C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  implicit constructor declaration (user annoyed) Goto page 1, 2  Next          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message sb Guest Posted: Tue Feb 24, 2004 3:29 am    Post subject: implicit constructor declaration (user annoyed) If there is at least one user-defined constructor, no constructors are implicitly declared. struct my_vec : public vector<int> { double foo; my_vec(size_t n) : vector<int>(2*n) {} // oops, no default ctor any more // have to do a lot of stupid typing now , like: my_vec() : vector<int>() {} // etc. }; Does anyone else find this behavior annoying? Instead, I would prefer for the compiler to implicitly declare all constructors that are not declared by the user. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Ivan Vecerina Guest Posted: Tue Feb 24, 2004 1:39 pm    Post subject: Re: implicit constructor declaration (user annoyed) "sb" <spam_bait101 (AT) yahoo (DOT) com> wrote Hi, Quote: If there is at least one user-defined constructor, no constructors are implicitly declared. Actually, only the default constructor is not implicitly declared anymore. A default copy-constructor is still implicitly defined (just as the default assignment operator is). Quote: struct my_vec : public vector<int> { NB: Because of the absence of a virtual destructor, deriving from an STL container tends to be dangerous. It is not a recommended to do so (a usual alternative is to use a data member instead, or replace the class with free functions). Quote: double foo; my_vec(size_t n) : vector<int>(2*n) {} // oops, no default ctor any more // have to do a lot of stupid typing now , like: my_vec() : vector<int>() {} 'a lot of stupid typing' ? Or just a 1-line declaration to add. In many designs (such as some polymorphic classes), it is inappropriate to have a default constructor. What would you do then? Quote: // etc. }; Does anyone else find this behavior annoying? Instead, I would prefer for the compiler to implicitly declare all constructors that are not declared by the user. As it is today, many people are having a hard time *disabling* implicitly generated functions: class NonCopyableStuff { ... private: // not implemented and private to disable implicit decls NonCopyableStuff(NonCopyableStuff const&); void operator=(NonCopyableStuff const&); }; Implicitly defined functions are actually quite dangerous. They might not do the right thing, and are bug-prone. Consider the very popular "rule of three" of C++: " Any class requiring a destructor, copy constructor, or an assignment operator must have all three implemented " Read for example: http://www.gotw.ca/gotw/069.htm This rule is so widely agreed on that some would like the next C++ standard to enforce it by disabling implicit function generations in some cases... Kind regards, Ivan -- http://ivan.vecerina.com/contact/?subject=NG_POST <- e-mail contact form [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Maciej Sobczak Guest Posted: Tue Feb 24, 2004 1:40 pm    Post subject: Re: implicit constructor declaration (user annoyed) Hi, sb wrote: Quote: If there is at least one user-defined constructor, no constructors are implicitly declared. struct my_vec : public vector<int> { double foo; my_vec(size_t n) : vector<int>(2*n) {} // oops, no default ctor any more // have to do a lot of stupid typing now , like: my_vec() : vector<int>() {} // etc. }; Does anyone else find this behavior annoying? There are people who find annoying the bare fact that the compiler does *anything* implicitly. Quote: Instead, I would prefer for the compiler to implicitly declare all constructors that are not declared by the user. Why? If you do not need the default constructor, there is no need "to do a lot of stupid typing". If you instead *do* need the default constructor, there is nothing stupid in writing it, is it? This rule is in the language because usually, if you write a constructor (any constructor), it is an indication that the class needs some specific initialization logic. The compiler humbly steps back here and says: "OK, you're the boss, if you say that the object needs to be *specially* constructed, then do it yourself". Have you also considered: my_vec(size_t n = 0) : vector<int>(2 * n) {} It may serve both needs (at least in this particular example). This is not the main topic here, buit please note also that deriving publicly from vector<T> is not considered to be a good programming practice. Most of the time this is a recipe for disaster or at least indication of fragile design. Other posters will for sure point that out. -- Maciej Sobczak : http://www.msobczak.com/ Programming : http://www.msobczak.com/prog/ [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Andre Heinen Guest Posted: Tue Feb 24, 2004 1:41 pm    Post subject: Re: implicit constructor declaration (user annoyed) On 23 Feb 2004 22:29:37 -0500, [email]spam_bait101 (AT) yahoo (DOT) com[/email] (sb) wrote: Quote: If there is at least one user-defined constructor, no constructors are implicitly declared. Unless I am mistaken, even if there are user-defined constructors the compiler still implicitly generates the copy constructor. Only the generation of the default constructor is disabled. Quote: snip Does anyone else find this behavior annoying? No, I like it. Quote: Instead, I would prefer for the compiler to implicitly declare all constructors that are not declared by the user. I would prefer the compiler not to implicitly generate any constructor at all. When you write your contructors yourself, you know what they do. When they are silently and implicitly generated, you always have to be on your guard and ensure they really do what you want. Actually, occasionally I go so far as to disable implicit generation by declaring the offending contructors (and the assignment operator) private, and by giving them no definition. -- Andre Heinen My address is "a dot heinen at europeanlink dot com" [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Risto Lankinen Guest Posted: Tue Feb 24, 2004 1:42 pm    Post subject: Re: implicit constructor declaration (user annoyed) Hi! "sb" <spam_bait101 (AT) yahoo (DOT) com> wrote Quote: If there is at least one user-defined constructor, no constructors are implicitly declared. Does anyone else find this behavior annoying? Instead, I would prefer for the compiler to implicitly declare all constructors that are not declared by the user. But some classes might absolutely require some construction argument. Letting default constructor be always synthesized would make it impossible for the programmer to enforce the requirement of a construction argument. But what you are asking makes sense in some cases. Since there already is a way to disable accidental auto-generation (by declaring said auto-generated functions in the "private" section, and dever defining them to prevent accidental use from the class itself), a corresponding mechanism to *enable* the auto-generation in the presence of additional ctors could be useful indeed. What would be a good syntax for such feature? I'd suggest the following: class C { public: C( SomeType arg ); using C(); // proposed syntax using C( const C & ); // proposed syntax }; Cheers! - Risto - [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Jeff Schwab Guest Posted: Tue Feb 24, 2004 7:14 pm    Post subject: Re: implicit constructor declaration (user annoyed) sb wrote: Quote: If there is at least one user-defined constructor, no constructors are implicitly declared. struct my_vec : public vector<int> { double foo; my_vec(size_t n) : vector<int>(2*n) {} // oops, no default ctor any more // have to do a lot of stupid typing now , like: my_vec() : vector<int>() {} // etc. }; Does anyone else find this behavior annoying? Instead, I would prefer for the compiler to implicitly declare all constructors that are not declared by the user. "All" constructors? How would the compiler know which constructors are needed? Default constructors would not be meaningful for some types. Anyway, your example is tedious. You don't need to indicate the base class's default constructor explicitly (that's why it's called a default), and you could have used a default argument to your first constructor. struct my_vec: vector<int> { double foo; my_vec(size_t n =0) : vector<int>(2*n) {} }; [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Craig Henderson Guest Posted: Tue Feb 24, 2004 7:19 pm    Post subject: Re: implicit constructor declaration (user annoyed) You're typing more than you need my_vec() { } is sufficient, you don't need to explicitly call the default ctor of the base class. Alternatively, you could give the ctor parameter a default value, so it can be used either way: my_vec(size_t n=0) : vector<int>(2*n) { } FWIW, I don't find it annoying, as I like to be explicit about what's defined for a class, so I generally define all compiler generated functions myself anyway, even if I don't implement them. Maybe that's just me :-) -- Craig "sb" <spam_bait101 (AT) yahoo (DOT) com> wrote Quote: If there is at least one user-defined constructor, no constructors are implicitly declared. struct my_vec : public vector<int> { double foo; my_vec(size_t n) : vector<int>(2*n) {} // oops, no default ctor any more // have to do a lot of stupid typing now , like: my_vec() : vector<int>() {} // etc. }; Does anyone else find this behavior annoying? Instead, I would prefer for the compiler to implicitly declare all constructors that are not declared by the user. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Andre Heinen Guest Posted: Tue Feb 24, 2004 7:21 pm    Post subject: Re: implicit constructor declaration (user annoyed) On 24 Feb 2004 08:40:51 -0500, Maciej Sobczak <no.spam (AT) no (DOT) spam.com> wrote: Quote: This is not the main topic here, buit please note also that deriving publicly from vector<T> is not considered to be a good programming practice. Most of the time this is a recipe for disaster or at least indication of fragile design. Other posters will for sure point that out. Well, Ivan did, and told us why: because of the absence of a virtual destructor. I must say it surprises me. Of course, I know that no virtual destructor means no polymorphism. But many classes are designed that way. How can deriving from std::vector be more dangerous than deriving from any other non-virtual class? As an example, in the standard library itself, aren't file stream classes derived from one another without any virtual destructor? -- Andre Heinen My address is "a dot heinen at europeanlink dot 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: Tue Feb 24, 2004 7:26 pm    Post subject: Re: implicit constructor declaration (user annoyed) [email]spam_bait101 (AT) yahoo (DOT) com[/email] (sb) wrote in message news:<221dd125.0402231407.7275ea1d (AT) posting (DOT) google.com>... Quote: If there is at least one user-defined constructor, no constructors are implicitly declared. struct my_vec : public vector<int> { double foo; my_vec(size_t n) : vector<int>(2*n) {} // oops, no default ctor any more // have to do a lot of stupid typing now , like: my_vec() : vector<int>() {} // etc. }; Does anyone else find this behavior annoying? No. Quote: Instead, I would prefer for the compiler to implicitly declare all constructors that are not declared by the user. Actually, I'd prefer for the compiler not to declare anything implicitly. But that would create problems of C compatibility; the current rule seems fairly reasonable. (Even better might be for the presense of any user defined constructor, destructor or assignment operator to suppress all of the compiler generated defaults -- including copy construction and assignment.) -- 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 Steve Dewhurst Guest Posted: Tue Feb 24, 2004 7:27 pm    Post subject: Re: implicit constructor declaration (user annoyed) [email]spam_bait101 (AT) yahoo (DOT) com[/email] (sb) wrote in message news:<221dd125.0402231407.7275ea1d (AT) posting (DOT) google.com>... Quote: If there is at least one user-defined constructor, no constructors are implicitly declared. .... Does anyone else find this behavior annoying? Instead, I would prefer for the compiler to implicitly declare all constructors that are not declared by the user. Not me. I like to have the option of forcing users of my classes to supply an initializer explicitly. Steve Steve Dewhurst www.semantics.org [ 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: Tue Feb 24, 2004 7:28 pm    Post subject: Re: implicit constructor declaration (user annoyed) In message <T1E_b.9936$k4.214112 (AT) news1 (DOT) nokia.com>, Risto Lankinen <rlankine (AT) hotmail (DOT) com> writes Quote: What would be a good syntax for such feature? I'd suggest the following: class C { public: C( SomeType arg ); using C(); // proposed syntax using C( const C & ); // proposed syntax }; Cheers! Well my proposal goes somewhat further and will be discussed at the next WG21 meeting. Here is a relevant extract from my paper: 2 The Proposals 1) Provide a specific mechanism to turn off all implicit member function generation. For this I propose that we enhance the grammar of class definition to allow a class to be qualified as explicit. In such a class none of the four special functions will be implicitly declared. 2) Provide a mechanism to explicitly declare a default constructor, copy constructor, copy assignment and destructor. For this purpose I propose that 'default' preceding the declaration explicitly requires the member function be compiler generated. The generation of an explicitly declared default constructor is not suppressed by the declaration of other constructors. The rules for compiler generation of the four members are as they are for the generation of the implicit versions. [12.1 para. 7 for implicit default constructor, 12.8 para. 4-8 for the copy constructor, 12.8 para. 10-14 for copy assignment, 12.4 para. 3-5,] Example: explicit class example{ public: default example(); example(int * i_ptr):val(*i_ptr){} default virtual ~example(); private: int val; }; // note the class is silly as such but is sufficient to demonstrate the combined // use of explicit and default. class derived: public example{ public: // public interface members private: std::string s; }; int main(){ int i(12); example e1; // OK, uses compiler generated default example e2(&i); // OK uses second constructor example e3(e1); // ERROR no implicit copy ctor derived d1; // OK derived d2(d1); //ERROR, cannot generate copy ctor example* d_ptr = new derived; delete d_ptr; // calls an implicitly generated // ~derived() first which calls ~string } -- 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 Francis Glassborow Guest Posted: Wed Feb 25, 2004 2:53 pm    Post subject: Re: implicit constructor declaration (user annoyed) In message <221dd125.0402231407.7275ea1d (AT) posting (DOT) google.com>, sb <spam_bait101 (AT) yahoo (DOT) com> writes Quote: If there is at least one user-defined constructor, no constructors are implicitly declared. struct my_vec : public vector<int> { double foo; my_vec(size_t n) : vector<int>(2*n) {} // oops, no default ctor any more // have to do a lot of stupid typing now , like: my_vec() : vector<int>() {} // etc. }; Does anyone else find this behavior annoying? Instead, I would prefer for the compiler to implicitly declare all constructors that are not declared by the user. Well learn to live with the current behaviour because any changes will not be in the direction that will reduce your annoyance. You preference would break several well established and commonly used C++ idioms. -- 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 Nicola Musatti Guest Posted: Wed Feb 25, 2004 2:54 pm    Post subject: Re: implicit constructor declaration (user annoyed) Andre Heinen <nospam (AT) nospam (DOT) invalid> wrote [...] Quote: Well, Ivan did, and told us why: because of the absence of a virtual destructor. I must say it surprises me. Of course, I know that no virtual destructor means no polymorphism. But many classes are designed that way. How can deriving from std::vector be more dangerous than deriving from any other non-virtual class? It isn't. Quote: As an example, in the standard library itself, aren't file stream classes derived from one another without any virtual destructor? No they aren't. The root of the hyerarchy is std::ios_base which has a virtual destructor. 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 Ben Hutchings Guest Posted: Wed Feb 25, 2004 3:00 pm    Post subject: Re: implicit constructor declaration (user annoyed) sb wrote: Quote: If there is at least one user-defined constructor, no constructors are implicitly declared. struct my_vec : public vector<int> { double foo; my_vec(size_t n) : vector<int>(2*n) {} // oops, no default ctor any more // have to do a lot of stupid typing now , like: my_vec() : vector<int>() {} // etc. }; Does anyone else find this behavior annoying? Yes. Quote: Instead, I would prefer for the compiler to implicitly declare all constructors that are not declared by the user. That would add to the existing problem of having to disable implicitly-declared functions in classes that aren't value classes. There is a proposal to modify the C++ syntax to allow the programmer to explicitly enable or disable the special member functions that are now implicitly declared: <http://std.dkuug.dk/jtc1/sc22/wg21/docs/papers/2003/n1445.htm>. {However note that this is superseded by n1582 in the pre-Sydney mailing. -mod/fwg} [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Drew Eckhardt Guest Posted: Wed Feb 25, 2004 3:01 pm    Post subject: Re: implicit constructor declaration (user annoyed) In article <221dd125.0402231407.7275ea1d (AT) posting (DOT) google.com>, sb <spam_bait101 (AT) yahoo (DOT) com> wrote: Quote: If there is at least one user-defined constructor, no constructors are implicitly declared. Actually if the user has not declared a copy constructor one will be declared as public and implemented using the members' copy constructors. This can have interesting effects. Quote: Does anyone else find this behavior annoying? It's good. If there aren't reasonable defaults I want a compile-time error without having to trick the compiler class foo { private: // Never instantiate. Code (I don't know how all my // template libraries are implemented - they're black boxes) is likely // to fail to compile because of the access, and if it does the link // fail although that might not happen until run-time at a // customer's site and I get an angry call at some oscene // time... foo(const foo&); }; to supress the behavior. Personally, I'd have preferred the implicit default constructor, copy constructor, and assignment operator generation to be limited to structs with no member functions (for backwards compatability with 'C'). -- <a href="http://www.poohsticks.org/drew/">Home Page</a> Life is a terminal sexually transmitted disease. [ 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.