C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  empty base class optimization          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message Thomas Mang Guest Posted: Fri Jul 30, 2004 9:54 am    Post subject: empty base class optimization Hello, While reading another thread about sizeof base-classes, I ran the following program on GCC 3.2 #include <iostream> #include <ostream> struct a{}; struct b : public a{}; int main() { std::cout << sizeof(b) << "n"; std::cout << sizeof(b::a); } The output is: 1 1 so we se in the first case EBO during work. However, I am not sure the second output is correct as the standard defines it. Yes, an object must practically have a size >= 1, but doen't the scope access operator direclty request a base-class object of b with size 0 ? Am I experiencing an restriction that is required, or is this a hole in the standard (the b::a access, where sizeof(a) = 0 in the EBO case)? thank you, Thomas [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Rob Williscroft Guest Posted: Fri Jul 30, 2004 3:29 pm    Post subject: Re: empty base class optimization Thomas Mang wrote in news:410A03FC.5DD800A0 (AT) unet (DOT) univie.ac.at in comp.lang.c++.moderated: Quote: Hello, While reading another thread about sizeof base-classes, I ran the following program on GCC 3.2 #include <iostream #include struct a{}; struct b : public a{}; int main() { std::cout << sizeof(b) << "n"; std::cout << sizeof(b::a); } The output is: 1 1 so we se in the first case EBO during work. However, I am not sure the second output is correct as the standard defines it. Yes, an object must practically have a size >= 1, but doen't the scope access operator direclty request a base-class object of b with size 0 ? Am I experiencing an restriction that is required, or is this a hole in the standard (the b::a access, where sizeof(a) = 0 in the EBO case)? No the EBO (Empty ...) is missleading. The Standard alows a base-class that has no data members (and no virtuals) to share the same address as another subobject with a distinct type. In essence: b binstance (void *)&(a &)binstance == (void *)&binstance; However if you write: struct b: a { a ainstance; }; Then: (void *)&(a &)binstance != (void *)&binstance.ainstance; I.e. the two 'a' subobjects must have a distinct address. In fact none of the examples sofar exhibit EBO, all we have is a derived class that "contains" a base class and *nothing* else. struct Empty {}; struct Another {}; struct EBO : Empty, Another {}; struct EBO2: Empty { char data; }; struct EBO2 : Empty, Another { char data; }; If you compiler supports EBO (and its enabled) all the above should have a size of 1, compiler dependant alignment requirments aside . struct no_EBO : EBO, Empty { }; no_EBO should have size of 2 as it contains 2 "Empty" subobjects and they are required to have a distinct address. Rob. -- http://www.victim-prime.dsl.pipex.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 Mang Guest Posted: Fri Jul 30, 2004 3:32 pm    Post subject: Re: empty base class optimization Thomas Mang schrieb: Quote: Hello, While reading another thread about sizeof base-classes, I ran the following program on GCC 3.2 #include <iostream #include struct a{}; struct b : public a{}; int main() { std::cout << sizeof(b) << "n"; std::cout << sizeof(b::a); } I apologize for a copy&pase error. The real test program should read as following: #include #include <ostream> struct a{}; struct b : public a{char c;}; int main(int argc, char *argv[]) { std::cout << "sizeof(a): " << sizeof(a) << "nsizeof(b): " << sizeof(b) << "nsizeof(b::a): " << sizeof(b::a); return 0; } However, the issue remains the same: EBO is taking place, and b::a has sizeof == 1. regards, Thomas [ 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 Jul 30, 2004 3:33 pm    Post subject: Re: empty base class optimization In article <410A03FC.5DD800A0 (AT) unet (DOT) univie.ac.at>, Thomas Mang <a9804814 (AT) unet (DOT) univie.ac.at> writes Quote: Hello, While reading another thread about sizeof base-classes, I ran the following program on GCC 3.2 #include <iostream #include struct a{}; struct b : public a{}; int main() { std::cout << sizeof(b) << "n"; std::cout << sizeof(b::a); } The output is: 1 1 so we se in the first case EBO during work. However, I am not sure the second output is correct as the standard defines it. Yes, an object must practically have a size >= 1, but doen't the scope access operator direclty request a base-class object of b with size 0 ? Am I experiencing an restriction that is required, or is this a hole in the standard (the b::a access, where sizeof(a) = 0 in the EBO case)? The compiler is telling you the truth. You asked for the sizeof a, albeit in the context of a b. It correctly told you the sizeof a. The fact that the 'storage' for a overlaps that for b is of no concern to the compiler. If you want something even more extreme try: #include <iostream> #include <ostream> struct a{}; struct a1{}; struct b : public a, public a1{char c;}; int main() { std::cout << sizeof(b) << "n"; std::cout << sizeof(b::a); std::cin.get(); } Which still outputs (quite correctly) 1 1 The rule is very simply that padding bytes maybe recycled. However now try: #include #include <ostream> struct a{}; struct b : public a {a ax;}; int main() { std::cout << sizeof(b) << "n"; std::cout << sizeof(b::a); std::cin.get(); } and the output is: 2 1 because the two instances of a (as a base and as a member) must have distinct addresses. -- 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 Peter C. Chapin Guest Posted: Fri Jul 30, 2004 3:36 pm    Post subject: Re: empty base class optimization Thomas Mang <a9804814 (AT) unet (DOT) univie.ac.at> wrote in news:410A03FC.5DD800A0 @unet.univie.ac.at: Quote: #include <iostream #include struct a{}; struct b : public a{}; int main() { std::cout << sizeof(b) << "n"; std::cout << sizeof(b::a); } The output is: 1 1 Yes, I believe that sizeof(b::a) == sizeof(a). The namespace qualification in this case only tells the compiler that you are talking about the type 'a' used as a base of 'b'. The type 'a' is defined above as an empty struct, but it has a size > 0. The expression sizeof(b::a) does not in any way refer to the size of the base subobject of b. Consider struct X { typedef int counter; }; Now sizeof(X::counter) = sizeof(int), right? That fact that you are using an int inside an X doesn't change the way sizeof works for int. Peter [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Alberto Barbati Guest Posted: Fri Jul 30, 2004 3:37 pm    Post subject: Re: empty base class optimization Thomas Mang wrote: Quote: so we se in the first case EBO during work. However, I am not sure the second output is correct as the standard defines it. Yes, an object must practically have a size >= 1, but doen't the scope access operator direclty request a base-class object of b with size 0 ? "b::a" is another way to say "a". The presence of the scope access operator does not request the "base-class subobject of b of type a". There is no way in C++ to actually refer to such entity as a type so you can't get its size. Alberto [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Dag Viken Guest Posted: Fri Jul 30, 2004 4:02 pm    Post subject: Re: empty base class optimization "Thomas Mang" <a9804814 (AT) unet (DOT) univie.ac.at> wrote Quote: Hello, While reading another thread about sizeof base-classes, I ran the following program on GCC 3.2 #include <iostream #include struct a{}; struct b : public a{}; int main() { std::cout << sizeof(b) << "n"; std::cout << sizeof(b::a); } The output is: 1 1 so we se in the first case EBO during work. However, I am not sure the second output is correct as the standard defines it. Yes, an object must practically have a size >= 1, but doen't the scope access operator direclty request a base-class object of b with size 0 ? Am I experiencing an restriction that is required, or is this a hole in the standard (the b::a access, where sizeof(a) = 0 in the EBO case)? I think it is a requirement of the C++ standard that all class and struct objects must have nonzero size, base class or not. If you have an empty base class with a derived class containing a single byte, both classes would be of size 1. Like it or not. Quote: thank you, Thomas [ 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 Jul 31, 2004 3:43 am    Post subject: Re: empty base class optimization Thomas Mang wrote: Quote: Am I experiencing an restriction that is required, or is this a hole in the standard (the b::a access, where sizeof(a) = 0 in the EBO case)? Your case exactly is described in 5.3.3/2. When applying sizeof to a base class subobject, the result is the size of that object's type. [ 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 Jul 31, 2004 3:48 am    Post subject: Re: empty base class optimization Dag Viken wrote: Quote: I think it is a requirement of the C++ standard that all class and struct objects must have nonzero size, base class or not. If you have an empty base class with a derived class containing a single byte, both classes would be of size 1. Like it or not. No. All objects have a non-zero size, but empty base classes may have zero size, as long as this would not place two such classes of the same type atthe same address. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Dag Viken Guest Posted: Sun Aug 01, 2004 11:04 am    Post subject: Re: empty base class optimization "Hyman Rosen" <hyrosen (AT) mail (DOT) com> wrote Quote: Dag Viken wrote: I think it is a requirement of the C++ standard that all class and struct objects must have nonzero size, base class or not. If you have an empty base class with a derived class containing a single byte, both classes would be of size 1. Like it or not. No. All objects have a non-zero size, but empty base classes may have zero size, as long as this would not place two such classes of the same type atthe same address. Yes, empty base classes may have zero size but still have sizeof() == 1. Does a virtual base class have size zero? I may be wrong but I am using VC7 and always get sizeof() >=1. Dag [ 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: Sun Aug 01, 2004 3:33 pm    Post subject: Re: empty base class optimization In article <SDYOc.4861$9Y6.2244 (AT) newsread1 (DOT) news.pas.earthlink.net>, Dag Viken <dag.viken (AT) earthlink (DOT) net> writes Quote: Yes, empty base classes may have zero size but still have sizeof() == 1. Does a virtual base class have size zero? Yes, but note that classes with virtual members are not empty (at least using the current implementations of virtual) Quote: I may be wrong but I am using VC7 and always get sizeof() >=1. -- 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 John Torjo Guest Posted: Mon Aug 02, 2004 10:32 am    Post subject: Re: empty base class optimization Francis Glassborow <francis (AT) robinton (DOT) demon.co.uk> wrote Quote: In article <SDYOc.4861$9Y6.2244 (AT) newsread1 (DOT) news.pas.earthlink.net>, Dag Viken <dag.viken (AT) earthlink (DOT) net> writes Yes, empty base classes may have zero size but still have sizeof() == 1. Does a virtual base class have size zero? If it's derived from, indeed it can have size zero. Quote: Yes, but note that classes with virtual members are not empty (at least using the current implementations of virtual) that is, sizeof(some_class) > 1, if the class has a virtual member - which applies to virtual destructors as well. So, if a base class has virtual members, the EBO is meaningless (there is nothing to be optimized). Is that correct? Best, John John Torjo Freelancer -- [email]john (AT) torjo (DOT) com[/email] Contributing editor, C/C++ Users Journal -- "Win32 GUI Generics" -- generics & GUI do mix, after all -- http://www.torjo.com/win32gui/ Professional Logging Solution for FREE -- http://www.torjo.com/code/logging.zip (logging - C++) -- http://www.torjo.com/logview/ (viewing/filtering - Win32) -- http://www.torjo.com/logbreak/ (debugging - Win32) (source code available) [ 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.