- Boost-users - lists.boost.org

opinions about a clean c++ ".net style" property system
by Isaac Lascasas 12 Mar '09

12 Mar '09

I have been strugling the past weeks designing a system to add ".net style" properties to my library. After some experiments like wrapping the class members inside member<type> m_member stuff I realized that I was wrong about what exactly a property is. So I rewrote everything again with the following idea in mind: A property is nothing more that a pack of some accessor functions and maybe some data. My objective was to support properties while keeping the class size the same and without any speed impact or weird requeriment in the get/set functions. The code is working right now but I think it can be more compact for the user, but unfortunately my c++ skills are insuficient to polish some things. Here is an example: This block shows how the system actually works: { rectangle_i r(1,2,3,4); size_i sz = get_property<size_i>(r, L"size"); get_property(sz, r, L"size"); set_property(r, L"size", size_i(10,20)); bool IsEmpty; get_property(IsEmpty, r, L"IsEmpty"); rectangle_i r2(10,1,1,1); get_property(r2, r, L"empty"); } And this section of a common rectangle class (ripped from .net api) shows how to register this "properties": template <class T_> class rectangle { // ... const point<T_>& get_location() const {return m_location;} void set_location(const point<T_>& value) {m_location = value;} static void initialize_property_pool(property_pool& out) { out.add<rectangle<T_>, point<T_>>(L"location", &rectangle<T_>::get_location, &rectangle<T_>::set_location); out.add<rectangle<T_>, size<T_>>(L"size", &rectangle<T_>::get_size, &rectangle<T_>::set_size); out.add<rectangle<T_>, T_>(L"height", &rectangle<T_>::get_height, &rectangle<T_>::set_height); out.add<rectangle<T_>, T_>(L"width", &rectangle<T_>::get_width, &rectangle<T_>::set_width); out.add<rectangle<T_>, T_>(L"bottom", &rectangle<T_>::get_bottom, &rectangle<T_>::set_bottom); out.add<rectangle<T_>, T_>(L"left", &rectangle<T_>::get_left, &rectangle<T_>::set_left); out.add<rectangle<T_>, T_>(L"right", &rectangle<T_>::get_right, &rectangle<T_>::set_right); out.add<rectangle<T_>, T_>(L"top", &rectangle<T_>::get_top, &rectangle<T_>::set_top); out.add_read_only<rectangle<T_>, bool>(L"IsEmpty", &rectangle<T_>::get_is_empty); out.add_static_read_only<rectangle<T_>, rectangle<T_>>(L"empty", &rectangle<T_>::get_empty); } // ... }; I want to know if there is a way of deducing the template parameters to reduce this verborreic initialize_property_pool function and to simplify things to the user and to replace "add_read_only" and "add_static_read_only" with just another pair of plain "add" functions. Any observation will be apreciated, thanks, Isaac Lascasas.

1 0

[thread] call_once question
by Agustín K-ballo Bergé 12 Mar '09

12 Mar '09

Is the following code supposed to work (instead of resulting on a stack overflow)? If so, what I'm doing wrong? I'm using Boost 1.37.0 on MS Visual Studio 2008, Windows. #include <boost/thread/once.hpp> boost::once_flag initialization_flag = BOOST_ONCE_INIT; void call_me_once() { boost::call_once( initialization_flag, call_me_once ); } int main( int, char*[] ) { boost::call_once( initialization_flag, call_me_once ); return 0; } Additionaly, - Can a once_flag be safelly used in a class or local static? - Is BOOST_ONCE_INIT guaranteed to statically initialize the flag? Agustín K-ballo Bergé.-

2 2

how to detect if a type is among a know list inside enable_if
by Claude Brisson 12 Mar '09

12 Mar '09

Hi, I need to enable only if a type T is among a specific known list. I tried to build an operator is_one_of< T, list of known types > condition that could be used within an enable_if (like others type_traits operators), like this: or< is_same< T, type1 >, or< is_same< T, type2>, ... > so I searched for an or< condition1, condition2> construct and didn't find anything... Thanks for your suggestions! Claude

2 1

[thread] thread_specific_ptr interface
by Agustín K-ballo Bergé 12 Mar '09

12 Mar '09

Is there a reason why thread_specific_ptr does not offer boolean evaluation as pointers do? Agustín K-ballo Bergé.-

1 0

Boost.date_time compilation error Debian GCC 4.3 Boost 1.35
by Joost Kraaijeveld 11 Mar '09

11 Mar '09

Hi, I cannot get this short program to compile with GCC 4.3.2: #include <iostream> #include <boost/date_time/gregorian/gregorian.hpp> #include <boost/date_time/gregorian/formatters.hpp> int main(int argc, char** argv){ boost::gregorian::date date = boost::gregorian::day_clock::local_day(); std::cout << boost::gregorian::to_sql_string_type(date) << std::endl; return 0; } The error is: g++ -O0 -g3 -Wall -c -fmessage-length=0 -MMD -MP -MF"Main.d" -MT"Main.d" -o"Main.o" "../Main.cpp" ../Main.cpp: In function ‘int main(int, char**)’: ../Main.cpp:18: error: no matching function for call to ‘to_sql_string_type(boost::gregorian::date&)’ I cannot find any solution anywhere hence my question: what am I doing wrong, and more important, how can I get it working? -- Groeten, Joost Kraaijeveld Askesis B.V. Molukkenstraat 14 6524NB Nijmegen tel: 024-3888063 / 06-51855277 fax: 024-3608416 web: www.askesis.nl

4 3

C++03 unique_ptr emulation
by Howard Hinnant 11 Mar '09

11 Mar '09

I've put up a new version of an C++03 emulated unique_ptr here: http://home.roadrunner.com/~hinnant/unique_ptr03.html This is a "boost-ized" emulation (uses boost tools and is in boost namespace), though I have reason to believe it may not work well on VC+ +05 unless the /Za option is used (I've tested only on gcc 4.0). I'm aware of Ion's interprocess version (http://www.boost.org/doc/libs/1_35_0/doc/html/interprocess/interprocess_sma… ), and thank Ion for his support over the years. This new version is not meant to replace Ion's version. It is simply my current best shot at emulating the unique_ptr behavior as specified in the C++0X CD1 draft. Booster's are welcome to take parts of this implementation and use them in places like interprocess/unique_ptr. Or to just use this unique_ptr directly. It consists of only one header: unique_ptr.hpp, and requires only boost headers, not boost cpp files. Basic documentation is included at the link. -Howard

5 8

Compiling probs with Filesystem in 1.38.0
by Loreene Dixon 11 Mar '09

11 Mar '09

I successfully used Boost 1.33.1 with VS 2005. Now I have the task of upgrading our software to VS 2008 ASAP and in so doing, I upgraded our Boost to 1.38.0. Apparently, Boost's Filesystem got revamped in 1.34.0 and the contents of exception.hpp got moved to path.hpp and into Interprocess (!) library's errors.hpp. We cannot call the error() routine anymore as it's no longer a member of filesystem_error. Here is an extract of our code (for Boost 1.33.1): -------------------------------------------------- #include <boost/filesystem/convenience.hpp> #include <boost/filesystem/exception.hpp> #include <boost/filesystem/operations.hpp> . . . vit::ViToolkitException& createViToolkitException(boost::filesystem::filesystem_error& e) { switch (e.error()) { case boost::filesystem::other_error: case boost::filesystem::path_error: case boost::filesystem::not_found_error: case boost::filesystem::not_directory_error: case boost::filesystem::already_exists_error: throw vit::BadPathException(e.what()); default: throw vit::GeneralException(e.what()); } } boost::filesystem::path createPath(const std::wstring& pathName) { try { return boost::filesystem::path(vit::wideStringToString(pathName), boost::filesystem::native); } catch (boost::filesystem::filesystem_error& e) { throw createViToolkitException(e); } } -------------------------------------------------- So basically we call Boost's methods and if an error occurs, we catch it and throw createViToolkitException where it examines the error to determine whether it's a bad path or general exception and throws a new appropriate error. When I attempted to compile Boost 1.38.0, my compiler errors were: File.cc File.cc(28) : error C2039: 'error' : is not a member of 'boost::filesystem::basic_filesystem_error<Path>' with [ Path=boost::filesystem::path ] File.cc(29) : error C2039: 'other_error' : is not a member of 'boost::filesystem' File.cc(29) : error C2065: 'other_error' : undeclared identifier File.cc(29) : error C2051: case expression not constant File.cc(30) : error C2039: 'path_error' : is not a member of 'boost::filesystem' File.cc(30) : error C2065: 'path_error' : undeclared identifier . . . I changed the #include <boost/filesystem/exception.hpp> to #include <boost/filesystem/path.hpp>. I could change all 'boost::filesystem::filesystem_error' references to 'boost::filesystem::basic_filesystem_error'. What can I do about the ' switch (e.error()) {' call since the error() call is no longer available? Is there a different method that I should call? And do I replace 'boost::filesystem::path_error' with 'boost::interprocess::path_error'? What do I need to do in order to recompile in Boost 1.38.0 and still have the same functionality? I inherited this code from our east coast developers and am somewhat a Boost newbie. Thanks!, Loreene

1 0

[Build/make tools] My strong interest in some 'boost-config' script...
by François Mauger 11 Mar '09

11 Mar '09

Dear boosters, As a programmer under Unix-like systems, I really appreciate when a library (say `mylib') provides some `mylib-config' script that helps its users to build compiler flags. Several industrial strength libraries have such a tool: GSL, MySQL and, in the field of particle physics software I work: CLHEP, ROOT, CERNLIB... Here is an example for the GSL (GNU Scientific Library): >>> $ gsl-config Usage: gsl-config [OPTION] Known values for OPTION are: --prefix show GSL installation prefix --libs print library linking information, with cblas --libs-without-cblas print library linking information, without cblas --cflags print pre-processor and compiler flags --help display this help and exit --version output version information An external CBLAS library can be specified using the GSL_CBLAS_LIB environment variable. The GSL CBLAS library is used by default. <<< I makes really easy to write rather portable makefiles using things like: CPPFLAGS+=$(shell gsl-config --cflags) LDFLAGS+=$(shell gsl-config --libs) Up to now and AFAIK it seems Boost does not provide such script that could be generated at build stage for any Unix-like system. Of course, I wrote my own script for my own usage (and for people who work with me). But at the end, I think it would be better to use an `official' tool. For my experience, it is also a strong argument to convince newbies to use Boost on top of standard C++, because it comes in some rather transparent way. So I have a few questions: - Is there such a script hidden in the Boost source tree? - If not, is there any plan to provide one? - Is someone else interested with it? Thank you very much for your comments. [hope this topic has not been addressed in the past! I apologize in advance for that!] regards frc [PS: I previously posted this mail within the wrong thread... sorry!] -- F.Mauger

4 4

[Interprocess] Deadlock when using interprocess_semaphore
by Derek Kivi 11 Mar '09

11 Mar '09

We are seeing a deadlock when making use of the interprocess_semaphore. In our case we have two running threads where the first thread is stuck waiting at a timed_wait() call while the second thread is stuck at a post() call. Also, a symptom we are seeing is that the lock value is set to 1 before both threads try to acquire the lock -- we think this is why the threads get stuck in the lock calls. We are running 1.36 at the moment. Have there been any fixes in 1.37 or 1.38 that might help us to avoid this problem? Thanks, Derek Here is what the stacks for each thread look like from the Visual C++ debugger: Thread1 ------- Singletons.dll!boost::interprocess::scoped_lock<boost::interprocess::int erprocess_mutex>::scoped_lock<boost::interprocess::interprocess_mutex>(b oost::interprocess::interprocess_mutex & m={...}) Line 78 + 0x27 bytes C++ Singletons.dll!boost::interprocess::interprocess_semaphore::timed_wait(c onst boost::posix_time::ptime & abs_time={...}) Line 55 + 0xc bytes C++ Singletons.dll!QSemaphoreImpl::Wait(const unsigned long timeoutMilliseconds=1000) Line 62 + 0xf bytes C++ Singletons.dll!QSemaphore::Wait(const unsigned long timeoutMilliseconds=1000) Line 24 C++ Singletons.dll!QUtilityThread::DefaultAction(quic::HpcSharedPtr<IThread, std::allocator> data={...}) Line 105 + 0x18 bytes C++ Thread2 ------- ntdll.dll!7c90e4f4() [Frames below may be incorrect and/or missing, no symbols loaded for ntdll.dll] ntdll.dll!7c90d1fc() kernel32.dll!7c8023f1() kernel32.dll!7c802455() Singletons.dll!boost::interprocess::winapi::sched_yield() Line 281 + 0xe bytes C++ Singletons.dll!boost::interprocess::detail::thread_yield() Line 51 + 0x8 bytes C++ Singletons.dll!boost::interprocess::interprocess_mutex::lock() Line 40 C++ Singletons.dll!boost::interprocess::interprocess_condition::notify(unsig ned long command=1) Line 50 C++ Singletons.dll!boost::interprocess::interprocess_condition::notify_one() Line 33 C++ Singletons.dll!boost::interprocess::interprocess_semaphore::post() Line 31 C++ Singletons.dll!QSemaphoreImpl::Post() Line 45 C++ Singletons.dll!QSemaphore::Post() Line 18 C++ Singletons.dll!QUtilityThread::RegisterTask(IBackgroundTask * pBackgroundTask=0x01002660) Line 54 C++ -- Derek Kivi Senior Software Developer QuIC Financial Technologies Inc. Office: +1 403 210 8282 Mobile: +1 403 863 5204 derek.kivi(a)quic.com Risk. Managed. www.quic.com Confidentiality Notice: The information transmitted is intended only for the person(s) or entity(ies) to which it is addressed and may contain confidential and/or privileged material. Any review, retransmission, dissemination, or other use of, or taking of any action in reliance upon this information by persons or entities other than the intended recipient is prohibited. If you received this email in error, please contact the sender and delete the email and any related material from any computer. ver. QuIC 0707

2 5

How to suppress vcvarsall.bat call?
by Peter Klotz 11 Mar '09

11 Mar '09

Hello When compiling Boost 1.38.0 with this command, bjam 3.1.17 insists on calling vcvarsall.bat before executing the actual build commands. bjam -q -sNO_COMPRESSION=1 --build-type=complete --abbreviate-paths --toolset=msvc-8.0 architecture=x86 address-model=32 However my environment already includes all necessary settings to execute cl, link,... So I would like to skip the call to vcvarsall.bat. Is this possible using a bjam command line argument? Regards, Peter.

3 4