C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  when an inline function doesn't inline, why doesn't compile          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message Manas Guest Posted: Thu Aug 28, 2003 3:05 pm    Post subject: when an inline function doesn't inline, why doesn't compile hi, suppose i have the following files: one.cpp ---------------------------------------------- #include <iostream> using namespace std::cout inline void g(void) { cout<<"One cppn"; } void f(void); int main(void) { f(); g(); return 0; } ---------------------------------------------------- two.cpp ---------------------------------------------------- #include using namespace std::cout inline void g(void) { cout<<"two cppn"; } void f(void) { cout<<"F of two cppn"; g(); } --------------------------------------------------------- Now when i compile and run the code, i get the following output: F of two cpp One cpp One cpp Clearly the function is not inlined(actually gcc doesn't inline if the code is not optimized and by default the code is not optimized). Question is : If the compile can't inline a function, why doesn't it give a compile time warning for multiple function definition. Also, why does it take definition of g() in one.cpp , not that in two.cpp. thanks manas [ 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 Aug 29, 2003 8:34 am    Post subject: Re: when an inline function doesn't inline, why doesn't com In article <9635763f.0308272107.abddb99 (AT) posting (DOT) google.com>, Manas <bhatt_manas (AT) hotmail (DOT) com> writes Quote: suppose i have the following files: one.cpp ---------------------------------------------- #include using namespace std::cout That will not compile. Quote: inline void g(void) { cout<<"One cppn"; } void f(void); int main(void) { f(); g(); return 0; } ---------------------------------------------------- two.cpp ---------------------------------------------------- #include using namespace std::cout inline void g(void) { cout<<"two cppn"; } void f(void) { cout<<"F of two cppn"; g(); } --------------------------------------------------------- Now when i compile and run the code, i get the following output: F of two cpp One cpp One cpp No you do not. The code does not compile. Quote: Clearly the function is not inlined(actually gcc doesn't inline if the code is not optimized and by default the code is not optimized). Question is : If the compile can't inline a function, why doesn't it give a compile time warning for multiple function definition. Because the language explicitly requires that it does not do so. Quote: Also, why does it take definition of g() in one.cpp , not that in two.cpp. I think you need to check on the meaning of the inline keyword. It explicitly allows multiple definitions and it is up to the implementation to deal with the result. If the definition of an inline function is not the same in two TUs then you have broken the one definition rule and are in the land of undefined behaviour. -- Francis Glassborow ACCU 64 Southfield Rd Oxford OX4 1PA +44(0)1865 246490 All opinions are mine and do not represent those of any organisation [ 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: Fri Aug 29, 2003 8:44 am    Post subject: Re: when an inline function doesn't inline, why doesn't com [email]bhatt_manas (AT) hotmail (DOT) com[/email] (Manas) wrote in message news:<9635763f.0308272107.abddb99 (AT) posting (DOT) google.com>... Quote: hi, suppose i have the following files: one.cpp ---------------------------------------------- #include using namespace std::cout inline void g(void) { cout<<"One cppn"; } void f(void); int main(void) { f(); g(); return 0; } ---------------------------------------------------- two.cpp ---------------------------------------------------- #include using namespace std::cout inline void g(void) { cout<<"two cppn"; } void f(void) { cout<<"F of two cppn"; g(); } --------------------------------------------------------- Now when i compile and run the code, i get the following output: F of two cpp One cpp One cpp I'm going to take a stab at this one, and warn you, that I'm a bit sketchy in this area. I *think* that you have violated the one-definition rule. The fact that the code is inlined has nothing to do with it's linkage. You have two symbols named "g" with the same signature. You have asked the compiler to optimize the code by inlining it, but it still must make extern'ed symbols. I would say that you *should* get an error from your linker. However, on my system, I get no error, and I get the expected: F of two cpp two cpp One cpp Either way, the "correct" way is to hide "g" from the other translation units. The old way was to specify static linkage: static inline void g() { } The new way is to wrap it in an anonymous namespace: namespace { inline void g() { } } 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 White Wolf Guest Posted: Fri Aug 29, 2003 11:21 am    Post subject: Re: when an inline function doesn't inline, why doesn't com Manas wrote: [SNIP] Quote: Clearly the function is not inlined(actually gcc doesn't inline if the code is not optimized and by default the code is not optimized). Question is : If the compile can't inline a function, why doesn't it give a compile time warning for multiple function definition. Also, why does it take definition of g() in one.cpp , not that in two.cpp. You have defined two non-static functions with different body and exactly the same name. It does not matter a bit if they are inline or not. As long as they are non-static (and not hidden in an unnamed namespace) you have just violated the ODR (One Definition Rule) and it is undefined behaviour. Your compiler can do whatever it wants to do and it does not even need to explain it (in a documentation etc.). -- WW aka Attila [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Ulrich Eckhardt Guest Posted: Fri Aug 29, 2003 11:28 am    Post subject: Re: when an inline function doesn't inline, why doesn't com Manas wrote: Quote: one.cpp inline void g(void) { cout<<"One cppn"; } two.cpp inline void g(void) { cout<<"two cppn"; } Violation of the 'one definition rule'. Undefined behaviour, anything can happen. Anyways, some compilers do warn and inline is not mandatory to the compiler. Uli -- Questions ? see C++-FAQ Lite: http://parashift.com/c++-faq-lite/ first ! [ 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: Sat Aug 30, 2003 10:12 am    Post subject: Re: when an inline function doesn't inline, why doesn't com Francis Glassborow <francis (AT) robinton (DOT) demon.co.uk> wrote [...] Quote: No you do not. The code does not compile. The code should not compile, according to the standard. I've got lots of code that should not compile, according to the standard, but does. (Not intentionally, of course, but I was writing code before there was a standard.) In this case, it's not a case of accepting prestandard code, but you still can't exclude a buggy compiler. Quote: Clearly the function is not inlined(actually gcc doesn't inline if the code is not optimized and by default the code is not optimized). Question is : If the compile can't inline a function, why doesn't it give a compile time warning for multiple function definition. Because the language explicitly requires that it does not do so. Almost. The standard doesn't say anything about warnings, so a warning would be permitted. A compile time warning would be very, very difficult, since it would require the compiler to know files what I will later link. A link time warning would be quite possible, however. In his particular case, of course, even a link time error is acceptable. Since he violated the one definition rule, the system can do whatever it wants. IMHO, a link time error or warning would be a good thing in such cases. And technically, it shouldn't be too difficult to do, except in portable compilers which use the native linker (e.g. EDG, g++): just stuff an MD5 hash of whatever comes under the one definition rule in the object file, and check that all instances have the same hash. Why no one has bothered to do it to date is beyond me. -- 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 Andrey Tarasevich Guest Posted: Sat Aug 30, 2003 1:21 pm    Post subject: Re: when an inline function doesn't inline, why doesn't com Manas wrote: Quote: ... Question is : If the compile can't inline a function, why doesn't it give a compile time warning for multiple function definition. ... Why should it? And why do you care? The C++ standard explicitly states that inline functions are allowed to have multiple definitions in multiple translation units (at most one per translation unit, of course). Just read the ODR- and inline-related parts of the standard. -- Best regards, Andrey Tarasevich Brainbench C and C++ Programming MVP [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Jeff Mirwaisi Guest Posted: Sat Aug 30, 2003 11:43 pm    Post subject: Re: when an inline function doesn't inline, why doesn't com Stroustrup:TC++PL 9.2.3 That is, two definitions of a class, template, or inline function are accepted as examples of the same unique definition if and only if [1] they appear in different translation units, and [2] they are tokenfortoken identical, and [3] the meanings of those tokens are the same in both translation units. and 3.2 para 3 iso 14882 ... An inline function shall be defined in every translation unit in which it is used. 3.2 para 5 iso 14882 There can be more than one definition of a class type (clause 9), enumeration type (7.2), inline function with external linkage (7.1.2), class template (clause 14), nonstatic function template (14.5.5), static data member of a class template (14.5.1.3), member function template (14.5.1.1), or template specialization for which some template parameters are not specified (14.7, 14.5.4) in a program provided that each definition appears in a different translation unit, and provided the definitions satisfy the following requirements. ... Therefore: <foo.cpp> inline void f() {} <bar.cpp> inline void f() {} is legal and defined by the ODR -Jeff Mirwaisi [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Andrey Tarasevich Guest Posted: Sun Aug 31, 2003 8:01 am    Post subject: Re: when an inline function doesn't inline, why doesn't com Andrey Tarasevich wrote: Quote: ... Question is : If the compile can't inline a function, why doesn't it give a compile time warning for multiple function definition. ... Why should it? And why do you care? The C++ standard explicitly states that inline functions are allowed to have multiple definitions in multiple translation units (at most one per translation unit, of course). Just read the ODR- and inline-related parts of the standard. ... Although the problem with your code is not related to actual inlining or non-inlining. You made two non-matching definitions of the same inline function with external linkage in two different translation units. That violates ODR for inline functions. -- Best regards, Andrey Tarasevich Brainbench C and C++ Programming MVP [ 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: Sun Aug 31, 2003 12:22 pm    Post subject: Re: when an inline function doesn't inline, why doesn't com In article <Xns93E747412ABCFjeffmirwaisiyahoocom (AT) 207 (DOT) 217.77.202>, Jeff Mirwaisi <jeff_mirwaisi (AT) yahoo (DOT) com> writes Quote: Therefore: foo.cpp inline void f() {} bar.cpp inline void f() {} is legal and defined by the ODR But inline void f(){NULL} would break the ODR. We had immense difficulty with defining this aspect of the ODR because we could find no set of rules that allowed all clearly functionally equivalent definitions or even all definitions that generated the same code that did not also let in cases which were clearly breaches of the ODR. The most uncomfortable aspect was the recognition that token-for-token equivalence was not sufficient because the local file context could modify the meaning of a token. One implementation (Metaware High C++) did diagnose some ODR breaches at link time. If two definitions of the 'same' function were not byte for byte identical in the object files being linked it issued a diagnostic. -- Francis Glassborow ACCU 64 Southfield Rd Oxford OX4 1PA +44(0)1865 246490 All opinions are mine and do not represent those of any organisation [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Daveed Vandevoorde Guest Posted: Sun Aug 31, 2003 5:18 pm    Post subject: Re: when an inline function doesn't inline, why doesn't com Francis Glassborow <francis (AT) robinton (DOT) demon.co.uk> wrote: [...] Quote: Question is : If the compile can't inline a function, why doesn't it give a compile time warning for multiple function definition. Because the language explicitly requires that it does not do so. I don't think it does. First, even if the program were valid, an implementation is allowed to issue diagnostics as long as it accepts the program. (E.g., the EDG front end has an option to issue such a diagnostic if its simple inliner is enabled but unable to inline an inline function.) Second (and ignoring the syntactical error you correctly identified), this code violates the one definition rule by providing two different definitions of an inline function. That results in undefined behavior and therefore a compiler is allowed to issue warning, hard errors, or anything else for this. (And again, the EDG front end can be configured to issue such errors.) Quote: Also, why does it take definition of g() in one.cpp , not that in two.cpp. Many compilers issue the non-inlined copy in a special object file section identified by name (e.g., the mangled name of the function). When the linker sees several such sections with the same name, it discards all but one of them. The linker does not check if they're identical because small differences in the code-generation state can result in harmless differences from one translation unit to another. Try changing the order of the files on the command line: I suspect that might actually change the output. Daveed [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Thomas Mang Guest Posted: Mon Sep 01, 2003 1:11 am    Post subject: Re: when an inline function doesn't inline, why doesn't com Francis Glassborow schrieb: Quote: I think you need to check on the meaning of the inline keyword. It explicitly allows multiple definitions and it is up to the implementation to deal with the result. If the definition of an inline function is not the same in two TUs then you have broken the one definition rule and are in the land of undefined behaviour. Say I put the definition of a simple function into a header file: in foo.h: void foo() {std::cout << "foo() called";} The header will be #included by various files, so foo appears in different translation units. Is it now compulsory to declare foo() as inline (by prefixing it with the 'inline' - keyword), or is the definition as given above (without 'inline') OK? regards, Thomas [ 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: Mon Sep 01, 2003 7:05 pm    Post subject: Re: when an inline function doesn't inline, why doesn't com In article <3F51CC1A.E6634C0C (AT) unet (DOT) univie.ac.at>, Thomas Mang <a9804814 (AT) unet (DOT) univie.ac.at> writes Quote: Say I put the definition of a simple function into a header file: in foo.h: void foo() {std::cout << "foo() called";} The header will be #included by various files, so foo appears in different translation units. Is it now compulsory to declare foo() as inline (by prefixing it with the 'inline' - keyword), or is the definition as given above (without 'inline') OK? If you put it in a header you must either make it inline or use some mechanism that allows multiple uses of a name (such as putting it in an unnamed namespace or qualifying it as static). The basic point is that you put things in header files that need to be available across TUs. In general those are declarations of names + definitions of types. If you want to make inlining a function easy for the implementation (modern implementations are becoming increasingly adept at inlining small functions at link time) it has to go in a header. There is no other motive for putting the definition of an ordinary function in a header file other than not understanding the C and C++ mechanism for separation of declaration from definition. -- Francis Glassborow ACCU 64 Southfield Rd Oxford OX4 1PA +44(0)1865 246490 All opinions are mine and do not represent those of any organisation [ 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.