C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  static virtual functions Goto page 1, 2  Next          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message kiran Guest Posted: Mon Dec 18, 2006 9:01 am    Post subject: static virtual functions Why cannot a static member function of class be declared as virtual? thanks, Kiran -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Guest Posted: Mon Dec 18, 2006 10:52 pm    Post subject: Re: static virtual functions Static member functions can be called without an object (no this pointer), but virtual functions dont make sense without an object. /Patrik kiran wrote: Quote: Why cannot a static member function of class be declared as virtual? thanks, Kiran -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Ondra Holub Guest Posted: Mon Dec 18, 2006 11:05 pm    Post subject: Re: static virtual functions kiran napsal: Quote: Why cannot a static member function of class be declared as virtual? thanks, Kiran Because it may not be called as virtual function. You can consider static class member function as global function with changed visibility (it is simillar to namespace, but it may be private or protected too). You do not need instance to call static member function, so there is no virtual table available and virtual call mechanism may not be used. -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Nicola Musatti Guest Posted: Mon Dec 18, 2006 11:05 pm    Post subject: Re: static virtual functions kiran wrote: Quote: Why cannot a static member function of class be declared as virtual? Because in order to dispatch a virtual member function you need an instance of the class, whose dynamic type determines the choice of function to call. No such instance is available for a static member function. On the other hand you can always define a non static, virtual member function and not refer to any non static member in its implementation. Cheers, Nicola Musatti -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Ron Natalie Guest Posted: Mon Dec 18, 2006 11:06 pm    Post subject: Re: static virtual functions kiran wrote: Quote: Why cannot a static member function of class be declared as virtual? What would it mean if it were? Static members aren't associated with any particular object, there's no object type (static or dyanmic) to determine the overrider. -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Arun Guest Posted: Mon Dec 18, 2006 11:11 pm    Post subject: Re: static virtual functions kiran wrote: Quote: Why cannot a static member function of class be declared as virtual? Function resolution (mapping a function call to its definition) for a virtual function is dynamic(Happens at run time with the help of vptr and vtable). Where as static function resolution happens at compile time. At compile time itself all call references to a static function will be resolved by the compiler. This is the reason why we can not have (even conceptually) a static member function, virtual. Note that the static member function is nothing but a static function with a specified class scope. We do not need objects to call a static function, where as a virtual function works only with objects. -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Dino Guest Posted: Mon Dec 18, 2006 11:12 pm    Post subject: Re: static virtual functions Then it wouldn't be static. You could have a templated static. Cheers, D. kiran wrote: Quote: Why cannot a static member function of class be declared as virtual? thanks, Kiran -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Guest Posted: Tue Dec 19, 2006 1:44 am    Post subject: Re: static virtual functions Bob a icrit : Quote: kiran wrote: Why cannot a static member function of class be declared as virtual? This question is probably a candidate for the C++ FAQ, if it's not already thre. A virtual member function is invoked in context of an object. For example; object->virtual_function(); where the actual function called depends on the actual type of object. A static member function is invoked without regard to any object, so the actual function called cannot depend on the type of an object. So the two concepts (virtual and static member functions) are mututally exclusive. Well, Object Pascal has had the moral equivalent of virtual static members for more than a decade now. They come in very handy to implement factories or other kind of metaclass information. I am not claiming that C++ needs them. I am claiming that the concept makes a lot of sense and is not at all inconsistent nor even useless ... --- Raoul Ref: See Class methods at : http://info.borland.com/techpubs/delphi/delphi5/oplg/classes.html -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top John G Harris Guest Posted: Tue Dec 19, 2006 3:25 am    Post subject: Re: static virtual functions In message <memo.20061218121050.248B (AT) brangdon (DOT) cix.compulink.co.uk>, Dave Harris <brangdon (AT) cix (DOT) co.uk> writes Quote: kiran.tangeeda (AT) gmail (DOT) com (kiran) wrote (abridged): Why cannot a static member function of class be declared as virtual? Because classes are not reified as run-time objects, which means static member functions don't have a "this" or a vtable to keep track of the dynamic type of the class. Which in turn is done for variety and efficiency. It didn't have to be that way, and in other languages (eg Smalltalk) the equivalent of static member functions are virtual, and jolly useful they are too. Smalltalk comes with a full meta-class system. If you want that in C++ you can build the meta-classes yourself. snip Why stop at classes ? If you could have a reference to complex<double> then the next thing to ask for is a reference to complex :-) John -- John Harris [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Bob Guest Posted: Tue Dec 19, 2006 10:10 am    Post subject: Re: static virtual functions hasta_l3 (AT) hotmail (DOT) com wrote: Quote: Well, Object Pascal has had the moral equivalent of virtual static members for more than a decade now. They come in very handy to implement factories or other kind of metaclass information. I am not claiming that C++ needs them. I am claiming that the concept makes a lot of sense and is not at all inconsistent nor even useless ... Class methods are the Object Pascal equivalent of C++ static member functions. -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Nicola Musatti Guest Posted: Wed Dec 20, 2006 12:04 am    Post subject: Re: static virtual functions hasta_l3 (AT) hotmail (DOT) com wrote: Quote: Bob a écrit : hasta_l3 (AT) hotmail (DOT) com wrote: Well, Object Pascal has had the moral equivalent of virtual static members for more than a decade now. They come in very handy to implement factories or other kind of metaclass information. [...] Class methods are the Object Pascal equivalent of C++ static member functions. and they can be virtual... The implicit this (aka Self) argument is - sort of - a reference to a metaclass. That's precisely the point: in order to implement static virtual member functions you need an explicit representation of classes as objects, which C++ does not provide. Cheers, Nicola Musatti -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top James Kanze Guest Posted: Wed Dec 20, 2006 12:20 am    Post subject: Re: static virtual functions Dave Harris wrote: Quote: kiran.tangeeda (AT) gmail (DOT) com (kiran) wrote (abridged): Why cannot a static member function of class be declared as virtual? Because classes are not reified as run-time objects, which means static member functions don't have a "this" or a vtable to keep track of the dynamic type of the class. That's not quite true. What does typeid( MyClass ) give you, if it isn't a concrete run-time object associated with the class. Quote: Which in turn is done for variety and efficiency. It didn't have to be that way, and in other languages (eg Smalltalk) the equivalent of static member functions are virtual, and jolly useful they are too. Smalltalk comes with a full meta-class system. If you want that in C++ you can build the meta-classes yourself. It would probably be better to extend type_info. Quote: In other words, for every class MyClass, you add a (meta) class MyClass_Class, and create an instance of the (meta) class as static member of the class called class_. Where class A inherits from class B, (meta) class A_Class inherits from B_Class. Instead of A::function() you have A::class_.function(). That calls a virtual function of A_Class on the static member of A, and it all works normally. (I put "meta" in brackets here because C++ doesn't have real meta-classes, but that is what you are modelling.) I'm not sure what this buys us compared to something like the Java approach, where there is just one type Class (a very extended version of C++'s type_info). Quote: The fact that you can build a structure like this by hand when you need it, is an argument for the language not forcing it upon you when you don't need it. Building the structure isn't the problem; using it is. With regards to static virtual functions, of course, you don't need it. A static function can be called in two ways: MyClass::staticFunc() ; ptrObj->staticFunc() ; The first way, you get the class member---no virtual resolution (and you wouldn't get virtual resolution even if the function were non-static), and in the second, you have an object, so there's no problem using virtual resolution, even if you don't pass the this pointer to the function. The question is whether it is really useful. It still doesn't allow being virtual without having an object. -- James Kanze (GABI Software) email:james.kanze (AT) gmail (DOT) com Conseils en informatique orientie objet/ Beratung in objektorientierter Datenverarbeitung 9 place Simard, 78210 St.-Cyr-l'Icole, France, +33 (0)1 30 23 00 34 -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Guest Posted: Wed Dec 20, 2006 3:45 am    Post subject: Re: static virtual functions Nicola Musatti a écrit : Quote: hasta_l3 (AT) hotmail (DOT) com wrote: Bob a écrit : hasta_l3 (AT) hotmail (DOT) com wrote: Well, Object Pascal has had the moral equivalent of virtual static members for more than a decade now. They come in very handy to implement factories or other kind of metaclass information. [...] Class methods are the Object Pascal equivalent of C++ static member functions. and they can be virtual... The implicit this (aka Self) argument is - sort of - a reference to a metaclass. That's precisely the point: in order to implement static virtual member functions you need an explicit representation of classes as objects, which C++ does not provide. Well, Object Pascal doesn't exactly represent classes as objects. They don't have data member, for example. And some kind of metaclass information already made its way into C++ under the disguise of type_info. Nevertheless, I dont think that C++ should ever have virtual static functions. For most of their power come from virtual constructor functions, and *that* concept is pretty foreign to C++. My point in this thread was to stress that the answer to the FAQ raised is not that virtual static functions are meaningless or useless. It is that they don't mix well with a few C++ key design decisions... Regards, Nicola --- Raoul -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Guest Posted: Wed Dec 20, 2006 3:46 am    Post subject: Re: static virtual functions Ron Natalie a écrit : Quote: hasta_l3 (AT) hotmail (DOT) com wrote: Class methods are the Object Pascal equivalent of C++ static member functions. and they can be virtual... The implicit this (aka Self) argument is - sort of - a reference to a metaclass. You still haven't indicated where this "class identity" comes from where it would be different from the class the method is defined in? In a nutshell : (I dont want to make the thread too much off-topic :-) If one has a type T that happens to be a class, then "class of T" is the type of the meta-class. The identifier T actually refers to the meta-class (and one can call a class method F with the syntax T.F) One can have variables of type "class of T" holding references to the metaclass, initialized from the value of the identifier T. If T2 derives from T, then one can have a method T2.F which overrides T.F, and have variables of type "class of T2". There is an implicit conversion from "class of T2" to "class of T", and one may therefore have a variable v with static type "class of T" but dynamic type "class of T2. The expression v.F will then call T2.F Have a nice day, Ron --- Raoul -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Dave Harris Guest Posted: Wed Dec 20, 2006 3:48 am    Post subject: Re: static virtual functions james.kanze (AT) gmail (DOT) com (James Kanze) wrote (abridged): Quote: Because classes are not reified as run-time objects, which means static member functions don't have a "this" or a vtable to keep track of the dynamic type of the class. That's not quite true. What does typeid( MyClass ) give you, if it isn't a concrete run-time object associated with the class. Yes, but all those objects have the same type, so don't help us here. Quote: It would probably be better to extend type_info. I don't see how that would help. Quote: I'm not sure what this buys us compared to something like the Java approach, where there is just one type Class (a very extended version of C++'s type_info). It buys us virtual static member functions. Static functions are not virtual in Java, partly because they failed to provide a full meta-class system. (Which I think is a shame, incidently. They missed a trick there.) Perhaps a more elaborate example would help? Given: struct A { static virtual int func() { return 1; } static int test() { return func(); } }; struct B : A { static virtual int func() { return 2; } } int x = B::test(); the compiler would produce something like: struct A { struct __Class { virtual int func() { return 1; } int test() { return func(); } }; static __Class __class; }; struct B : A { struct __Class : A::__Class { virtual int func() { return 2; } }; static __Class __class; }; int x = B::__class.test(); Notice that A::__class and B::__class are not just different objects, they have different types, too. This allows them to have different behaviour. Quote: A static function can be called in two ways: MyClass::staticFunc() ; ptrObj->staticFunc() ; These would be equivalent. In both cases, the object used would be MyClass::__class, not *ptrObj. Quote: The first way, you get the class member---no virtual resolution That misses the point. We want to get virtual resolution despite not having supplied an object explicitly. If we can supply an object we might as well just use virtual functions on that. -- Dave Harris, Nottingham, UK. -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Display posts from previous: All Posts1 Day7 Days2 Weeks1 Month3 Months6 Months1 Year Oldest FirstNewest First         C++Talk.NET Forum Index -> C++ Language (Moderated) All times are GMT Goto page 1, 2  Next Page 1 of 2   Jump to: Select a forum C/C++----------------C LanguageC Language (Moderated)C++ Language (Moderated)C++ language (comp.lang.c++)C++ language, library and standards Non-English C/C++ Newsgroups----------------C++ (German)C++ (French)  You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum Powered by phpBB © 2001, 2006 phpBB Group SEO toolkit © 2004-2006 webmedic.