C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  forward declaration of classes          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message HappyHippy Guest Posted: Wed Nov 23, 2005 8:40 pm    Post subject: forward declaration of classes Hi, I have a question about forward declaration of classes. First of all, a simple working example, which compiles and works with no problems. ///file class_a.h:/ #ifndef __CLASS_A_H__ #define __CLASS_A_H__ class classA { public: int getX() const { return m_nX; } void setX(int nX) { m_nX = nX; } private: int m_nX; }; #endif ///file class_b.h:/ #ifndef __CLASS_B_H__ #define __CLASS_B_H__ *class classA; //forward declaration of classA class* class classB { public: classB(classA *pA): m_pA(pA){} int getData() const; void setData(int nData); private: classA *m_pA; }; #endif ///file class_b.cpp:/ #include "class_b.h" *#include "class_a.h" //here we need to include "class_a.h"* int classB::getData() const { if(m_pA) return m_pA->getX(); return -1; //let's assume, -1 is an error code } void classB::setData(int nData) { if(m_pA) m_pA->setX(nData); } It is enough to just put "class classA;" in class_b.h file, since classB does not have classA members (only pointer-to-classA member), so forward declaration is enough for compiler, and "#include "class_a.h" is needed only in implementation file. That's fine, BUT... Often you can see a class or a struct declared as: ///file class_a.h:/ #ifndef __CLASS_A_H__ #define __CLASS_A_H__ typedef class { public: int getX() const { return m_nX; } void setX(int nX) { m_nX = nX; } private: int m_nX; } TClassA; #endif In this case the previous approach (all other files are the same) does not work... MS VC++ 7.1 gives me the following output: class_a.h(17) : error C2371: 'classA' : redefinition; different basic types class_b.h(1) : see declaration of 'classA' class_b.cpp(7) : error C2027: use of undefined type 'classA' class_b.h(1) : see declaration of 'classA' class_b.cpp(7) : error C2227: left of '->getX' must point to class/struct/union class_b.cpp(15) : error C2027: use of undefined type 'classA' class_b.h(1) : see declaration of 'classA' class_b.cpp(15) : error C2227: left of '->setX' must point to class/struct/union So, my question is: Is there any way to use forward declaration for classA to avoid using '#include "class_a.h"' in class_b.h file in the second case? Thank you! [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Bob Hairgrove Guest Posted: Thu Nov 24, 2005 8:03 am    Post subject: Re: forward declaration of classes On 23 Nov 2005 15:40:09 -0500, HappyHippy <karenman (AT) mail (DOT) ru> wrote: Quote: Hi, I have a question about forward declaration of classes. First of all, a simple working example, which compiles and works with no problems. ///file class_a.h:/ #ifndef __CLASS_A_H__ #define __CLASS_A_H__ Do not use identifiers beginning with leading underscores (see section 17.4.3.1.2 of the C++ standard). [Isn't this in the FAQ somewhere?? If not, it should be.] Quote: class classA { public: int getX() const { return m_nX; } void setX(int nX) { m_nX = nX; } private: int m_nX; }; #endif ///file class_b.h:/ #ifndef __CLASS_B_H__ #define __CLASS_B_H__ *class classA; //forward declaration of classA class* class classB { public: classB(classA *pA): m_pA(pA){} int getData() const; void setData(int nData); private: classA *m_pA; }; #endif ///file class_b.cpp:/ #include "class_b.h" *#include "class_a.h" //here we need to include "class_a.h"* int classB::getData() const { if(m_pA) return m_pA->getX(); return -1; //let's assume, -1 is an error code } void classB::setData(int nData) { if(m_pA) m_pA->setX(nData); } It is enough to just put "class classA;" in class_b.h file, since classB does not have classA members (only pointer-to-classA member), so forward declaration is enough for compiler, and "#include "class_a.h" is needed only in implementation file. That's fine, BUT... Often you can see a class or a struct declared as: ///file class_a.h:/ #ifndef __CLASS_A_H__ #define __CLASS_A_H__ typedef class { public: int getX() const { return m_nX; } void setX(int nX) { m_nX = nX; } private: int m_nX; } TClassA; #endif I never see this in C++ code I work on, only in legacy C code for struct declarations. Quote: In this case the previous approach (all other files are the same) does not work... MS VC++ 7.1 gives me the following output: class_a.h(17) : error C2371: 'classA' : redefinition; different basic types class_b.h(1) : see declaration of 'classA' class_b.cpp(7) : error C2027: use of undefined type 'classA' class_b.h(1) : see declaration of 'classA' class_b.cpp(7) : error C2227: left of '->getX' must point to class/struct/union class_b.cpp(15) : error C2027: use of undefined type 'classA' class_b.h(1) : see declaration of 'classA' class_b.cpp(15) : error C2227: left of '->setX' must point to class/struct/union So, my question is: Is there any way to use forward declaration for classA to avoid using '#include "class_a.h"' in class_b.h file in the second case? Look at your declaration with the typedef. Maybe you need to use classA instead of TClassA, or the other way around? -- Bob Hairgrove [email]NoSpamPlease (AT) Home (DOT) com[/email] [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Daniel Krügler Guest Posted: Thu Nov 24, 2005 2:03 pm    Post subject: Re: forward declaration of classes HappyHippy wrote: [snip] Quote: ///file class_b.h:/ #ifndef __CLASS_B_H__ #define __CLASS_B_H__ *class classA; //forward declaration of classA class* This is invalid C++. But you could write: class classA; with the remainder of the header unchanged. Quote: class classB { public: classB(classA *pA): m_pA(pA){} int getData() const; void setData(int nData); private: classA *m_pA; }; #endif ///file class_b.cpp:/ #include "class_b.h" *#include "class_a.h" //here we need to include "class_a.h"* int classB::getData() const { if(m_pA) return m_pA->getX(); return -1; //let's assume, -1 is an error code } I would strongly recommend to either assert(m_pA) (or throw an exception, if pA == 0) in the c'tor of classB accepting a classA*. Then every member function can be written w/o any further test with dubious return values, like the one above. Quote: void classB::setData(int nData) { if(m_pA) m_pA->setX(nData); } Same thing here. Quote: It is enough to just put "class classA;" in class_b.h file, since classB does not have classA members (only pointer-to-classA member), so forward declaration is enough for compiler, and "#include "class_a.h" is needed only in implementation file. That's fine, BUT... Often you can see a class or a struct declared as: ///file class_a.h:/ #ifndef __CLASS_A_H__ #define __CLASS_A_H__ typedef class { public: int getX() const { return m_nX; } void setX(int nX) { m_nX = nX; } private: int m_nX; } TClassA; do you mean classA instead of TClassA? Quote: #endif In this case the previous approach (all other files are the same) does not work... MS VC++ 7.1 gives me the following output: class_a.h(17) : error C2371: 'classA' : redefinition; different basic types class_b.h(1) : see declaration of 'classA' class_b.cpp(7) : error C2027: use of undefined type 'classA' class_b.h(1) : see declaration of 'classA' class_b.cpp(7) : error C2227: left of '->getX' must point to class/struct/union class_b.cpp(15) : error C2027: use of undefined type 'classA' class_b.h(1) : see declaration of 'classA' class_b.cpp(15) : error C2227: left of '->setX' must point to class/struct/union So, my question is: Is there any way to use forward declaration for classA to avoid using '#include "class_a.h"' in class_b.h file in the second case? Strictly speaking: No. Please have a look at the follow-ups of the thread "How Constructor of typedef class works", where you will find similar restrictions in the case, where the roles of the typedef-definition and the class declaration are reversed. Greetings from Bremen, Daniel Krügler [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Alberto Ganesh Barbati Guest Posted: Thu Nov 24, 2005 5:11 pm    Post subject: Re: forward declaration of classes HappyHippy wrote: Quote: Often you can see a class or a struct declared as: ///file class_a.h:/ #ifndef __CLASS_A_H__ #define __CLASS_A_H__ typedef class { public: int getX() const { return m_nX; } void setX(int nX) { m_nX = nX; } private: int m_nX; } TClassA; #endif Often? Maybe in C programs or badly written C++ programs. C++ programs never need to use that idiom and a well-written program should avoid it. Quote: So, my question is: Is there any way to use forward declaration for classA to avoid using '#include "class_a.h"' in class_b.h file in the second case? No, and that's one good reason why it's bad C++ programming style to use that idiom. Ganesh [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Ulrich Eckhardt Guest Posted: Thu Nov 24, 2005 5:18 pm    Post subject: Re: forward declaration of classes HappyHippy wrote: Quote: I have a question about forward declaration of classes. One note here: there is no such thing as a forward declaration. What you mean is a declaration like e.g. "class A;". Quote: #ifndef __CLASS_A_H__ #define __CLASS_A_H__ These are reserved symbols because they contain either two consecutive underscores or start with an underscore followed by a capital letter. Quote: class classA { [...] }; This is called a class definition (as opposed to a declaration), btw. Quote: *class classA; //forward declaration of classA class* I assume the first asterisk here is a typo and you didn't compile with it. Quote: That's fine, BUT... Often you can see a class or a struct declared as: typedef class { public: int getX() const { return m_nX; } void setX(int nX) { m_nX = nX; } private: int m_nX; } TClassA; OK. Firstly, let's see how this works: the above defines a class without giving the class a name, so in other words it is an anonymous class. However, in the same move it uses a typedef to assign an alias name to this anonymous class. Now, for the reasons: in C, names of types and names of structs, enums, unions all had their own namespace. So, in order to create an object of type 'struct foo' you really had to write 'struct foo f;' not just 'foo f;' like in C++. Using typedef, you could alias a struct's name in the types namespace and from thereon refer to the alias without the struct keyword before it. This is the reason many people don't even bother assigning a name in the first place but only create an alias of the anonymous struct. Quote: MS VC++ 7.1 gives me the following output: class_a.h(17) : error C2371: 'classA' : redefinition; different basic types class_b.h(1) : see declaration of 'classA' class_b.cpp(7) : error C2027: use of undefined type 'classA' class_b.h(1) : see declaration of 'classA' MSC gets confused because the name is a 'proper' name in one case and a typedef in the other case. Quote: So, my question is: Is there any way to use forward declaration for classA to avoid using '#include "class_a.h"' in class_b.h file in the second case? You can use typedef class foo {...} foo; i.e. use the same name for the class and the typedef. However, this only makes sense when it is not a class but a struct/enum/union and needs to be usable for C, too. Otherwise, don't use this construct. Lastly, there is an exception which (I think) makes the typedef of an anonymous struct equivalent to the struct being defined with a proper name, which is mostly for compatibility between C and C++. I only remember this vaguely though, maybe someone can step in. Uli [ 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: Fri Nov 25, 2005 6:18 pm    Post subject: Re: forward declaration of classes In message <ai9g53-fc7.ln1 (AT) satorlaser (DOT) homedns.org>, Ulrich Eckhardt <eckhardt (AT) satorlaser (DOT) com> writes Quote: HappyHippy wrote: I have a question about forward declaration of classes. One note here: there is no such thing as a forward declaration. What you mean is a declaration like e.g. "class A;". snip Surely a class declaration whose purpose is forwardly declaring a class can be called a 'forward declaration' without any risk of confusion. 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 kanze Guest Posted: Sat Nov 26, 2005 2:37 pm    Post subject: Re: forward declaration of classes Ulrich Eckhardt wrote: Quote: HappyHippy wrote: One note here: there is no such thing as a forward declaration. What you mean is a declaration like e.g. "class A;". Technically, this defines the class as an incomplete type. I think it is common usage, however, to refer to this particular type of definition as a forward declaration. [...] Quote: class classA { [...] }; This is called a class definition (as opposed to a declaration), btw. *class classA; //forward declaration of classA class* I assume the first asterisk here is a typo and you didn't compile with it. I think he just added the * to draw attention to the line. Quote: That's fine, BUT... Often you can see a class or a struct declared as: typedef class { public: int getX() const { return m_nX; } void setX(int nX) { m_nX = nX; } private: int m_nX; } TClassA; OK. Firstly, let's see how this works: the above defines a class without giving the class a name, so in other words it is an anonymous class. However, in the same move it uses a typedef to assign an alias name to this anonymous class. §7.1.3/5: "If the typedef declaration defines an unnamed class (or enum), the first typedef-name declared by the declaration to be that class type (or enum type) is used to denote the class type (or enum type) for linkage purposes only. It's a hack, for C compatibility, because the above idiom was frequent in C. (The above works in C because C uses structural type matching, at least between compilation units. C++ uses name equivalence.) Quote: MS VC++ 7.1 gives me the following output: class_a.h(17) : error C2371: 'classA' : redefinition; different basic types class_b.h(1) : see declaration of 'classA' class_b.cpp(7) : error C2027: use of undefined type 'classA' class_b.h(1) : see declaration of 'classA' MSC gets confused because the name is a 'proper' name in one case and a typedef in the other case. I don't know what the current situation is, but in the past, VC++ has had problems in this area. Things like a forward declaration "class A ;" not matching a definition which started "struct A { ...". So this could easily be simply a bug in the compiler. On the other hand, I'm not really too sure about the exact meaning of "for linkage purposes only". I would have interpreted it to mean that the following was legal: class A ; typedef struct { int x ; int y ; } A ; but the two compilers available to me (Sun CC and g++) both disagree, although the error message from g++ is: equiv.cc:8: error: conflicting declaration 'typedef struct A A' equiv.cc:2: error: 'struct A' has a previous declaration as `struct A' which suggests something wrong interally -- it is, at any rate treating the anonymous structure as having the name A. It may be that the compilers are refusing the typedef for A because there is already a symbol A defined. But I don't really see how this is different from something like: struct A { int x ; int y ; } ; int A ; which both compilers accept without complaining. On the other hand, the distinction makes some sense: the rule is for C compatibility, the above is legal in C, where as the example with the typedef (even replacing class with struct in the forward declaration) isn't. Quote: So, my question is: Is there any way to use forward declaration for classA to avoid using '#include "class_a.h"' in class_b.h file in the second case? You can use typedef class foo {...} foo; i.e. use the same name for the class and the typedef. However, this only makes sense when it is not a class but a struct/enum/union and needs to be usable for C, too. Otherwise, don't use this construct. Most of the time that I've had problems with these sort of things, it has involved headers that I couldn't modify. -- James Kanze GABI Software Conseils en informatique orientée objet/ Beratung in objektorientierter Datenverarbeitung 9 place Sémard, 78210 St.-Cyr-l'École, 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 Dan McLeran Guest Posted: Sat Nov 26, 2005 2:52 pm    Post subject: Re: forward declaration of classes Quote: Often? Maybe in C programs or badly written C++ programs. C++ programs never need to use that idiom and a well-written program should avoid it. Are you saying that forward-declaration is a bad practice? If so, I think Andrei Alexandrescu and Herb Sutter would disagree with you. This technique is one of the recommended practices from their book, C++ Coding Standards ([url]http://www.gotw.ca/publications/c++cs.htm)[/url]. I can't remember which one, because I left my copy at work. [ 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: Sun Nov 27, 2005 3:58 am    Post subject: Re: forward declaration of classes Dan McLeran wrote: Quote: Often? Maybe in C programs or badly written C++ programs. C++ programs never need to use that idiom and a well-written program should avoid it. Are you saying that forward-declaration is a bad practice? No. He's saying that you shouldn't write: typedef class { ... } C ; Just class C { ... } ; does the job nicely. -- James Kanze mailto: [email]james.kanze (AT) free (DOT) fr[/email] Conseils en informatique orientée objet/ Beratung in objektorientierter Datenverarbeitung 9 pl. Pierre Sémard, 78210 St.-Cyr-l'École, 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 Alberto Ganesh Barbati Guest Posted: Sun Nov 27, 2005 4:01 am    Post subject: Re: forward declaration of classes Dan McLeran wrote: Quote: Often? Maybe in C programs or badly written C++ programs. C++ programs never need to use that idiom and a well-written program should avoid it. Are you saying that forward-declaration is a bad practice? Of course not! I was referring only to this idiom reported in the OP and clearly quoted in my post: typedef class /* NO NAME HERE */ { /* definition here */ } TClassA; which makes little sense in C++. One of the reason why it makes so little sense is that you can't forward declare such a class. (let alone the fact that you can't declare neither a constructor nor a destructor). Ganesh [ 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 Page 1 of 1   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.