blah )
{
    boost::optional<int> k = 42;
    EXPECT_EQ( 42, k ); // this will trigger the link error
}

Replacing the pointed line by
   EXPECT_EQ( 42, k.get() ); // OK

seems to fix the issue.

Sorry for the noise!