\nBoth C++ and C# have the concept of virtual functions.\nThese are functions which always invoke the most heavily\nderived implementation, even if called from a pointer to\nthe base class.\nHowever, the two languages differ on the semantics of\nvirtual functions during object construction and destruction.\n<\/p>\n
\nC# objects exist as their final type before construction begins,\nwhereas C++ objects change type during the construction process.\n<\/p>\n
\nHere’s an example:\n<\/p>\n
\nclass Base {\npublic:\n Base() { f(); }\n virtual void f() { cout << 1; }\n void g() { f(); }\n};\nclass Derived : public Base {\npublic:\n Derived() { f(); }\n virtual void f() { cout << 2; }\n};\n<\/pre>\n\nWhen a Derived<\/code> object is constructed,\nthe object starts as a Base<\/code>, then the\nBase::Base<\/code> constructor is executed.\nSince the object is still a Base<\/code>,\nthe call to f()<\/code> invokes Base::f<\/code>\nand not Derived::f<\/code>.\nAfter the Base::Base<\/code> constructor completes,\nthe object then becomes a Derived<\/code> and\nthe Derived::Derived<\/code> constructor is run.\nThis time, the call to f()<\/code> invokes\nDerived::f<\/code>.\n<\/p>\n\nIn other words, constructing a Derived<\/code>\nobject prints “12”.\n<\/p>\n\nSimilar remarks apply to the destructor.\nThe object is destructed in pieces, and a call to\na virtual function invokes the function corresponding\nto the stage of destruction currently in progress.\n<\/p>\n
\nThis is why some coding guidelines recommend against\ncalling virtual functions from a constructor or destructor<\/strong>.\nDepending on what stage of construction\/destruction is taking place,\nthe same call to f()<\/code> can have different effects.\nFor example, the function Base::g()<\/code> above\nwill call Base::f<\/code>\nif called from the Base::Base<\/code>\nconstructor or destructor,\nbut will call Derived::f<\/code>\nif called after the object has been constructed and before\nit is destructed.\n<\/p>\n\nOn the other hand, if this sample were written\n(with suitable syntactic changes) in C#,\nthe output would be “22”\nbecause a C# object is created as its final type.\nBoth calls to f()<\/code> invoke Derived::f<\/code>,\nsince the object is always a Derived<\/code>.\nNotice that means a method can be invoked on an object\nbefore its constructor has run<\/strong>.\nSomething to bear in mind.\n<\/p>\n\nSometimes your C++ program may crash with the error\n“R6025 – pure virtual function call”.\nThis message comes from a function called __purecall.\nWhat does it mean?\n<\/p>\n
\nC++ and C# both have the concept of a “pure virtual function”\n(which C# calls “abstract”).\nThis is a method which is declared by the base class,\nbut for which no implementation is provided.\nIn C++ the syntax for this is “=0”:\n<\/p>\n
\nclass Base {\npublic:\n Base() { f(); }\n virtual void f() = 0<\/font>;\n};\n<\/pre>\n\nIf you attempt to create a Derived<\/code> object,\nthe base class will attempt to call Base::f<\/code>,\nwhich does not exist since it is a pure virtual function.\nWhen this happens, the “pure virtual function call” error\nis raised and the program is terminated.\n<\/p>\n\nOf course, the mistake is rarely as obvious as this.\nTypically, the call to the pure virtual function occurs\ndeep inside the call stack of the constructor.\n<\/p>\n