C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  'unreachable code' in a throw statement          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message Hovik Melikyan Guest Posted: Wed Oct 29, 2003 1:29 am    Post subject: 'unreachable code' in a throw statement I have a throw statement which looks like throw new X("You made a boo-boo"); where X is defined like class X { // ... X(const std::string&); // ... } MSVC7 gives a "unreachable code" warning pointing to the throw statememnt. The warning disappears as soon as I remove the operator new in the throw statement OR make the X constructor accept (const char*) instead of (const string&). So dynamic allocation and the dynamic string together confuse the ..NET compiler. I looked at the assembly code and found nothing that can be left out when an exception is thrown. Also, no memory leaks seem to occur. Can anyone explain if this warning message is erroneous or it really warns of a potential bug? Thanks, -- Hovik Melikyan [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Carl Daniel Guest Posted: Wed Oct 29, 2003 4:45 pm    Post subject: Re: 'unreachable code' in a throw statement Hovik Melikyan wrote: Quote: Can anyone explain if this warning message is erroneous or it really warns of a potential bug? I'd suggest posting a minimal complete compiland that produces the warning to microsoft.public.dotnet.languages.vc where someone will no doubt be able to determine exactly what's going on. You can post to this group on news.microsoft.com if your ISP doesn't carry it. -cd [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Ron Guest Posted: Thu Oct 30, 2003 3:25 pm    Post subject: Re: 'unreachable code' in a throw statement Quote: I have a throw statement which looks like throw new X("You made a boo-boo"); where X is defined like class X { // ... X(const std::string&); // ... } MSVC7 gives a "unreachable code" warning pointing to the throw statememnt. The warning disappears as soon as I remove the operator new in the throw statement OR make the X constructor accept (const char*) instead of (const string&). So dynamic allocation and the dynamic string together confuse the .NET compiler. I looked at the assembly code and found nothing that can be left out when an exception is thrown. Also, no memory leaks seem to occur. Can anyone explain if this warning message is erroneous or it really warns of a potential bug? I don't see what the compiler is complaining about. However, you have a leak. The runtime system copies the pointer to the X object you create, s.15.1(3-4), and passes it to any handlers. Afterward it throws away the copy, s.15.1(4), but nothing ever deletes the X object you created. Why not use throw X ("This will not leak"); instead? -Ron [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Hovik Melikyan Guest Posted: Fri Oct 31, 2003 2:56 am    Post subject: Re: 'unreachable code' in a throw statement "Ron" <rcrane (AT) ictv (DOT) com> wrote Quote: I don't see what the compiler is complaining about. However, you have a leak. The runtime system copies the pointer to the X object you create, s.15.1(3-4), and passes it to any handlers. Afterward it throws away the copy, s.15.1(4), but nothing ever deletes the X object you created. The catcher is responsible for deleting the object, of course. Anyway, it's been confirmed as a compiler bug already at microsoft.public.dotnet.languages.vc. Quote: Why not use throw X ("This will not leak"); instead? Well, sometimes you have a complex exception object and you don't want it to be copied multiple times while rewinding the stack. You gain some performance at the price of 'delete x' in the final catcher. Why not? -- Hovik Melikyan [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Thomas Richter Guest Posted: Sat Nov 01, 2003 9:56 am    Post subject: Re: 'unreachable code' in a throw statement Hi, Quote: Well, sometimes you have a complex exception object and you don't want it to be copied multiple times while rewinding the stack. You gain some performance at the price of 'delete x' in the final catcher. Why not? Well, if performance (!) is an issue, one shouldn't use exceptions in first place. Throwing an exception should be, errr, the exceptional case, right. (-; Stack unwinding takes its time. And passing a pointer is at least error prone, so I really don't see why this could be a good thing to do. So long, Thomas [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Volker Glave Guest Posted: Sat Nov 01, 2003 9:57 am    Post subject: Re: 'unreachable code' in a throw statement "Hovik Melikyan" <hovik (AT) melikyan (DOT) com> wrote Quote: "Ron" <rcrane (AT) ictv (DOT) com> wrote in message news:c668c1f8.0310291328.14284b95 (AT) posting (DOT) google.com... I don't see what the compiler is complaining about. However, you have a leak. The runtime system copies the pointer to the X object you create, s.15.1(3-4), and passes it to any handlers. Afterward it throws away the copy, s.15.1(4), but nothing ever deletes the X object you created. The catcher is responsible for deleting the object, of course. Anyway, it's been confirmed as a compiler bug already at microsoft.public.dotnet.languages.vc. Why not use throw X ("This will not leak"); instead? Well, sometimes you have a complex exception object and you don't want it to be copied multiple times while rewinding the stack. You gain some performance at the price of 'delete x' in the final catcher. Why not? Because the final catcher in general is not supposed to, and should not, and does not have knowledge about if the object was or was not created by "new" in the first place. (Except if he does. So do what you want to. If it's working it's good enough.) Volker Glave [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Ali Cehreli Guest Posted: Sat Nov 01, 2003 10:29 am    Post subject: Re: 'unreachable code' in a throw statement "Hovik Melikyan" <hovik (AT) melikyan (DOT) com> wrote Quote: "Ron" <rcrane (AT) ictv (DOT) com> wrote in message news:c668c1f8.0310291328.14284b95 (AT) posting (DOT) google.com... I don't see what the compiler is complaining about. However, you have a leak. The runtime system copies the pointer to the X object you create, s.15.1(3-4), and passes it to any handlers. Afterward it throws away the copy, s.15.1(4), but nothing ever deletes the X object you created. The catcher is responsible for deleting the object, of course. But the catcher can never be sure whether the object is allocated on the heap. What if someone writes throw X("..."); The catcher can still catch by 'X const *' but should not delete. Quote: Why not use throw X ("This will not leak"); instead? Well, sometimes you have a complex exception object and you don't want it to be copied multiple times while rewinding the stack. You gain some performance at the price of 'delete x' in the final catcher. Why not? If the catchers catch by 'X const &', there is no copy. There should be only one exception object constructed in the special memory area that is used for thrown objects. The standard idiom is to throw by value and catch by const reference. Scott Meyers's 'More Effective C++', item 13, is specifically on this subject. Ali [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Hovik Melikyan Guest Posted: Sun Nov 02, 2003 1:20 am    Post subject: throw new ... [Was: 'unreachable code' in a throw statement] "Ali Cehreli" <acehreli (AT) yahoo (DOT) com> wrote Quote: But the catcher can never be sure whether the object is allocated on the heap. What if someone writes throw X("..."); The catcher can still catch by 'X const *' but should not delete. 'X const *' won't catch it. Actually 'X const &' will, like you said (below). Quote: If the catchers catch by 'X const &', there is no copy. There should be only one exception object constructed in the special memory area that is used for thrown objects. Well, that works fine except some rather exotic situations when an exception object can live 'beyond' the scope of exception handling. I'm experimenting with some non-trivial error recovery methods and also with using C++ exceptions for returning from nested calls. Exceptions are intened for exceptional situations, that's true, but it's conventional. It's not obvious that exceptions can't be used for returning values from 'deeply' nested functions, if they can replace couple of branches and return's. For the sake of simplicity and readability of the source code. In this regard 'throw new ...' makes sense as long as 'return new ...' makes sense. I'd appreciate any opinions and/or references on this topic. Thanks, -- Hovik Melikyan [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Joshua Lehrer Guest Posted: Sun Nov 02, 2003 4:50 pm    Post subject: Re: 'unreachable code' in a throw statement [email]acehreli (AT) yahoo (DOT) com[/email] (Ali Cehreli) wrote in message news:<f0070508.0310311621.4bdf7cf7 (AT) posting (DOT) google.com>... Quote: But the catcher can never be sure whether the object is allocated on the heap. What if someone writes throw X("..."); The catcher can still catch by 'X const *' but should not delete. This is incorrect. Catch statements match similar to function calls. If you call a function passing in an X, it won't bind to an X*. Likewise, with exceptions, if you throw an instance, you can catch by copy/value or reference - if you throw a pointer, you must catch by pointer. joshua lehrer factset research systems [ 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.