- Boost - lists.boost.org

RE: [boost] shared_ptr, interface and dlls
by Stefan Seefeld 08 Dec '04

08 Dec '04

> > extern "C" InterfacePtr DLLCreate( void ) > > { > > return InterfacePtr( new InterfaceImpl ); > > } > > > > At the moment this code doesn't compile because VS7.1 complains that > > it a function with extern "C" linkage can't return a > > shared_ptr<Interface> instance. How could I implement the desired > > behavior? > > Either omit the extern "C" or, if you must use a C-style > interface, do this: > > extern "C" Interface* DLLCreate() > { > return new InterfaceImpl; > } > > extern "C" void DLLDestroy( Interface* p ) > { > delete p; > } > > and then construct an InterfacePtr( DLLCreate(), DLLDestroy > ). You'll need a > public virtual destructor in this case, though. When writing plugins under linux (i.e. with 'dlopen' etc.), I believe I have already used a function returning a std::auto_ptr as entry point for the loader. What is wrong with that ? Is your advice above simply due to a limitation with MSVC ? Regards, Stefan

2 1

shared_ptr, interface and dlls
by Dirk Gregorius 08 Dec '04

08 Dec '04

Hi, is it possible to return a shared_ptr to an interface from a dl?. I thought of the following what unfortunately doesn't work: // Interface.h - exe project class Interface { public: vitrtual void f( void ) = 0; virtual void g( void ) = 0; protected: ~Interface( void ) {} } typedef shared_ptr<Interface> InterfacePtr; InterfacePtr Create( void ) { // Load a special create function from the dll //.... return DLLCreate(); } // InterfaceImpl.h - dll project #include Interface class InterfaceImpl : public Interface { public: void f( void ) { /* ... */ }; void g( void ) { /* ... */ }; private: // No copy operations } extern "C" InterfacePtr DLLCreate( void ) { return InterfacePtr( new InterfaceImpl ); } At the moment this code doesn't compile because VS7.1 complains that it a function with extern "C" linkage can't return a shared_ptr<Interface> instance. How could I implement the desired behavior? Some other related questions regarding this topic: 1.) I link to the multithreaded C run-time dll - is this allowed? I read in an older post that only static CRT linking is allowed. 2.) Does the above code - loading the factory from the DLL as well and (hopefully) constructing the instance there - solve the boundary problems? I read some articles regarding this issue, but to be honest I haven't completly understood the problems that may occure in this case so far. Any help is greatly appreiciated.. -Dirk

2 1

[Boost-bugs] [ boost-Support Requests-1081415 ] isdigit, problem with namespace
by SourceForge.net 08 Dec '04

08 Dec '04

Support Requests item #1081415, was opened at 2004-12-08 07:41 Message generated for change (Tracker Item Submitted) made by Item Submitter You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=207586&aid=1081415&group_… Category: None Group: None Status: Open Resolution: None Priority: 5 Submitted By: Nobody/Anonymous (nobody) Assigned to: Nobody/Anonymous (nobody) Summary: isdigit, problem with namespace Initial Comment: Hi! In one of my projects I use the regex library, and it works great. But unfortunately it seems to mess up my namespaces, or something like that. I included #include "boost/regex.hpp". Using the boost library I am to call functions from the std namespace (memset, isdigit, ...) like ::isdigit(). Without the namespace-operator the compiler seems to have two functions called isdigit. Now it happens that I am to use another library along with regex (one of my company), which uses an isdigit() call (without namespace- operator), and which i must not alter. Honestly, I'm a little bit puzzled at all that the regex- library forces me to use the namespace-operator with functions from std-namespace. Is this the way it was ment, or did I somethig wrong? Thanks for your help, Markus ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=207586&aid=1081415&group_… ------------------------------------------------------- SF email is sponsored by - The IT Product Guide Read honest & candid reviews on hundreds of IT Products from real users. Discover which products truly live up to the hype. Start reading now. http://productguide.itmanagersjournal.com/ _______________________________________________ Boost-bugs mailing list Boost-bugs(a)lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/boost-bugs

1 0

[date_time] Sunpro wide test errors should be marked expected
by Caleb Epstein 08 Dec '04

08 Dec '04

-- Caleb Epstein caleb dot epstein at gmail dot com

2 1

[external_allocators] any interest?
by Thorsten Ottosen 08 Dec '04

08 Dec '04

Dear all, I have sometimes spoken with people here that had an interest in allocators that could be reset within containers. For example, some have wanted a allocator& container::get_allocator(); member function. One way to solve that problem could be a class like this: template< class Allocator > class external_allocator { public: // forward to Allocator* member }; And then it can be used like this typedef external_allocator< std::allocator<T> > ex_allocator; std::allocator<T> alloc; ex_allocator ex_alloc(alloc); std::vector<T,ex_allocator> vec(ex_alloc); and can now control the allocator from the outside. There seems to only be a small overhead with this approach. One usage scenario I can see would be to use a stack-based allocator on performance critical paths. (because we need the copy-semantics of the container's allocator to be shallow) If there is interest, I can also add the specified member function to the smart containers. or an appropriate member to external_allocator<>. Feedback is welcome -Thorsten

2 2

[ublas] matrix.hpp doesn't compile with Borland bcc32 5.6.4
by Nicola Musatti 08 Dec '04

08 Dec '04

Hallo, A translation unit that only includes #include <boost/numeric/ublas/matrix.hpp> from Boost 1.32.0 doesn't compile with Borland's bcc32 5.6.4 (BCB6 with Update 4), but it used to with Boost 1.31.0 . I get the following error messages: Build [C++ Error] traits.hpp(349): E2285 Could not find a match for 'std::abs(const _STL::complex<float>)' [C++ Error] traits.hpp(354): E2285 Could not find a match for 'std::sqrt(const _STL::complex<float>)' [C++ Error] traits.hpp(419): E2285 Could not find a match for 'std::abs(const _STL::complex<double>)' [C++ Error] traits.hpp(424): E2285 Could not find a match for 'std::sqrt(const _STL::complex<double>)' [C++ Error] traits.hpp(486): E2285 Could not find a match for 'std::abs(const _STL::complex<long double>)' [C++ Error] traits.hpp(491): E2285 Could not find a match for 'std::sqrt(const _STL::complex<long double>)' [C++ Error] vector.hpp(813): E2356 Type mismatch in redeclaration of 'T' [C++ Error] vector.hpp(804): E2344 Earlier declaration of 'T' [C++ Error] vector.hpp(819): E2356 Type mismatch in redeclaration of 'T' [C++ Error] vector.hpp(805): E2344 Earlier declaration of 'T' [C++ Error] vector.hpp(1030): E2356 Type mismatch in redeclaration of 'T' [C++ Error] vector.hpp(1022): E2344 Earlier declaration of 'T' Cheers, Nicola Musatti

1 0

Solaris+gcc regression errors should be marked expected
by Caleb Epstein 08 Dec '04

08 Dec '04

Sorry, I think I sent an empty message just before this one. Please ignore it. There are several test failures for date_time and other Boost libs on sunos5+gcc (see for example http://tinyurl.com/4y6tx) that are due to incomplete wide character support on that platorm. This situation appears to be similar to a number other configurations (e.g. win32+mingw) which have a similar or identical tests marked as expected failures in the test matrix. I think sunos5+gcc should be treated similarly. Here is the relevant section of boost/config/stdlib/libstdcpp3.hpp: #if !defined(_GLIBCPP_USE_WCHAR_T) && !defined(_GLIBCXX_USE_WCHAR_T) # define BOOST_NO_CWCHAR # define BOOST_NO_CWCTYPE # define BOOST_NO_STD_WSTRING # define BOOST_NO_STD_WSTREAMBUF #endif -- Caleb Epstein caleb dot epstein at gmail dot com

1 0

Name checking in filesystem
by Walter Landry 08 Dec '04

08 Dec '04

Greetings, I have come across an annoying problem with the filesystem library. I have wrapped filesystem with Swig so that I can call it from python. I had to rewrite the filesystem::path implementation a little because Swig can't handle nested classes. In any case, once I got that sorted out, I wanted to do something like p=path(".bar") which doesn't work because filenames starting with a period are not allowed by the default name checking routine. Argh. Unfortunately, I can't figure out how to change the default name checker from python. If I worked at it long enough, I could probably work it out. But I am sure that I don't want to have to run something like path.default_name_check(no_check) at the beginning of every 10 line script. It is a particular piece of functionality (name checking) that just gets in the way. I would wager that most people using boost::filesystem from scripts would not want to have to deal with it, since it is completely different from every other portable, cross platform filesystem API. So to make a long story short, I think the default for boost::filesystem should be not to check for "portable" paths. This is different from a position I took earlier, where I thought that it is ok if I only have to set it once at the beginning. That is acceptable for large programs, but not for small ones. Regards, Walter Landry wlandry(a)ucsd.edu

6 14

[review-results] Smart Container Library
by Pavol Droba 08 Dec '04

08 Dec '04

Hi, I hereby announce that the Smart Container library is ACCEPTED into Boost. Congratulations and thanks to Thorsten Ottosen for providing us with this library. Most of the reviewers were in favor of acceptance, but there were several requests for improvements. I have consulted them with Thorsten and he proposed to do following changes. I consider them sufficient for acceptance. 1. add support for boost.range in insert functions, constructors etc 2. use void* implementation 3. improve docs as discussed during the review 4. add support for null values: - remove preconditions - add functions is_null( size_t idx ); is_null( iterator ); 5. add more debug checks after access to ptr_begin(), ptr_end() 6. reconfigure policies: ptr_vector<T,clone_manager> becomes ptr_vector<T,clone_allocator = something, std::allocator<T*> = std::allocator<T*> > or perhaps ptr_vector<T,clone_allocator = something< std::allocator<T* > > 7. improve tests I also suggest, that before the inclusion of the library into the boost distribution, the library should be brought in the discussion to verify, that all issues above have been addressed. Best Regards, Pavol Droba Review Manager

2 1

[date_time] SunPRO patch for date_names_put facet
by Caleb Epstein 07 Dec '04

07 Dec '04

The following patch is required to build date_time using the Sun Workshop C++ compiler. I mentioned this previously in another thread, but I fear it has been lost. --- boost/date_time/date_names_put.hpp~ 2004-12-01 08:58:32.846228000 -0500 +++ boost/date_time/date_names_put.hpp 2004-12-01 08:59:38.334228000 -0500 @@ -52,6 +52,11 @@ typedef std::basic_string<charT> string_type; static std::locale::id id; + +#if defined (__SUNPRO_CC) && defined (_RWSTD_VER) + std::locale::id& __get_id (void) const { return id; } +#endif + void put_special_value(iter_type& oitr, special_value_enum sv) const { do_put_special_value(oitr, sv); -- Caleb Epstein caleb dot epstein at gmail dot com

2 1