C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  stack trace info in exceptions          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message jlom Guest Posted: Sat Aug 16, 2003 8:04 am    Post subject: stack trace info in exceptions I'm writing unmanaged code in VC++.NET 2003, and thought that the existence of std::exception (and subclasses thereof) was going to provide some access to the stack snapshot at the time the exception is thrown. However, there doesn't seem to be anything in those classes that allows me to do that. While I've realized that most likely I'll never be able to achieve the same basic functionality of Java exceptions's printStackTrace if I don't make symbolic info available (which is not going to happen in production releases), I'm still quite curious to understand how I could at least start approaching that goal. Thanks. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Pete Becker Guest Posted: Sat Aug 16, 2003 1:45 pm    Post subject: Re: stack trace info in exceptions jlom wrote: Quote: I'm writing unmanaged code in VC++.NET 2003, and thought that the existence of std::exception (and subclasses thereof) was going to provide some access to the stack snapshot at the time the exception is thrown. However, there doesn't seem to be anything in those classes that allows me to do that. While I've realized that most likely I'll never be able to achieve the same basic functionality of Java exceptions's printStackTrace if I don't make symbolic info available (which is not going to happen in production releases), I'm still quite curious to understand how I could at least start approaching that goal. Unlike Java, C++ does not put a debugger into your application. -- Pete Becker Dinkumware, Ltd. (http://www.dinkumware.com) [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Daniel Anderson Guest Posted: Sun Aug 17, 2003 12:00 am    Post subject: Re: stack trace info in exceptions I once wrote a class: FuncTrace, to do something similar to what you are looking for. In the simplest form the object constructor take a string as a parameter that it save for reporting. In the destructor you verify if the destruction is due to exception by checking uncaught_exception(). if so you can use the string saved in the constructor to build a stack strace. Now in every function, the first object constructed should be FuncTrace with the fonction name the parameter. Some compiler offers function names as a macro. Danderson "jlom" <noone (AT) nowhere (DOT) net> a écrit dans le message de news: [email]3F3E2000.7080200 (AT) nowhere (DOT) net[/email]... Quote: I'm writing unmanaged code in VC++.NET 2003, and thought that the existence of std::exception (and subclasses thereof) was going to provide some access to the stack snapshot at the time the exception is thrown. However, there doesn't seem to be anything in those classes that allows me to do that. While I've realized that most likely I'll never be able to achieve the same basic functionality of Java exceptions's printStackTrace if I don't make symbolic info available (which is not going to happen in production releases), I'm still quite curious to understand how I could at least start approaching that goal. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Eugene Gershnik Guest Posted: Sun Aug 17, 2003 11:48 am    Post subject: Re: stack trace info in exceptions jlom wrote: Quote: I'm writing unmanaged code in VC++.NET 2003, and thought that the existence of std::exception (and subclasses thereof) was going to provide some access to the stack snapshot at the time the exception is thrown. However, there doesn't seem to be anything in those classes that allows me to do that. It is possible on your particular compiler and platform. To provide a short answer here, you need to redefine or intercept the compiler-provided function used to throw exceptions. For more details ask this question on one of Microsoft's Visual C++ newsgroups or look at http://www.windevnet.com/documents/s=7547/win0212a/ Quote: While I've realized that most likely I'll never be able to achieve the same basic functionality of Java exceptions's printStackTrace if I don't make symbolic info available (which is not going to happen in production releases), I'm still quite curious to understand how I could at least start approaching that goal. You are right that it is not possible to achieve the _exact_ functionality of Java exceptions but from practical point of view all you need is to have all the information required to diagnose a problem. Eugene [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Aaron Bentley Guest Posted: Mon Aug 18, 2003 10:47 am    Post subject: Re: stack trace info in exceptions jlom wrote: Quote: While I've realized that most likely I'll never be able to achieve the same basic functionality of Java exceptions's printStackTrace if I don't make symbolic info available (which is not going to happen in production releases), I'm still quite curious to understand how I could at least start approaching that goal. In our project, we have a Context class and a ContextStack Singleton. As a variation on "Resource Acquisition Is Instantiation", the Context constructor registers the object with the ContextStack, and the destructor deregisters the object with the ContextStack. The Context object has no other operations. void randfunc() { Context cx1("randfunc()"); for (i=0; i<5; ++i) { Context cx2("Loop 1"); } } Since cx1 is created at the beginning of randfunc(), it stays in the stack for the duration of the function. cx2 is constructed and destroyed during each loop, so it is registered and destroyed 5 times. This is flexible, but it requires you to scatter Context objects like breadcrumbs throughout your code. Aaron -- Aaron Bentley www.aaronbentley.com [ 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: Tue Aug 19, 2003 12:03 am    Post subject: Re: stack trace info in exceptions Aaron Bentley <aaron.bentley (AT) utoronto (DOT) ca> wrote Quote: jlom wrote: While I've realized that most likely I'll never be able to achieve the same basic functionality of Java exceptions's printStackTrace if I don't make symbolic info available (which is not going to happen in production releases), I'm still quite curious to understand how I could at least start approaching that goal. In our project, we have a Context class and a ContextStack Singleton. As a variation on "Resource Acquisition Is Instantiation", the Context constructor registers the object with the ContextStack, and the destructor deregisters the object with the ContextStack. The Context object has no other operations. void randfunc() { Context cx1("randfunc()"); for (i=0; i<5; ++i) { Context cx2("Loop 1"); } } Since cx1 is created at the beginning of randfunc(), it stays in the stack for the duration of the function. cx2 is constructed and destroyed during each loop, so it is registered and destroyed 5 times. This is flexible, but it requires you to scatter Context objects like breadcrumbs throughout your code. May I suggest a way to make these even more useful? You can use these context objects to do debug printing along the way. You can pass a debug level to the constructor, and it only prints if some global debugging level exceeds the level of that stamp. Thus, if the global value is 0, no debug output is printed, and setting it to a high number causes all stamps to print. Second, you can have the constructor use a varargs signature, and work like printf: void randfunc() { Context cx1(1,"randfunc()"); for (i=0; i<5; ++i) { Context cx2(2,"Loop %d",i); } } Finally, you can set conditional breakpoints in your code by causing the context object's constructor to signal the debugger when a given context object is reached. We also use context-like objects to control the name of the currently running thread. That way, when a thread crashes, we know what the last context object was that was hit. joshua lehrer factset research systems NYSE:FDS [ 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: Tue Aug 19, 2003 12:09 am    Post subject: Re: stack trace info in exceptions In article <3F3E2000.7080200 (AT) nowhere (DOT) net>, jlom wrote: Quote: I'm writing unmanaged code in VC++.NET 2003, and thought that the existence of std::exception (and subclasses thereof) was going to provide some access to the stack snapshot at the time the exception is thrown. However, there doesn't seem to be anything in those classes that allows me to do that. While I've realized that most likely I'll never be able to achieve the same basic functionality of Java exceptions's printStackTrace if I don't make symbolic info available (which is not going to happen in production releases), If you're happy to release Java bytecode with symbolic information (not that you have a lot of choice), why not C++ code? Quote: I'm still quite curious to understand how I could at least start approaching that goal. I assume you want stack backtraces for uncaught exceptions, not for all exceptions. This should be possible but VC++ won't do it because it unwinds the stack while searching for an exception handler, so by the time it detects an unhandled exception this information is lost. If you want to get stack backtraces in can't-happen cases, use assert() not exceptions. This should abort the program without unwinding the stack. On a machine without a just-in-time debugger installed, the "Dr Watson" program should save a dump of the program state which can be sent to the developer. (On Windows 9x, Dr Watson must be started beforehand and will save only partial information.) You can load the dump file into Visual Studio and see the full program state, not just the stack trace. Caveats: 1. You need to build every release with symbolic debugging information and keep that information even if you don't distribute it. 2. This doesn't work so well with Windows 9x. 3. Optimisations result in an unintuitive ordering of operations and can obscure the call stack, so it may be hard to interpret a dump. (But the same applies to interactive debugging.) For information on Dr Watson, see: <http://support.microsoft.com/default.aspx?scid=kb;en-us;246084> <http://support.microsoft.com/default.aspx?scid=kb;en-us;275481> <http://support.microsoft.com/default.aspx?scid=kb;en-us;308538> Similarly under Unix, a failed assertion normally results in a core dump (dump file named "core" after an old memory technology). [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Shannon Barber Guest Posted: Tue Aug 19, 2003 12:13 am    Post subject: Re: stack trace info in exceptions jlom <noone (AT) nowhere (DOT) net> wrote Quote: I'm writing unmanaged code in VC++.NET 2003, and thought that the existence of std::exception (and subclasses thereof) was going to provide some access to the stack snapshot at the time the exception is thrown. However, there doesn't seem to be anything in those classes that allows me to do that. While I've realized that most likely I'll never be able to achieve the same basic functionality of Java exceptions's printStackTrace if I don't make symbolic info available (which is not going to happen in production releases), I'm still quite curious to understand how I could at least start approaching that goal. It's not as nice as a stack dump, but it's easy to pick-up the line, file, and code that generates the error using a macro: #define MKH_THROWONFAILURE(expr) { int error=0; if( (error=(expr)) != 0) throw MKH::System::Exception(MKH_ERROR(error), #expr); } #define MKH_ERROR(err) MKH::System::Error(err, __LINE__, __FILE__) struct Exception : std::exception { Exception(const Error& err, const char* code) { make_msg(err, desc); } //... } [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Markus Werle Guest Posted: Tue Aug 19, 2003 12:38 pm    Post subject: Re: stack trace info in exceptions jlom wrote: Quote: I'm writing unmanaged code in VC++.NET 2003, and thought that the existence of std::exception (and subclasses thereof) was going to provide some access to the stack snapshot at the time the exception is thrown. However, there doesn't seem to be anything in those classes that allows me to do that. While I've realized that most likely I'll never be able to achieve the same basic functionality of Java exceptions's printStackTrace if I don't make symbolic info available (which is not going to happen in production releases), I'm still quite curious to understand how I could at least start approaching that goal On linux with gdb as debugger one can use break __cxa_throw (must be the first command!) and then at the breakpoint the stack is still available via backtrace (until gdb supports "catch throw", which was promised to me by one of the developers and maybe is already in the newest version) I am pretty sure that similar functionalities are available with other debuggers on Your desaster platform (or run it through cygwin://gdb if possible). If not and if nothing hints at a solution to come then it is time to say hello to some new platform - but then You need to manage your code of course ... Markus -- Build your own Expression Template Library with Daixtrose! Visit http://daixtrose.sourceforge.net/ [ 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.