C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  use of std::uncaught_exception()          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message John Torjo Guest Posted: Wed Oct 27, 2004 1:04 am    Post subject: use of std::uncaught_exception() Hi all, I'd like to know if you've used/relied of correct behavior of std::uncaught_exception in your code - that is in destructors (code shown below). Of course, simple examples always work (not of VC6, on other compilers:)) Thus, I'd like to ask you if you've used/relied on std::uncaught_exception() in commercial code (or put in another way, non-toy programs). Thanks. #include <iostream> struct test { test() { std::cout << "test" << std::endl; } void do_throw() { throw 1; } ~test() { std::cout << "~test" << std::endl; if ( !std::uncaught_exception() ) throw 2; } }; int main() { try { test o; o.do_throw(); } catch( int & i) { std::cout << "i= " << i << std::endl; } return 0; } // expected output: // test // ~test // i= 1 In particular, I want to know if I can throw an exception from a destructor in a way similar to shown above - and to know if modern compilers can take it. (There is a possibility that my destructor could throw. I want to throw an exception only if there's no exception thrown yet) Best, John John Torjo, Contributing editor, C/C++ Users Journal -- "Win32 GUI Generics" -- generics & GUI do mix, after all -- http://www.torjo.com/win32gui/ -- v1.5 - tooltips at your fingertips (work for menus too!) + bitmap buttons, tab dialogs, hyper links, lite html [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Bob Hairgrove Guest Posted: Wed Oct 27, 2004 1:59 pm    Post subject: Re: use of std::uncaught_exception() On 26 Oct 2004 21:04:54 -0400, [email]jtorjo (AT) yahoo (DOT) com[/email] (John Torjo) wrote: [snip] Quote: (There is a possibility that my destructor could throw. I want to throw an exception only if there's no exception thrown yet) Destructors which throw are evil. Read Scott Meyer's "More Effective C++", Item 11 (publ. by Addison-Wesley, ISBN 0-201-63371-X) titled: "Prevent exceptions from leaving destructors". -- Bob Hairgrove [email]NoSpamPlease (AT) Home (DOT) com[/email] [ 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: Thu Oct 28, 2004 2:06 pm    Post subject: Re: use of std::uncaught_exception() "John Torjo" <jtorjo (AT) yahoo (DOT) com> wrote Quote: Hi all, I'd like to know if you've used/relied of correct behavior of std::uncaught_exception in your code - that is in destructors (code shown below). <snip> Quote: In particular, I want to know if I can throw an exception from a destructor in a way similar to shown above - and to know if modern compilers can take it. (There is a possibility that my destructor could throw. I want to throw an exception only if there's no exception thrown yet) More to the point than item 11 in More Effictive C++ is item 19 in More Exceptional C++. Herb refers to your code (which I snipped) as the 'wrong solution.' Here are the relevant bits: "Why the Wrong Solution is Unsound: The first problem is that the idiom .... won't actually work as intended.... because it can end up using the path that doesn't throw even when it would be safe to throw [because the exception would be caught higher up]." "Why the Wrong Solution Is Immoral: ... it is always poor design to allow an operation to report the same error in two different ways....It complicates the interface and the semantics. And it makes the caller's life harder because the caller must be able to handle both flavors of error reporting." Of course, that doesn't answer the question whether the wrong solution has been used successfully in commercial code ;-) Quote: Best, John Jonathan [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Peter Koch Larsen Guest Posted: Thu Oct 28, 2004 2:14 pm    Post subject: Re: use of std::uncaught_exception() "John Torjo" <jtorjo (AT) yahoo (DOT) com> skrev i en meddelelse news:c638aac5.0410260717.1346a222 (AT) posting (DOT) google.com... Quote: Hi all, I'd like to know if you've used/relied of correct behavior of std::uncaught_exception in your code - that is in destructors (code shown below). std::uncaught_exception asks whether there is an uncaught exception, and I'm not sure that is the question you want to have answered. Most of the time, the interesting question is whether your exception will call terminate. You can not use uncaught_exception for that. Apart from that your question is rather platform-specific, and as I have never used that feature, i can't give you an answer. [snip] Quote: In particular, I want to know if I can throw an exception from a destructor in a way similar to shown above - and to know if modern compilers can take it. (There is a possibility that my destructor could throw. I want to throw an exception only if there's no exception thrown yet) My best advice is: don't throw. I am aware that there are occasions (such as closing of files) that could cause an error which it would be nice to tell about via an exception. In that case you could perhaps use an alternative function ("close"), that could throw. That principle is the best i can think of. Kind regards Peter Quote: Best, John John Torjo, Contributing editor, C/C++ Users Journal -- "Win32 GUI Generics" -- generics & GUI do mix, after all ^^^^^^^^^^^^^^^^^^ this library is just way cool, John. [snip} [ 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 Oct 29, 2004 1:44 am    Post subject: Re: use of std::uncaught_exception() Bob Hairgrove wrote: Quote: On 26 Oct 2004 21:04:54 -0400, [email]jtorjo (AT) yahoo (DOT) com[/email] (John Torjo) wrote: [snip] (There is a possibility that my destructor could throw. I want to throw an exception only if there's no exception thrown yet) Destructors which throw are evil. Read Scott Meyer's "More Effective C++", Item 11 (publ. by Addison-Wesley, ISBN 0-201-63371-X) titled: "Prevent exceptions from leaving destructors". Please notice that that article had been written before uncaught_exception() was introduced into the C++ standard (at least that's what footnote 4 in my edition suggests). Alberto [ 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 Oct 29, 2004 1:46 am    Post subject: Re: use of std::uncaught_exception() John Torjo wrote: Quote: Thus, I'd like to ask you if you've used/relied on std::uncaught_exception() in commercial code (or put in another way, non-toy programs). The most notable use of std::uncaught_exception() is in the C++ standard library itself! Specifically it's in the destructor of basic_ostream::sentry (27.6.2.3/4): "If ((os.flags() & ios_base::unitbuf) && !uncaught_exception()) is true, calls os.flush()." That's because os.flush() may throw, so the check is necessary to avoid the library call terminate() accidentally. Quote: In particular, I want to know if I can throw an exception from a destructor in a way similar to shown above - and to know if modern compilers can take it. You are allowed to throw from destructors and it's better to check uncaught_exception() before doing that. Whether throwing from destructors is good or bad coding style, I'm not going to tell you as I am pretty sure a flame war will soon start on such subject ;-) I tried gcc 3.3.3 and VC++ 7.1 and both of them correctly implement uncaught_exception(). IIRC older versions of VC++ had problems with that. Alberto [ 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: Fri Oct 29, 2004 2:15 am    Post subject: Re: use of std::uncaught_exception() Bob Hairgrove <invalid (AT) bigfoot (DOT) com> wrote Quote: On 26 Oct 2004 21:04:54 -0400, [email]jtorjo (AT) yahoo (DOT) com[/email] (John Torjo) wrote: [snip] (There is a possibility that my destructor could throw. I want to throw an exception only if there's no exception thrown yet) Destructors which throw are evil. Read Scott Meyer's "More Effective C++", Item 11 (publ. by Addison-Wesley, ISBN 0-201-63371-X) titled: "Prevent exceptions from leaving destructors". Yes, I'm aware of that. Still, I want my original question to be answered ;) Best, John John Torjo, Contributing editor, C/C++ Users Journal -- "Win32 GUI Generics" -- generics & GUI do mix, after all -- http://www.torjo.com/win32gui/ -- v1.5 - tooltips at your fingertips (work for menus too!) + bitmap buttons, tab dialogs, hyper links, lite html [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Bob Bell Guest Posted: Fri Oct 29, 2004 2:14 pm    Post subject: Re: use of std::uncaught_exception() [email]jtorjo (AT) yahoo (DOT) com[/email] (John Torjo) wrote in message news:<c638aac5.0410260717.1346a222 (AT) posting (DOT) google.com>... Quote: Hi all, I'd like to know if you've used/relied of correct behavior of std::uncaught_exception in your code - that is in destructors (code shown below). Of course, simple examples always work (not of VC6, on other compilers:)) Thus, I'd like to ask you if you've used/relied on std::uncaught_exception() in commercial code (or put in another way, non-toy programs). Some compilers can take it, some can't. But I don't ever write code this way. Others have given some good reasons why not. Here are a couple of others. If you've got to handle the problem somehow when you think you can't throw, why not handle it the same way even when you think you can throw? Wouldn't that be simpler for everyone? No one expects destructors to throw. For example: // Use "two-step" strong exception safety: void F() { // Step one: operations that might throw: A a; // ... // Step two: operations that don't throw: // ... } a.~A() is called in the non-throwing step; it would be quite surprising to have an exception thrown here, after the state of the system was updated. If you've got some kind of cleanup operation in your class that must be able to throw, make it an ordinary member function and require your clients to call it. If you really need to, you can add a flag to your class that tracks whether this cleanup function is called, so the destructor can assert() if an object is destroyed without having that function called. Yes, this is ugly, and yes, it's much better than a throwing destructor. Bob [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Hubert Schmid Guest Posted: Fri Oct 29, 2004 2:36 pm    Post subject: Re: use of std::uncaught_exception() Alberto Barbati <AlbertoBarbati (AT) libero (DOT) it> writes: Quote: I tried gcc 3.3.3 and VC++ 7.1 and both of them correctly implement uncaught_exception(). std::uncaught_exception() was broken (unusable) in GCC3 a long time. It is still broken in gcc 3.3.5 and it has been fixed in gcc 3.4. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14026 regards, Hubert [ 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: Sun Oct 31, 2004 11:00 am    Post subject: check_invariant (was Re: use of std::uncaught_exception()) [email]jtorjo (AT) yahoo (DOT) com[/email] (John Torjo) wrote in message news:<c638aac5.0410260717.1346a222 (AT) posting (DOT) google.com>... Quote: Hi all, I'd like to know if you've used/relied of correct behavior of std::uncaught_exception in your code - that is in destructors (code shown below). [...] Thanks for your answers. Here's what I want to do. I want to have a check_invariant class, which checks that an invariant is true when entering a function, and when exiting it. That is : void some_class::f() { check_invariant chk(this, &some_class::is_valid); ... // code } The invariant, in case it does not hold, it can choose to throw an exception. I don't want a debate whether that's a good thing or not - I just want to know if this works. Here's the code for it: // this is version 1.0 - anyway, I'm interested in // whether the destructor works as expected struct check_invariant : private boost::noncopyable { template<class type> check_invariant(type * self, void (type::*func)() const) : checker( boost::bind( boost::mem_fn(func), self)) { checker(); } check_invariant( void (*func)() ) : checker(func) { checker(); } ~check_invariant() { bool any_exc = std::uncaught_exception(); try { checker(); } catch(...) { if ( !any_exc) throw; } } private: boost::function<void()> checker; }; The idea is: The invariant will be checked when exiting the function, even if the function is exited by an exception. If it's exited by an exception, it lets that exception propage. Best, John John Torjo, Contributing editor, C/C++ Users Journal -- "Win32 GUI Generics" -- generics & GUI do mix, after all -- http://www.torjo.com/win32gui/ -- v1.5 - tooltips at your fingertips (work for menus too!) + bitmap buttons, tab dialogs, hyper links, lite html [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top J Lindrud Guest Posted: Sun Oct 31, 2004 11:04 am    Post subject: Re: use of std::uncaught_exception() [email]jtorjo (AT) yahoo (DOT) com[/email] (John Torjo) wrote in message news:<c638aac5.0410272346.8843a41 (AT) posting (DOT) google.com>... Quote: Bob Hairgrove <invalid (AT) bigfoot (DOT) com> wrote in message [snip] Destructors which throw are evil. Read Scott Meyer's "More Effective [snip] Yes, I'm aware of that. Still, I want my original question to be answered ;) Best, John I sure don't want to come across as evil... but FWIW here's a scenario I've come across. I have a macro called THROW, which is used like this: void func(int x, int y) { // ... THROW( std::runtime_error, "Something went wrong")(x)(y); } The result is the throwing of a std::runtime_error, with the given message, along with the variable names and values of the appended arguments, in this case x and y. So the exception message look like this: "Something went wrong: x=1, y=2," . The macro is implemented by constructing a temporary variable at the macro invocation site, and the actual throw statement is located in the destructor of the temporary variable. There's no way of getting around that, since I want the user to be able to append any number of expressions to the macro invocation, without requiring a special final function call to invoke the throw. The problem is that when an expression is appended to the macro invocation (the (x) or (y) in the example), it could trigger an exception while writing to a std::ostringstream etc, and then we certainly don't want to throw again while destroying the temporary, since that would terminate the program. Calling std::uncaught_exception() in the temporary's destructor comes in very useful to keep that from happening. std::uncaught_exception() doesn't give a precise answer to the question "is it safe to throw now?" (sometimes it's safe to throw regardless of what std::uncaught_exception() tells you), but if it returns false, at least you're guaranteed that a throw won't terminate the program. /Jarl [ 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.