- Boost-users - lists.boost.org

Suggestions/advice using boost::test
by Ahmed Badran 21 Jul '09

21 Jul '09

Hi, I want to use boost test in a particular way, I'm not sure if it's even possible, suggestions/advice are greatly appreciated. I have a shared library consisting of many classes. I would like to write my BOOST_AUTO_TEST_CASE at the end of the implementation of each class so as to encourage always writing tests. So this is what I've done, included <boost/test/unit_test.hpp> in the implementation file of each class, and wrote the test cases for each class at the end of the .cpp file for that class. In one of the .cpp files, I've stuck the following at the top to produce a canned 'main': #ifdef TEST #define BOOST_TEST_MAIN #include <boost/test/included/unit_test.hpp> #endif and compiled this particular file twice, so now I have a library without a main, and an executable with a main. The self-contained executable runs the tests perfectly as intended. The problem is that now the library is dependent on boost::test, any program that links against this library now needs to also link against the boost::test library (otherwise I get linker errors) which is _highly_ undesirable. It is perfectly fine however, if the tests get included/instantiated in the library but never get called, that's a perfectly valid solution. I also don't want to wrap the inclusion/code for the tests in ifdefs, since this will mean that the code (which is a lot) will have to be compiled twice and this process gets messy and tedious. Is there a way to accomplish this? Regards, Ahmed

1 0

[python][lambda] thin wrapper functions using boost.lambda
by Václav Šmilauer 21 Jul '09

21 Jul '09

Hi, I would like to use lambda functions to define thin wrapper functions in boost::python, like struct A{ void meth1(); }; [...] python::class_<A>("A").def("meth1",lambda::bind(&A::meth1,lambda::_1)); and I am getting "error: no matching function for call to 'get_signature [...]' " from the compiler. Reading boost::lambda documentation (http://www.boost.org/doc/libs/1_39_0/doc/html/lambda/using_library.html#lam…) """The basic rule is that a lambda function can be called with arguments of any types, as long as the lambda expression with substitutions performed is a valid C++ expression.""" Is there a way to enforce parameter types, so that boost::python can get the signature? Or, more generally, is it somehow possible to use such in-place functions to create the thin wrappers? Many thanks, Vaclav

3 3

variant<T&> with throwing initialization code.
by Zachary Turner 20 Jul '09

20 Jul '09

I have a couple of variants whose inner type are all references. For example, variant<a1&, a2&, ...> var1; variant<b1&, b2&, ...> var2; Because this variant will not be assignable due to the references, they need to be initialized in a constructor initializer list. However, it's possible for the function that creates these references in the first place to throw, and moreover the inner types might contain unmanaged resources that need to be properly cleaned up on a throw. Also the construction of var_k, might depend upon the successful construction of var_j where j<k (like for example I might need to use some member variable in one of the previously constructed objects to pass to one of the later objects to construct it). Is there a decent way to handle this? The problem is that if, say, something throws while trying to create var2, but var1 has already been created, then var1's unmanaged resources need to be properly freed. I tried to package all the variants together in a class such as typedef variant<a1&, a2&, ...> v1_type; typedef variant<b1&, b2&, ...> v2_type; struct package { package(v1_type& v1, v2_type& v2) : var1_(v1), var2_(v2) { } v1_type var1_; v2_type var2_; }; and then only have one instance of package in my class. Then, in the initializer list call package_(create_package()), and define create_package() as: package create_package() { optional<v1_type> v1; optional<v2_type> v2; try { v1 = create_v1(); v2 = create_v2(); return package(v1.get(), v2.get()); } catch(...) { } } but for some reason this fails to compile, saying that it's trying to call variant::operator= in the package constructor. Which doesn't make sense since I don't think it should be.

1 0

BGL: copy_graph does not work with bundled properties
by caligulaemperorofrome＠yahoo.com 20 Jul '09

20 Jul '09

Hi, copy_graph is not working with bundled property. See this minimal example program (40 lines) at http://tinyurl.com/lxh577. Just download make and execute - very easy. Thanks - Caligula

1 0

remove boost variable names from gcc compiled file
by ndgs＠lpemail.com 20 Jul '09

20 Jul '09

After I compile my file with g++ using the -s flag all that is striped out are my own variable names. I used xxd to look a the compiled output and "boost" is still all over the binary. How can I remove them? I used g++ with /usr/lib/librt.a and /usr/local/lib/libboost_thread-gcc43-mt.a I want to release this as a binary with no source code and I would prefer it if there wasn't boost this and boost that in the binary. Thanks, Piano

2 1

[phoenix] using dereference and increment operator
by Kim Kuen Tang 20 Jul '09

20 Jul '09

Hi all, the following piece of code compiled with boost 1.39: std::vector<double> u(11,1.0), w(11,3.0); std::vector<double>::const_iterator it=w.begin(); double value=(*ph::ref(it)++)(); But if i use the expression *ph::ref(it)++ as a functional in an algorithm, std::generate(u.begin(),u.end(),*phoenix::ref(it)++ ); the compilation fails with the following error: >c:\boost\boost_msvc_2008\include\boost-1_39\boost\fusion\container\vector\detail\vector_n.hpp(68) : error C2440: '<function-style-cast>' : cannot convert from 'const boost::phoenix::composite<EvalPolicy,EvalTuple>' to 'boost::fusion::vector_data1<Derived,T0>' 1> with 1> [ 1> EvalPolicy=boost::phoenix::post_increment_eval, 1> EvalTuple=boost::fusion::vector<boost::phoenix::reference<std::_Vector_const_iterator<double,std::allocator<double>>>,boost::fusion::void_,boost::fusion::void_,boost::fusion::void_,boost::fusion::void_,boost::fusion::void_,boost::fusion::void_,boost::fusion::void_,boost::fusion::void_,boost::fusion::void_> 1> ] 1> and 1> [ 1> Derived=boost::fusion::vector1<boost::phoenix::reference<std::_Vector_const_iterator<double,std::allocator<double>>>>, 1> T0=boost::phoenix::reference<std::_Vector_const_iterator<double,std::allocator<double>>> 1> ] 1> No constructor could take the source type, or constructor overload resolution was ambiguous I remember that the compilation was fine when using boost 1.38. ( I need to check this.) Here is the code that produces the errors: # include <algorithm> # include <vector> # include <boost/spirit/include/phoenix.hpp> using namespace boost; int main() { std::vector<double> u(11,1.0), w(11,3.0); std::vector<double>::const_iterator it=w.begin(); double value=(*phoenix::ref(it)++)(); it=w.begin(); std::generate(u.begin(),u.end(),*phoenix::ref(it)++ ); return 0; } With regards, Kim Tang

2 3

Definition of string in shared memory
by Samuel Gallard 20 Jul '09

20 Jul '09

Hello, I can't useful examples about how to store string in shared memory. I have to share a queue of Element objects (queue<Element> *) between processes. This is the signature of Element class : class Element { ... private : string dir; string file; ... } I used a shared_memory_object to use the queue. When I test it with thread, it is working fine. But when i test it with process, I obtained some segmentation fault because it can't retrieve the attributes for an Element object. it is normal because I am creating the Element object in one process the second can't access it. I have seen in boost documentation the template interprocess:basic_string. But I have no clue how to use it. I would like to have something like : typedef basic_string< string, segment> mystring class Element { ... private : mystring dir; mystring file; ... } Could you someone show me how to do that in the creation but also in the implementation. Thanks you in advance Regards Samuel

1 0

Locale and Boost.Serialization
by Luca Cappa 20 Jul '09

20 Jul '09

Hello, on the Windows platform, in my own application (built in unicode char econding) I am experiencing this problem with Boost 1.38: if I change the locale before loading an xml archive, then the application will hang in the loading of the archive file. I must add that this only occurs in a Release build on Windows Vista, but it never occurs on Windows XP. Also, if I run the application on Vista with the Compatibility Mode set to Windows XP SP2, it does work and does not hang while loading the archive. I tryed to get a stack trace with a ReleaseWithDebInfo on Windows Vista, here it is: Warning: stack unwind information unavailable. Following frames may be wrong. application!std::vector<double, std::allocator<double> > ::push_back+0xblabl1 application!std::vector<double, std::allocator<double> > ::push_back+0xblabl2 application!std::vector<double, std::allocator<double> > ::push_back+0xblabl3 application!boost::archive::detail::oserializer<boost::archive::xml_oarchive, boost::shared_ptr<TriggerUnit> > ::save_object_data+0x18b50 application!0xblabla Any hint what to look at to resolve this issue? Many thanks. Best Regards, Luca

1 0

Spirit 1.39 compilation times
by Avi Bahra 20 Jul '09

20 Jul '09

I am using boost spirit 1.39. On gcc the compilation times are acceptable. However on AIX with v10.1 compiler, it fails to even compile. After 24hrs waiting, I gave up. Ok I realise that the grammer needs to split. Is there any _small_ examples out there, that demonstrates this technique for a beginner ? In my case I have one grammer and ~70 rules. Take the following rule: rule<ScannerT, string_closure::context_t> identifier ; identifier = lexeme_d [ (alnum_p || ch_p('_')) >> *(alnum_p || ch_p('_')) ] [ identifier.name = construct_<std::string>(arg1, arg2) ] ; What is the technique for converting it into a grammer, that is separately compiled ? Also how would I reference it again, without including the header file, from another rule dotPath = ( str_p(".") >> +( str_p("/") >> identifier ) ) [ dotPath.name = construct_<std::string>(arg1, arg2) ] ; Any help would greatly appreciated. Best regards, Ta, Avi

2 1

[Asio] How do I dynamically select a port?
by Robert Jones 19 Jul '09

19 Jul '09

Hi All I'm new to Asio, and not a tcp/ip expert either, so if I'm approaching this all wrong then please tell me. I have two process on one machine, one of which streams a file to the other. In fact there may be several 'others', but one other will suffice to phrase the question. I need to dynamically select port numbers on which to stream the files to these other processes, based on whether that port number is available, and once the viability of using a particular port is established that port number can be communicated to the other process(es) by other means. So, the code sequence I have is this. At what point can I detect that use of this particular port number is ok? Is it appropriate to use from_string("127.0.0.1"), or should I use some other form for local sockets? Comments appreciated. using namespace boost; using asio :: ip :: tcp; unsigned short port = 2001; asio :: io_service io; tcp :: endpoint endpoint( asio :: ip :: address :: from_string( "127.0.0.1" ), port ); tcp :: acceptor acceptor( io, endpoint ); tcp :: iostream socket_stream; acceptor.accept( * socket_stream.rdbuf( ) ); Thanks, - Rob.

2 2