C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  Re: Destructor not called when object is declared globally          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message kanze@gabi-soft.fr Guest Posted: Thu Jun 26, 2003 7:11 pm    Post subject: Re: Destructor not called when object is declared globally Francis Glassborow <francis.glassborow (AT) ntlworld (DOT) com> wrote Quote: In message <eb906752.0306232126.56eaf816 (AT) posting (DOT) google.com>, Sarang Aravamuthan <sarang_rajan (AT) yahoo (DOT) com> writes I'm observing a rather strange phenomenon with the following piece of code: /* This seems to work only when redirected to a file, for eg. through sometests > filename Also it works only when X is declared the way it is but not when declared starter X(); */ #include <iostream using namespace std; class starter { public: starter() {cout << "Initialize n";} ~starter() {cout << "Clean upn"; }; }; starter X; void main() returns an int not a void { cout << "in main()n"; } When I compile and run this code in debug mode or in command line, I get the following output (i.e. the destructor is not invoked). Initialize in main() Well your code (after correcting the return type of main()) ran exactly the same in both cases when I used G++. I suspect that your environment (and you do not say what that is) is ignoring output after exit from main() but the output is still generated as is shown by the redirection. While I think you're right in this case, it is worth pointing out that cout is only guaranteed to be constructed after the first instance of ios::Init has been constructed (or main has been entered, whichever comes first). As it happens, g++ (like many compilers) declares a static instance of ios::Init in transition from the classical iostreams (which required this) easier. But the standard doesn't requirer it; I'd add a definition of an ios::Init object somewhere to be sure (probably as a static variable in the constructor). The other point is that the destructor of the last ios::Init object will call flush on the stream. Again, if there isn't one, no one calls flush, and the output doesn't get to the OS. (The standard stream objects are never destructed, so the data won't get flushed there either.) The easiest way to be sure here is to use std::endl instead of 'n' -- I'd say that this is a generally preferable solution except when you explicitly know that there will be a batch of output (or a close or program termination) immediately following. Basically, std::endl always works, whereas 'n' doesn't. -- James Kanze GABI Software mailto:kanze (AT) gabi-soft (DOT) fr Conseils en informatique orientée objet/ Beratung in objektorientierter Datenverarbeitung 11 rue de Rambouillet, 78460 Chevreuse, France, Tél. : +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 MiniDisc_2k2 Guest Posted: Thu Jun 26, 2003 7:42 pm    Post subject: Re: Destructor not called when object is declared globally Try flushing the buffer (note I replaced your n's with endl's because that'll flush the buffer). If you don't want to end the line but need to flush the buffer (for future reference), you need to make a call to cout.flush(). <snip> Quote: class starter { public: starter() {cout << "Initialize " << endl;} ~starter() {cout << "Clean up" << endl; }; Quote: }; starter X; void main() { cout << "in main()" << endl; } When I compile and run this code in debug mode or in command line, I get the following output (i.e. the destructor is not invoked). Initialize in main() That's because the last line is not flushed. It remains in the buffer while the program ends. Quote: However, if I redirect the output to a file through "sometest.exe outfile", outfile contains Initialize in main() Clean up That's because files do not need to be flushed. Quote: as one would expect. Moreover, if the class object is declared as starter X(); then neither the constructor, nor the destructor is invoked in any way I generate the output, i.e. the output is just the string "in main()". Not quite sure why that is, but perhaps it's because of doing something like this: #include void a(); void b(); int main() { return (0); } void a() { b(); } void b() { cout << "in b()"; } Nothing would be displayed, even though you have a call to b(). Just like the constructor. It's like you're never calling it, because you never actually reach that code. Quote: Can someone shed any insight on why the program exits without calling the destructor--or is this a problem with my compiler? Neither. The destructor is called and your compiler is fine. -- MiniDisc_2k2 To reply, replace nospam.com with cox dot net [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top MiniDisc_2k2 Guest Posted: Thu Jun 26, 2003 7:43 pm    Post subject: Re: Destructor not called when object is declared globally Try flushing the buffer (note I replaced your n's with endl's because that'll flush the buffer). If you don't want to end the line but need to flush the buffer (for future reference), you need to make a call to cout.flush(). <snip> Quote: class starter { public: starter() {cout << "Initialize " << endl;} ~starter() {cout << "Clean up" << endl; }; Quote: }; starter X; void main() { cout << "in main()" << endl; } When I compile and run this code in debug mode or in command line, I get the following output (i.e. the destructor is not invoked). Initialize in main() That's because the last line is not flushed. It remains in the buffer while the program ends. Quote: However, if I redirect the output to a file through "sometest.exe outfile", outfile contains Initialize in main() Clean up That's because files do not need to be flushed. Quote: as one would expect. Moreover, if the class object is declared as starter X(); then neither the constructor, nor the destructor is invoked in any way I generate the output, i.e. the output is just the string "in main()". Not quite sure why that is, but perhaps it's because of doing something like this: #include void a(); void b(); int main() { return (0); } void a() { b(); } void b() { cout << "in b()"; } Nothing would be displayed, even though you have a call to b(). Just like the constructor. It's like you're never calling it, because you never actually reach that code. Quote: Can someone shed any insight on why the program exits without calling the destructor--or is this a problem with my compiler? Neither. The destructor is called and your compiler is fine. -- MiniDisc_2k2 To reply, replace nospam.com with cox dot net [ 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 Jun 27, 2003 12:24 am    Post subject: Re: Destructor not called when object is declared globally In message <30869baf.0306240556.12847b (AT) posting (DOT) google.com>, David Cattarin <dittoC (AT) ix (DOT) netcom.com> writes Quote: I suspect the destructor is getting called, but I think your iostreams implementation might have a flushing problem when the streams are destructed. I would have thought that the standard requires cout to flush when it is destroyed, but that might not be the case. Actually the standard requires that cout is not destroyed :-) -- ACCU Spring Conference 2003 April 2-5 The Conference you should not have missed ACCU Spring Conference 2004 Late April Francis Glassborow ACCU [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Tony Guest Posted: Fri Jun 27, 2003 4:50 pm    Post subject: Re: Destructor not called when object is declared globally Quote: However, if I redirect the output to a file through "sometest.exe outfile", outfile contains Initialize in main() Clean up That's because files do not need to be flushed. I tried this in MS VC++ 6.0. I am not sure if this is a flush problem. Even I added "endl" to the dtor. The print in dtor is not done. If the output is redirecting to a file. The print is correct just as it is expected. Then I tried it in SGI machine (But I am not sure what compiler I am using -. The print in dtor is done correctly even without "endl". Tony [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top David Cattarin Guest Posted: Fri Jun 27, 2003 8:15 pm    Post subject: Re: Destructor not called when object is declared globally Francis Glassborow <francis.glassborow (AT) ntlworld (DOT) com> wrote Quote: In message <30869baf.0306240556.12847b (AT) posting (DOT) google.com>, David Cattarin <dittoC (AT) ix (DOT) netcom.com> writes I suspect the destructor is getting called, but I think your iostreams implementation might have a flushing problem when the streams are destructed. I would have thought that the standard requires cout to flush when it is destroyed, but that might not be the case. Actually the standard requires that cout is not destroyed Really? Where does it say that? Now I'm confused. I thought cout was a global static object and would be destroyed on program termination. I also thought that note 209 might imply that it got flushed at this point (though I know it says underlying C stream). Dave [ 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 Jun 28, 2003 3:41 pm    Post subject: Re: Destructor not called when object is declared globally In message <30869baf.0306270733.75c7f444 (AT) posting (DOT) google.com>, David Cattarin <dittoC (AT) ix (DOT) netcom.com> writes Quote: Really? Where does it say that? Now I'm confused. I thought cout was a global static object and would be destroyed on program termination. I also thought that note 209 might imply that it got flushed at this point (though I know it says underlying C stream). See 27.3 para 2 last sentence: The objects are not destroyed during program execution. And if you look at the section where footnote 209 is you will see that the dtors for static objects are called before the C streams are closed. Nothing about dtors being called latter and calling a dtor for cin, cout etc. without closing the stream to which it is attached might cause some problems. -- ACCU Spring Conference 2003 April 2-5 The Conference you should not have missed ACCU Spring Conference 2004 Late April Francis Glassborow ACCU [ 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 Jun 30, 2003 10:34 am    Post subject: Re: Destructor not called when object is declared globally [email]tongru_huo (AT) yahoo (DOT) com[/email] (Tony) wrote in message news:<1f87458d.0306270549.46a18217 (AT) posting (DOT) google.com>... Quote: However, if I redirect the output to a file through "sometest.exe outfile", outfile contains Initialize in main() Clean up That's because files do not need to be flushed. I tried this in MS VC++ 6.0. I am not sure if this is a flush problem. Even I added "endl" to the dtor. The print in dtor is not done. If the output is redirecting to a file. The print is correct just as it is expected. Then I tried it in SGI machine (But I am not sure what compiler I am using -. The print in dtor is done correctly even without "endl". Formally, unless you've done something to ensure that cout was constructed before the print in the constructor, you have undefined behavior. Practically, I don't think that this is your problem -- the most likely symptom if the code wasn't going to work would be a core dump before entering main. If the print in the destructor doesn't appear even when endl is used, the compiler is not conformant. In practice, with most of the implementations I've seen, the print in the destructor will appear even without endl, but this is not guaranteed, *unless* you have defined an instance of std::ios::Init before your static object in the same file. In practice, if no instance of std::ios::Init has been constructed before your constructor executes, you will almost surely core dump, and if an instance has been constructed with static lifetime, you don't need the endl. -- 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, Tél. : +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 David Cattarin Guest Posted: Mon Jun 30, 2003 6:49 pm    Post subject: Re: Destructor not called when object is declared globally Francis Glassborow <francis.glassborow (AT) ntlworld (DOT) com> wrote Quote: In message <30869baf.0306270733.75c7f444 (AT) posting (DOT) google.com>, David Cattarin <dittoC (AT) ix (DOT) netcom.com> writes Really? Where does it say that? Now I'm confused. I thought cout was a global static object and would be destroyed on program termination. I also thought that note 209 might imply that it got flushed at this point (though I know it says underlying C stream). See 27.3 para 2 last sentence: The objects are not destroyed during program execution. Sure, but on program termination, wouldn't you agree that they are destroyed? Quote: And if you look at the section where footnote 209 is you will see that the dtors for static objects are called before the C streams are closed. Nothing about dtors being called latter and calling a dtor for cin, cout etc. without closing the stream to which it is attached might cause some problems. I realize that "Standard C++ IOStreams & Locales" is not the standard, but it gave me the impression that cout is associated with stdout. I would have thought, though I can't see it guaranteed in the standard, that when cout is destroyed, its data is written to stdout which is guaranteed to be flushed before termination. What am I missing? Dave [ 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 Jul 01, 2003 7:32 pm    Post subject: Re: Destructor not called when object is declared globally [email]dittoC (AT) ix (DOT) netcom.com[/email] (David Cattarin) wrote in message news:<30869baf.0306300657.49f2427a (AT) posting (DOT) google.com>... Quote: Francis Glassborow <francis.glassborow (AT) ntlworld (DOT) com> wrote in message news:<Uo2w+nDFlMicrosoft+EwZ6 (AT) robinton (DOT) demon.co.uk>... In message <30869baf.0306270733.75c7f444 (AT) posting (DOT) google.com>, David Cattarin <dittoC (AT) ix (DOT) netcom.com> writes Really? Where does it say that? Now I'm confused. I thought cout was a global static object and would be destroyed on program termination. I also thought that note 209 might imply that it got flushed at this point (though I know it says underlying C stream). See 27.3 para 2 last sentence: The objects are not destroyed during program execution. Sure, but on program termination, wouldn't you agree that they are destroyed? The standard says no. In practice, I guess it depends on your definition of "destroyed", the destructor isn't called, but of course, they are present in the process memory space, and once the process terminates, that memory space is deallocated. The important thing, I guess, is that there is no way a conforming C++ program can tell that they have disappeared. Quote: And if you look at the section where footnote 209 is you will see that the dtors for static objects are called before the C streams are closed. Nothing about dtors being called latter and calling a dtor for cin, cout Quote: etc. without closing the stream to which it is attached might cause some problems. I realize that "Standard C++ IOStreams & Locales" is not the standard, but it gave me the impression that cout is associated with stdout. I would have thought, though I can't see it guaranteed in the standard, that when cout is destroyed, its data is written to stdout which is guaranteed to be flushed before termination. That would be true, except that in the sense you are using the word here (that of the C++ standard), cout is never destroyed. -- 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, Tél. : +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 David Cattarin Guest Posted: Wed Jul 02, 2003 10:09 pm    Post subject: Re: Destructor not called when object is declared globally [email]kanze (AT) gabi-soft (DOT) fr[/email] wrote in message news:<d6652001.0307010039.7592aeb (AT) posting (DOT) google.com>... Quote: dittoC (AT) ix (DOT) netcom.com (David Cattarin) wrote in message news:<30869baf.0306300657.49f2427a (AT) posting (DOT) google.com>... Francis Glassborow <francis.glassborow (AT) ntlworld (DOT) com> wrote in message news:<Uo2w+nDFlMicrosoft+EwZ6 (AT) robinton (DOT) demon.co.uk>... In message <30869baf.0306270733.75c7f444 (AT) posting (DOT) google.com>, David Cattarin <dittoC (AT) ix (DOT) netcom.com> writes Really? Where does it say that? Now I'm confused. I thought cout was a global static object and would be destroyed on program termination. I also thought that note 209 might imply that it got flushed at this point (though I know it says underlying C stream). See 27.3 para 2 last sentence: The objects are not destroyed during program execution. Sure, but on program termination, wouldn't you agree that they are destroyed? The standard says no. In practice, I guess it depends on your definition of "destroyed", the destructor isn't called, but of course, they are present in the process memory space, and once the process terminates, that memory space is deallocated. The important thing, I guess, is that there is no way a conforming C++ program can tell that they have disappeared. [snip] I think my problem was that I was distinguishing between program execution and program termination. After going back and rereading the long definition of program execution (1.9), I can see that it includes program termination. So, yep, the standard definitely says (27.3) that cout et al are not destroyed. Thanks, Dave [ 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.