C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  typecasting the base class to derived class. Goto page 1, 2  Next          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message cman Guest Posted: Fri Feb 20, 2004 8:53 pm    Post subject: typecasting the base class to derived class. #include <stdio.h> class Base { public: Base() { printf("Base()n"); } ~Base() { printf("~Base()n"); } }; class Derived:public Base { public: Derived() { printf("Derived()n"); } void myConst() const { printf("Derived's myConst function x=%d",x); } ~Derived() { printf("~Derived()n"); } int x; }; main() { Derived *dptr; dptr=(Derived *)new Base(); dptr->x=10; dptr->myConst(); delete dptr; } In the above code "dptr->myConst()" succeeds without giving any error on Solaris, but generates an exception when ran on Winodws after compiling in VC++ My Question is Why does it not fail? since typecasting should not define x... this call should fail. I replaced "x" with "x[1024]" and then used the last element in the array above, to isolate the possiblity of having a fluke. But it still worked ! Thanks. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Thomas Tutone Guest Posted: Sat Feb 21, 2004 10:49 am    Post subject: Re: typecasting the base class to derived class. "cman" <complangman (AT) yahoo (DOT) com> wrote: Quote: #include <stdio.h class Base { public: Base() { printf("Base()n"); } Since you're deriving from it, you ordinarily should make the destructor virtual if there's any chance you'll use a Base pointer to refer to a Derived object. Quote: ~Base() { printf("~Base()n"); } }; class Derived:public Base { public: Derived() { printf("Derived()n"); } void myConst() const { printf("Derived's myConst function x=%d",x); } ~Derived() { printf("~Derived()n"); } int x; }; main() main returns int. Always. Please see the FAQ. Quote: { Derived *dptr; dptr=(Derived *)new Base(); OK, so you've casted it to something it's not. Probably better to get in the habit of using a static_cast<> in these circumstances... Quote: dptr->x=10; undefined behavior. Quote: dptr->myConst(); undefined behavior. Quote: delete dptr; } In the above code "dptr->myConst()" succeeds without giving any error on Solaris, but generates an exception when ran on Winodws after compiling in VC++ My Question is Why does it not fail? Your program has undefined behavior. "Appearing to work" and "crashing the computer" both fall in the category of undefined behavior. In fact, "appearing to work the first 5000 times I run it, then crashing on the 5001st time" also falls in the category of undefined behavior. Presumably what is actually happening - although it is obviously implementation-dependent - is that the computer is trashing random memory when it assigns to the nonexistent x member. On the Solaris, it so happens that trashing that memory doesn't cause any immediately visible effects - it may even be trashing memory that didn't have anything important in it anyway. On the Visual C++ platform, presumably you weren't so lucky. By the way, I don't know what you mean by "exception." A C++ exception? A hardware exception? A .NET structured exception? That is an overloaded term, and you haven't provided enough info to determine which meaning you intended. But I suppose that any of those still falls in the category of undefined behavior. Best regards, Tom __________________________________ Do you Yahoo!? Yahoo! Mail SpamGuard - Read only the mail you want. http://antispam.yahoo.com/tools [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Jonathan Turkanis Guest Posted: Sat Feb 21, 2004 1:59 pm    Post subject: Re: typecasting the base class to derived class. "cman" <complangman (AT) yahoo (DOT) com> wrote in message: <snip> Quote: main() { Derived *dptr; dptr=(Derived *)new Base(); dptr->x=10; dptr->myConst(); delete dptr; } <snip> Quote: My Question is Why does it not fail? It is undefined behavior, by 5.4/5 and 5.2.9/5. No diagnostic or explosion is required. Jonathan [ 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: Sat Feb 21, 2004 4:14 pm    Post subject: Re: typecasting the base class to derived class. cman wrote: Quote: Derived *dptr; dptr=(Derived *)new Base(); dptr->x=10; Undefined behaviour, all bets are off. Just don't do that. Uli -- Questions ? see C++-FAQ Lite: http://parashift.com/c++-faq-lite/ first ! [ 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: Sat Feb 21, 2004 4:25 pm    Post subject: Re: typecasting the base class to derived class. In message <20040220224944.15405.qmail (AT) web60907 (DOT) mail.yahoo.com>, Thomas Tutone <thomas8675309 (AT) yahoo (DOT) com> writes Quote: "cman" <complangman (AT) yahoo (DOT) com> wrote: #include class Base { public: Base() { printf("Base()n"); } Since you're deriving from it, you ordinarily should make the destructor virtual if there's any chance you'll use a Base pointer to refer to a Derived object. To be absolutely precise, if there is any chance that you will try to delete a dynamic instance through a base-class pointer. -- Francis Glassborow ACCU Author of 'You Can Do It!' see http://www.spellen.org/youcandoit For project ideas and contributions: http://www.spellen.org/youcandoit/projects [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Pete Becker Guest Posted: Sat Feb 21, 2004 6:23 pm    Post subject: Re: typecasting the base class to derived class. Francis Glassborow wrote: Quote: In message <20040220224944.15405.qmail (AT) web60907 (DOT) mail.yahoo.com>, Thomas Tutone <thomas8675309 (AT) yahoo (DOT) com> writes "cman" <complangman (AT) yahoo (DOT) com> wrote: #include class Base { public: Base() { printf("Base()n"); } Since you're deriving from it, you ordinarily should make the destructor virtual if there's any chance you'll use a Base pointer to refer to a Derived object. To be absolutely precise, if there is any chance that you will try to delete a dynamic instance through a base-class pointer. There's always a chance that someone will write incorrect code. The true test is not probabilistic. It is whether the design of the base class calls for deletion of derived types through base pointers. -- Pete Becker Dinkumware, Ltd. (http://www.dinkumware.com) [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top stelios xanthakis Guest Posted: Sun Feb 22, 2004 1:47 am    Post subject: Re: typecasting the base class to derived class. Pete Becker <petebecker (AT) acm (DOT) org> wrote Quote: There's always a chance that someone will write incorrect code. The true test is not probabilistic. It is whether the design of the base class calls for deletion of derived types through base pointers. The problem is an illegal downcast. This is why the new C++ casts (static/reintepret/dynamic/const) were invented by its designers: they can warn on such errors. As long as a base object is casted to a derrived object, the program is ill. It may not crash because accedentially the objects have the same size, but it is not ok. sx [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Pete Becker Guest Posted: Sun Feb 22, 2004 5:31 pm    Post subject: Re: typecasting the base class to derived class. stelios xanthakis wrote: Quote: Pete Becker <petebecker (AT) acm (DOT) org> wrote There's always a chance that someone will write incorrect code. The true test is not probabilistic. It is whether the design of the base class calls for deletion of derived types through base pointers. The problem is an illegal downcast. Please don't snip context. I was replying to these assertions: Quote: Since you're deriving from it, you ordinarily should make the destructor virtual if there's any chance you'll use a Base pointer to refer to a Derived object. To be absolutely precise, if there is any chance that you will try to delete a dynamic instance through a base-class pointer. Illegal downcasts are not involved here. The design issue raised here is when classes should have virtual destructors. -- Pete Becker Dinkumware, Ltd. (http://www.dinkumware.com) [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top kanze@gabi-soft.fr Guest Posted: Mon Feb 23, 2004 7:41 pm    Post subject: Re: typecasting the base class to derived class. Pete Becker <petebecker (AT) acm (DOT) org> wrote Quote: Francis Glassborow wrote: In message <20040220224944.15405.qmail (AT) web60907 (DOT) mail.yahoo.com>, Thomas Tutone <thomas8675309 (AT) yahoo (DOT) com> writes "cman" <complangman (AT) yahoo (DOT) com> wrote: #include class Base { public: Base() { printf("Base()n"); } Since you're deriving from it, you ordinarily should make the destructor virtual if there's any chance you'll use a Base pointer to refer to a Derived object. To be absolutely precise, if there is any chance that you will try to delete a dynamic instance through a base-class pointer. There's always a chance that someone will write incorrect code. The true test is not probabilistic. It is whether the design of the base class calls for deletion of derived types through base pointers. Well, there is one other case when need a virtual function: when the design calls for a checked cast from Base* to Derived*. That's what the original poster did, and given the original problem (and not just the bit of code you were responding to), one could argue that the "correct" solution would have been to add a virtual destructor to Base, and to use reinterpret_cast. That way, he would have gotten the exception or the error he seems to want. -- James Kanze GABI Software mailto:kanze (AT) gabi-soft (DOT) fr Conseils en informatique orientée objet/ http://www.gabi-soft.fr Beratung in objektorientierter Datenverarbeitung 11 rue de Rambouillet, 78460 Chevreuse, France, +33 (0)1 30 23 45 16 [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Jim Melton Guest Posted: Tue Feb 24, 2004 3:28 am    Post subject: Re: typecasting the base class to derived class. <kanze (AT) gabi-soft (DOT) fr> wrote Quote: Pete Becker <petebecker (AT) acm (DOT) org> wrote in message news:<403786D3.5AE43460 (AT) acm (DOT) org>... Francis Glassborow wrote: In message <20040220224944.15405.qmail (AT) web60907 (DOT) mail.yahoo.com>, Thomas Tutone <thomas8675309 (AT) yahoo (DOT) com> writes "cman" <complangman (AT) yahoo (DOT) com> wrote: #include <stdio.h class Base { public: Base() { printf("Base()n"); } Since you're deriving from it, you ordinarily should make the destructor virtual if there's any chance you'll use a Base pointer to refer to a Derived object. To be absolutely precise, if there is any chance that you will try to delete a dynamic instance through a base-class pointer. There's always a chance that someone will write incorrect code. The true test is not probabilistic. It is whether the design of the base class calls for deletion of derived types through base pointers. Well, there is one other case when need a virtual function: when the design calls for a checked cast from Base* to Derived*. That's what the original poster did, and given the original problem (and not just the bit of code you were responding to), one could argue that the "correct" solution would have been to add a virtual destructor to Base, and to use reinterpret_cast. That way, he would have gotten the exception or the error he seems to want. OK, you lost me. The OP's problem was that he used a C-style cast, which blindly does precisely what you tell it. I would have thought that dynamic_cast would have been the correct solution (presuming any virtual functions at all -- usually required to achieve polymorphism). Here's my touchstone for C++-style casts: static_cast - invoke defined conversion const_cast - add/remove constness dynamic_cast - typesafe down-cast reinterpret_cast - DWIM (do what i mean): interpret the bits of the argument as if they were the template type Of the four, I think reinterpret_cast is the hardest to get right. But I didn't think you would get any type-safety from it. Is it possible your recommendation is a typo? -- Opinions posted are those of the author. My company doesn't pay me enough to speak for them. </disclaimer> -- Jim Melton Software Architect, Fusion Programs Lockheed Martin IS&S (303) 971-3846 [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top stelios xanthakis Guest Posted: Tue Feb 24, 2004 7:26 pm    Post subject: Re: typecasting the base class to derived class. Pete Becker <petebecker (AT) acm (DOT) org> wrote Quote: stelios xanthakis wrote: Pete Becker <petebecker (AT) acm (DOT) org> wrote There's always a chance that someone will write incorrect code. The true test is not probabilistic. It is whether the design of the base class calls for deletion of derived types through base pointers. The problem is an illegal downcast. Please don't snip context. I was replying to these assertions: I'm sorry. Just wanted to point out what's really wrong before the thread went too far. Quote: Since you're deriving from it, you ordinarily should make the destructor virtual if there's any chance you'll use a Base pointer to refer to a Derived object. To be absolutely precise, if there is any chance that you will try to delete a dynamic instance through a base-class pointer. Illegal downcasts are not involved here. The design issue raised here is when classes should have virtual destructors. That would had been the case if the code was Base *b = new Derrived; The case: Derrived *d = (Derrived*) new Base; Is ill and because there are people reading clc++ who are not experts, we wouldn't want to confuse this ill program with the virtual dtors. BTW, to contribute to the discussion, I think a virtual destructor is also needed in this case: class A { int x; }; class B { int y; }; class C : A, B { int z; }; int f () { B *b = new C; delete b; } Despite the fact that the classes have no dtors, a virtual dtor must be present because the above crashes. How does the std describes this? I don't know... Cheers, Stelios [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top kanze@gabi-soft.fr Guest Posted: Tue Feb 24, 2004 7:29 pm    Post subject: Re: typecasting the base class to derived class. Jim Melton <jim.melton (AT) lmco (DOT) com> wrote Quote: Since you're deriving from it, you ordinarily should make the destructor virtual if there's any chance you'll use a Base pointer to refer to a Derived object. To be absolutely precise, if there is any chance that you will try to delete a dynamic instance through a base-class pointer. There's always a chance that someone will write incorrect code. The true test is not probabilistic. It is whether the design of the base class calls for deletion of derived types through base pointers. Well, there is one other case when need a virtual function: when the design calls for a checked cast from Base* to Derived*. That's what the original poster did, and given the original problem (and not just the bit of code you were responding to), one could argue that the "correct" solution would have been to add a virtual destructor to Base, and to use reinterpret_cast. That way, he would have gotten the exception or the error he seems to want. OK, you lost me. No problem. Pete said that you needed a virtual destructor if and only if the design called for deletion of derived types through a pointer to base types. Given that the original poster was downcasting, and that he wanted an error, he probably wanted to use dynamic_cast. Which only works if there is at least one virtual function. In the absense of any other virtual function, a virtual destructor is as good a choice as any. Another possible reason for a virtual destructor. Quote: The OP's problem was that he used a C-style cast, which blindly does precisely what you tell it. I would have thought that dynamic_cast would have been the correct solution (presuming any virtual functions at all -- usually required to achieve polymorphism). Exactly. And as there weren't any other virtual functions in the example:-). Quote: Here's my touchstone for C++-style casts: static_cast - invoke defined conversion const_cast - add/remove constness dynamic_cast - typesafe down-cast reinterpret_cast - DWIM (do what i mean): interpret the bits of the argument as if they were the template type Of the four, I think reinterpret_cast is the hardest to get right. But I didn't think you would get any type-safety from it. Is it possible your recommendation is a typo? Ooops. Very much so. I meant dynamic_cast, of course. (The others wouldn't require any virtual functions. In a very real sense, they are all "do as I say, and stop asking questions" types of operators.) -- James Kanze GABI Software mailto:kanze (AT) gabi-soft (DOT) fr Conseils en informatique orientée objet/ http://www.gabi-soft.fr Beratung in objektorientierter Datenverarbeitung 11 rue de Rambouillet, 78460 Chevreuse, France, +33 (0)1 30 23 45 16 [ 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 Feb 25, 2004 2:39 pm    Post subject: Re: typecasting the base class to derived class. In message <8a018872.0402240653.15835ae4 (AT) posting (DOT) google.com>, stelios xanthakis <mayall (AT) freemail (DOT) gr> writes Quote: BTW, to contribute to the discussion, I think a virtual destructor is also needed in this case: class A { int x; }; class B { int y; }; class C : A, B { int z; }; int f () { B *b = new C; delete b; } Despite the fact that the classes have no dtors, a virtual dtor must be present because the above crashes. How does the std describes this? I don't know... No, if it there were virtual dtors there would not be a problem (actually I am surprised that the above code does anything worse than leak memory). Of course there are dtors because currently there is no way to avoid every class object having a dtor (well you might declare one and then not define it but then the linker would complain). -- Francis Glassborow ACCU Author of 'You Can Do It!' see http://www.spellen.org/youcandoit For project ideas and contributions: http://www.spellen.org/youcandoit/projects [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Pete Becker Guest Posted: Wed Feb 25, 2004 3:04 pm    Post subject: Re: typecasting the base class to derived class. [email]kanze (AT) gabi-soft (DOT) fr[/email] wrote: Quote: No problem. Pete said that you needed a virtual destructor if and only if the design called for deletion of derived types through a pointer to base types. No, I said that IN THE CONTEXT TO WHICH I WAS REPLYING, the true rule was that you need a virtual destructor only if your class design calls for deleting objects of derived types through pointers to the base type. That was in contrast to the assertion that "[t]o be absolutely precise, if there is any chance that you will try to delete a dynamic instance through a base-class pointer [you should have a virtual destructor]." The point was that decisions about virtual destructors should be based on concrete design criteria, not on general principles and not on guesses. -- Pete Becker Dinkumware, Ltd. (http://www.dinkumware.com) [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Thomas Tutone Guest Posted: Wed Feb 25, 2004 3:05 pm    Post subject: Re: typecasting the base class to derived class. "stelios xanthakis" <mayall (AT) freemail (DOT) gr> wrote: [snip] Quote: BTW, to contribute to the discussion, I think a virtual destructor is also needed in this case: class A { int x; }; class B { int y; }; class C : A, B { int z; }; OK, C inherits privately from A & B. Quote: int f () { B *b = new C; That shouldn't even compile - B is a private base class of C. Quote: delete b; } Despite the fact that the classes have no dtors, Sure they do - well they would but for the fact that your code doesn't compile - they would have the compiler-generated default destructor. I think you mean they have no user-defined destructors, but so what? Quote: a virtual dtor must be present because the above crashes. How does the std describes this? I don't know... OK, your example includes both private inheritance and multiple inheritance, both of which are red herrings that distract from the issue. Let me simplify your example: class Base {}; class Derived : public Base { int i; }; int main() { Base* b(new Derived); delete b; } That's undefined behavior, because Base doesn't have a virtual destructor (although it does have a nonvirtual, compiler-generated destructor). Since it's undefined, "crashes" falls within the type of behavior the program could exhibit. Best regards, Tom __________________________________ Do you Yahoo!? Yahoo! Mail SpamGuard - Read only the mail you want. http://antispam.yahoo.com/tools [ 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.