C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  References to Pointers And Polymorphism          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message Brian Neal Guest Posted: Fri Jan 23, 2004 10:03 am    Post subject: References to Pointers And Polymorphism Please consider the following code: class B {}; class D : public B {}; D d; D* dPtr = &d; B* bPtr = dPtr; // fine B& bRef = d; // fine D*& dPtrRef = dPtr; // fine B*& bPtrRef = dPtrRef; // ********* oops ************ bPtrRef = dPtr; // fine The line marked "oops" generates warnings on 2 compilers I use, saying a temporary will be used in order to initialize bPtrRef. For some reason I cannot wrap my mind around this. I was just going to add a cast, but the temporary warning makes me think it will not do what I want anyway. Can someone explain what is happening here? Thanks. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Francis Glassborow Guest Posted: Fri Jan 23, 2004 9:03 pm    Post subject: Re: References to Pointers And Polymorphism In message <26d6ee84.0401220539.299bcc30 (AT) posting (DOT) google.com>, Brian Neal <SurfCedarRapids (AT) netscape (DOT) net> writes Quote: Please consider the following code: class B {}; class D : public B {}; So all Ds are also Bs. That means that all pointers/references to B may also point/reference Ds. However it does not mean that all pointers to D are pointers to B. I.e. the relationship is between B and D NOT between B* and D*. This is something that is subtle but essential to understand. It begins to bite unpleasantly when you have a multiple inheritance hierarchy. The Compiler has no way to know that you do not somewhere else provide such a relationship: e.g.: class C : public D, public B {}; Now a B* might be pointing to either of two B sub-objects, whether it is a subobject of D depends on how you initialise it. Quote: D d; D* dPtr = &d; B* bPtr = dPtr; // fine B& bRef = d; // fine D*& dPtrRef = dPtr; // fine B*& bPtrRef = dPtrRef; And here you treat a D* as if it is a B* but it isn't (it is the wrong type) so the compiler has two options, it can outright reject your code or it can (incorrectly IMO) be try to be 'helpful' and create a temporary instance of a B* to bind to the reference. Incorrectly because the licence to bind references to temporaries is only allowed for const references. Quote: // ********* oops ************ bPtrRef = dPtr; // fine The line marked "oops" generates warnings on 2 compilers I use, saying a temporary will be used in order to initialize bPtrRef. For some reason I cannot wrap my mind around this. I was just going to add a cast, but the temporary warning makes me think it will not do what I want anyway. Can someone explain what is happening here? Thanks. The question is, of course, exactly what it is that you want because the compiler is correct in posting 'here be dragons' notices on your code. -- Francis Glassborow ACCU Author of 'You Can Do It!' see http://www.spellen.org/youcandoit [ 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: Fri Jan 23, 2004 9:06 pm    Post subject: Re: References to Pointers And Polymorphism Brian Neal wrote: Quote: class B {}; class D : public B {}; D d; D* dPtr = &d; D*& dPtrRef = dPtr; // fine B*& bPtrRef = dPtrRef; // ********* oops ************ The line marked "oops" generates warnings on 2 compilers I use, saying a temporary will be used in order to initialize bPtrRef. A pointer to D doesn't have an ISA relationship with a pointer to B. When converting from D* to B*, which is possible, a temporary is created. BTW, take a look at this: D* pd; B*& pb = pb; pb = new B;// oops, pd now points to a B! Uli [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Ben Hutchings Guest Posted: Fri Jan 23, 2004 9:11 pm    Post subject: Re: References to Pointers And Polymorphism Brian Neal wrote: Quote: Please consider the following code: class B {}; class D : public B {}; D d; D* dPtr = &d; B* bPtr = dPtr; // fine B& bRef = d; // fine D*& dPtrRef = dPtr; // fine B*& bPtrRef = dPtrRef; // ********* oops ************ bPtrRef = dPtr; // fine The line marked "oops" generates warnings on 2 compilers I use, saying a temporary will be used in order to initialize bPtrRef. These compilers are broken; they should not allow the conversion at all. Quote: For some reason I cannot wrap my mind around this. I was just going to add a cast, but the temporary warning makes me think it will not do what I want anyway. Can someone explain what is happening here? Suppose this kind of conversion was allowed on references. Then you could do: B*& bPtrRef = dPtrRef; // bPtrRef is alias for dPtr bPtrRef = new B; // dPtr now points to a B dPtr->some_method_on_D(); // oops! [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Hyman Rosen Guest Posted: Sat Jan 24, 2004 1:12 pm    Post subject: Re: References to Pointers And Polymorphism Brian Neal wrote: Quote: Can someone explain what is happening here? If your code were legal, then you could assign an arbitrary B* to bPtrRef, resulting in a D* variable (dPtr) holding a non-D* pointer. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Eugene Alterman Guest Posted: Mon Jan 26, 2004 12:05 am    Post subject: Re: References to Pointers And Polymorphism "Brian Neal" <SurfCedarRapids (AT) netscape (DOT) net> wrote Quote: Please consider the following code: class B {}; class D : public B {}; D d; D* dPtr = &d; B* bPtr = dPtr; // fine B& bRef = d; // fine D*& dPtrRef = dPtr; // fine B*& bPtrRef = dPtrRef; // ********* oops ************ bPtrRef = dPtr; // fine The line marked "oops" generates warnings on 2 compilers I use, saying a temporary will be used in order to initialize bPtrRef. For some reason I cannot wrap my mind around this. I was just going to add a cast, but the temporary warning makes me think it will not do what I want anyway. Can someone explain what is happening here? A similar thing is happening here: int i = 5; long l = i; // fine int& iRef = i; // fine long& lRef = iRef // ********* oops ************ lRef = i; // fine I suppose the above does not baffle you. An int is convertible to a long, but an int is not a long. D* is convertible to B* (since a D 'IS A' B), but a D* IS NOT a B*. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Allan W Guest Posted: Wed Jan 28, 2004 1:29 pm    Post subject: Re: References to Pointers And Polymorphism Quote: Brian Neal <SurfCedarRapids (AT) netscape (DOT) net> writes D*& dPtrRef = dPtr; // fine B*& bPtrRef = dPtrRef; Francis Glassborow <francis (AT) robinton (DOT) demon.co.uk> wrote Quote: here you treat a D* as if it is a B* but it isn't (it is the wrong type) agreed Quote: so the compiler has two options, it can outright reject your code or it can (incorrectly IMO) be try to be 'helpful' and create a temporary instance of a B* to bind to the reference. It really need to allow this code to be compiled -- for legacy support (pre-standard C++). I doubt there's much code left that needs it, though. The C++ standard mandates a 'diagnostic message' but it doesn't say that this can't just be a warning. Note: When you get a warning, it normally means you have made a mistake! Just because your code compiles doesn't mean it doesn't have mistakes. Quote: Incorrectly because the licence to bind references to temporaries is only allowed for const references. I don't have D&C in front of me, but I think I understand this reasoning. If you're using a const reference, you probably only want to read the values. Even if you don't realize that you have a temporary, it was constructed by copying values from the original object, so this will work mostly correctly. (I think I could construct counter-examples based on executing methods of the base class instead of the derived class, but in the absence of this type of problem it ought to get the correct values the vast majority of the time.) OTOH, if you use a non-const reference, you might very well want to make changes to the object. Now it matters that you're modifying a temporary instead of the original object -- if you don't realize you're using a temporary, your changes will be lost. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Francis Glassborow Guest Posted: Wed Jan 28, 2004 8:12 pm    Post subject: Re: References to Pointers And Polymorphism In message <7f2735a5.0401271248.1193f57a (AT) posting (DOT) google.com>, Allan W <allan_w (AT) my-dejanews (DOT) com> writes Quote: Francis Glassborow <francis (AT) robinton (DOT) demon.co.uk> wrote here you treat a D* as if it is a B* but it isn't (it is the wrong type) agreed so the compiler has two options, it can outright reject your code or it can (incorrectly IMO) be try to be 'helpful' and create a temporary instance of a B* to bind to the reference. It really need to allow this code to be compiled -- for legacy support (pre-standard C++). I doubt there's much code left that needs it, though. The C++ standard mandates a 'diagnostic message' but it doesn't say that this can't just be a warning. Yes, you are right. In general once a compiler has issued at least one diagnostic message a TU it can go ahead and do anything. In this case I would be happy with the quality of the implementation if it always warned (at all warning levels) about creating a temporary as an 'extension'. -- Francis Glassborow ACCU Author of 'You Can Do It!' see http://www.spellen.org/youcandoit [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Balog Pal Guest Posted: Wed Jan 28, 2004 8:34 pm    Post subject: Re: References to Pointers And Polymorphism "Allan W" <allan_w (AT) my-dejanews (DOT) com> wrote Quote: D*& dPtrRef = dPtr; // fine B*& bPtrRef = dPtrRef; so the compiler has two options, it can outright reject your code or it can (incorrectly IMO) be try to be 'helpful' and create a temporary instance of a B* to bind to the reference. If you're using a const reference, you probably only want to read the values. Even if you don't realize that you have a temporary, it was constructed by copying values from the original object, so this will work mostly correctly. (I think I could construct counter-examples based on executing methods of the base class instead of the derived class, but in the absence of this type of problem it ought to get the correct values the vast majority of the time.) In the example we have a pointer. So what you write can not apply. As a pointer is a builtin type. And if you go dereferencing it, any polymorphic bahavior will work fine. And nonpolymorphic behavior will work 'as requested' ;) B* const & bPtrRef = dPtrRef; could compile binding making bPtrRef bound to a copy of dPtr created in a temporary. The already mentioned problem of setting the pointer to point to something else can't strike due to constness. Though if the programmer is not aware of the copy, he van do delete dPtr; dPtr = new ... ; leaving bPtrRef dongling. Quote: OTOH, if you use a non-const reference, you might very well want to make changes to the object. Now it matters that you're modifying a temporary instead of the original object -- if you don't realize you're using a temporary, your changes will be lost. That is the other side of the problem. Paul [ 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.