Wouldn't the program need exactly one master type-of list? Since C and C++ don't have cross-module consolidation, just text dumps of common files
"Daryle Walker" <darylew@hotmail.com> wrote that
work by "coincidence", couldn't you break ODR by:
File1.cpp: #include BOOST_TYPEOF_BEGIN_REGISTRATION() BOOST_TYPEOF_REGISTER_TYPE(greater_action); //1 BOOST_TYPEOF_REGISTER_TYPE(less_action); //2 BOOST_TYPEOF_REGISTER_TYPE(and_action); //3 #include BOOST_TYPEOF_END_REGISTRATION() int my_func1() { /*...*/}
File2.cpp: #include BOOST_TYPEOF_BEGIN_REGISTRATION() BOOST_TYPEOF_REGISTER_TYPE(placeholder<1>); //1 BOOST_TYPEOF_REGISTER_TYPE(less_action); //2 (only one the same) BOOST_TYPEOF_REGISTER_TYPE(greater_action); //3 #include BOOST_TYPEOF_END_REGISTRATION() int my_func2() { /*...*/}
I could, but I use anonimous namespaces to avoid this. This ensures every compilation unit has it's own set of encoding/decoding classes, and takes care of ODR.
This wouldn't break if the preprocessor counter doesn't reset between files, but then the same type-of would get lots of Ids.
The same type may be encoded/decoded using different set of integral constants in different compilation units. I don't see a problem here since encoding/decoding always happens inside the same compile-time expression, and therefore inside the same translation unit. Regards, Arkadiy