- Boost-users - lists.boost.org

[pointer_container] and std::back_inserter
by Jeff Flinn 28 Mar '08

28 Mar '08

Are pointer_container's intended to be compatible with std::back_inserter? I've also tried calling insert and assign. I've tried: #include <boost/ptr_container/ptr_vector.hpp> #include <iterator> #include <vector> #include <memory> int main() { boost::ptr_vector<int> int_ptrs; *std::back_inserter(int_ptrs) = new int(123); *std::back_inserter(int_ptrs) = std::auto_ptr<int>(new int(456)); std::vector<int*> ptrs(1, new int(123)); int_ptrs.insert(int_ptrs.end(), ptrs.begin(), ptrs.end()); return 0; } Thanks, Jeff

3 5

Boost Pool - error: segmentation fault
by Joost Meijles 28 Mar '08

28 Mar '08

Hi All, I am trying to use the Boost Pool library, more specifically I am trying to use the singleton pool interface. Everything seems to work fine, till the moment I use it to create an object from a class containing a (STL) map, which results in a segmentation fault. Does anyone know why this happens, or how to work around it? What I find also strange is that the constructor and destructor of the class doesn't seem to be called when using the pool. Is this normal behavior? Please find below the example that causes the segmentation fault (and the not calling of the constructor/destructor issue). Thanks a lot in advance. Kind regards, Joost ========================================== class MyClass { public: int x; vector y; map z; }; int main() { typedef boost::singleton_pool myPool; MyClass * const t = (MyClass *) myPool::malloc(); t->x = 1; t->y.push_back(2); t->z.insert(make_pair(3,3));//causes segmentation fault myPool::purge_memory(); } ========================================== _________________________________________________________________ Probeer Live Search: de zoekmachine van de makers van MSN! http://www.live.com/?searchOnly=true

4 5

Build CSR graph from a file in graphviz format
by Johan Oudinet 28 Mar '08

28 Mar '08

Hi, I read the csr-example.cpp that explain how to use compressed_sparse_row_graph and output it in graphviz format (.dot), and I'd like building a csr graph from a file in DOT format. So I tried using read_graphviz but actually it doesn't work with CSR graph (see below for my code). Does someone successfully build a CSR graph from a DOT file ? Regards, //===== csr.cc ===== # include <iostream> # include <string> # include <fstream> # include <boost/graph/compressed_sparse_row_graph.hpp> # include <boost/graph/iteration_macros.hpp> # include <boost/graph/graphviz.hpp> struct Node { std::string name; }; int main () { typedef boost::compressed_sparse_row_graph<boost::directedS, Node> graph_t; // Try to read the CSR graph g from foo.dot graph_t g; std::ifstream in("foo.dot", std::ios::in); boost::dynamic_properties dp; dp.property ("node_id", boost::get (&Node::name, g)); bool status = read_graphviz(in, g, dp, "node_id"); // doesn't work, see below for the error message BGL_FORALL_VERTICES (v, g, graph_t) std::cout << g[v].name << std::endl; BGL_FORALL_EDGES (e, g, graph_t) std::cout << g[source (e, g)].name << " -> " << g[target (e, g)].name << std::endl; } //================ The error message with g++ 4.1.3: /usr/include/boost/graph/graphviz.hpp:721: error: conversion from 'boost::csr_edge_descriptor<unsigned int, unsigned int>' to non-scalar type 'std::pair<boost::csr_edge_descriptor<unsigned int, unsigned int>, bool>' requested <8==== foo.dot ===== digraph { 0 -> 3; 0 -> 4; 1 -> 3; 1 -> 4; 2 -> 3; 2 -> 4; 3 -> 4; } ================= -- Johan Oudinet

1 0

nondynamic: A utility class that promotes safe programming
by Daniel Lidström 28 Mar '08

28 Mar '08

Hi there! (I recently posted a similar message to microsoft.public.vc.language, but the response was so bad I thought I would post here instead.) I have created a utility class that I think is useful, called nondynamic. The purpose is sort of like boost::noncopyable, but I disallow any dynamic/ pointer usages. For example, & and -> are disabled. Operator new/delete as well. Let me show nondynamic: // all members are private class nondynamic { // disallow operator new static void* operator new(std::size_t); static void* operator new(std::size_t, const std::nothrow_t&) throw(); static void* operator new[](std::size_t); static void* operator new[](std::size_t, const std::nothrow_t&) throw(); // disallow operator delete static void operator delete(void*){} static void operator delete(void*, const std::nothrow_t&) throw(){} static void operator delete[](void*){} static void operator delete[](void*, const std::nothrow_t&) throw(){} // disallow address-of const nondynamic* operator&() const; /* */ nondynamic* operator&() /* */; // disallow -> const nondynamic* operator->() const; /* */ nondynamic* operator->() /* */; }; Using this class is as simple as using boost::noncopyable: struct DataIO : private nondynamic { ... }; Now you are only able to use DataIO as a value type. All other usages are disabled. I think this promotes safe programming, in that pointer usage is minimized (it seems with pointers, new/delete/null is never far away). Let me illustrate how this class can be used most effectively. I thought it would be nice to use nondynamic when you have a class hierarchy, and a factory method. Here's the base class: struct IDataIO { virtual void Write(const std::string& str) = 0; virtual void Read(char* buffer, std::size_t size) = 0; }; typedef boost::shared_ptr<IDataIO> IDataIOPtr; A sample subclass: struct SerialDataIO : IDataIO { void Write(const std::string&) { std::cout << __FUNCTION__ << std::endl; } void Read(char*, std::size_t) { std::cout << __FUNCTION__ << std::endl; } }; Here's the wrapper class that inherits from nondynamic: struct DataIO : IDataIO, private nondynamic { DataIO(IDataIOPtr dataIO) : mDataIO(dataIO) {} void Write(const std::string& str) { mDataIO->Write(str); } void Read(char* buffer, std::size_t size) { mDataIO->Read(buffer, size); } private: IDataIOPtr mDataIO; }; And finally, the factory: DataIO CreateDataIO(int i) { if( i==0 ) return IDataIOPtr(new SerialDataIO); else throw std::runtime_error("Could not create DataIO"); } Now the clients of the class hierarchy uses the wrapper, and are completely isolated from pointers: DataIO dataIO = CreateDataIO(0); dataIO.Write("Hello"); char buffer[1024]; dataIO.Read(buffer, sizeof(buffer)); It should be impossible to misuse DataIO. What do you think about nondynamic? Is it implemented correctly? Named appropriately? Useful at all? Any comments gladly accepted :-) Thanks in advance! -- Daniel

2 1

[move] How to use BOOST_LVALUE_ASSIGN?
by Terry G 28 Mar '08

28 Mar '08

How are BOOST_LVALUE_ASSIGN and BOOST_LVALUE_ASSIGN2 used? How do I know which one to use? Is David Abrahams move.hpp the best library to use for move semantics? I noticed something similar under interprocess. Is there a version of unique_ptr that works with Dave's move.hpp? If not, does it make sense to create one? Its been several years since this was a topic of discussion here. Unfortunately, I didn't need to use it then. Now I do, and C++0X is still over the horizon for a few years for embedded systems. I'm not sure, but I think the compiler I'll be using is GCC 3.4 or later. terry

1 0

intrusive_ptr and incomplete types
by gast128 27 Mar '08

27 Mar '08

Dear all. I was googling about incomplete types and smart pointers but couldn't find a definitive statement about intrusive pointers. >From the header file of intrusive_ptr it seems that it can deal with incomplete types. The constructor and destructor use the 'intrusive_ptr_add_ref' or 'intrusive_ptr_release'. If these are not invoked, use of an intrusive_ptr with incomplete types should be ok. So it seems that below example will compile: //foo.h: #include <boost/intrusive_ptr.hpp> class A; class Foo { public: Foo(); ~Foo(); //leave the dtor out, and it won't work boost::intrusive_ptr<A> m_ptr; }; //bla.cpp: #include "foo.h" void F() { Foo foo; } Is this correct? Or is my (non confirming?) compiler giving me the wrong impression. A good article about incomplete types is 'The Standard Librarian: Containers of Incomplete Types' from Austern. He states that member function which are not referenced in the client code are not checked for incomplete types. For intrusive_ptr this would be the constructor and destructor. wkr, me

2 2

Help linking to Boost in Xcode 3 debug
by Jeremy Conlin 27 Mar '08

27 Mar '08

I am using Boost program_options library. I can successfully compile my code under "Release" mode in Xcode, but under "Debug" mode there are some issues with linking. The compiler says it can't find vtables for some of the program_options stuff (see error messages pasted below). I have tried to compile my code and Boost with the same architecture (i386 only) and telling the linker to find the debug library with the linker flag OTHER_LDFLAGS = -lboost_program_options-d-1_34_1 Can someone help me to figure out why this won't link? I'm sorry if the information is too general. Thanks, Jeremy * "boost::program_options::validate(boost::any&, __gnu_debug_def::vector<std::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::basic_string<char, std::char_traits<char>, std::allocator<char> > > > const&, bool*, int)", referenced from:* * boost::program_options::typed_value<bool, char>::xparse(boost::any&, __gnu_debug_def::vector<std::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::basic_string<char, std::char_traits<char>, std::allocator<char> > > > const&) constin ArnoldiVsPower.o* * "boost::program_options::value_semantic_codecvt_helper<char>::parse(boost::any&, __gnu_debug_def::vector<std::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::basic_string<char, std::char_traits<char>, std::allocator<char> > > > const&, bool) const", referenced from:* * vtable for boost::program_options::typed_value<bool, char>in ArnoldiVsPower.o* * vtable for boost::program_options::typed_value<unsigned long, char>in ArnoldiVsPower.o* * vtable for boost::program_options::typed_value<double, char>in ArnoldiVsPower.o* * vtable for boost::program_options::typed_value<int, char>in ArnoldiVsPower.o* * "boost::program_options::detail::cmdline::cmdline(__gnu_debug_def::vector<std::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::basic_string<char, std::char_traits<char>, std::allocator<char> > > > const&)", referenced from:* * boost::program_options::basic_command_line_parser<char>::basic_command_line_parser(int, char**)in ArnoldiVsPower.o* *ld: symbol(s) not found* *collect2: ld returned 1 exit status*

1 0

Serializing Boost.Tuple
by Sohail Somani 27 Mar '08

27 Mar '08

Hi, Just thought I would share some code that serializes Boost.Tuple using Boost.Serialization: #include <boost/tuple/tuple.hpp> #include <boost/serialization/nvp.hpp> #include <boost/preprocessor/repetition.hpp> namespace boost { namespace serialization { #define GENERATE_ELEMENT_SERIALIZE(z,which,unused) \ ar & boost::serialization::make_nvp("element",t.get< which >()); #define GENERATE_TUPLE_SERIALIZE(z,nargs,unused) \ template< typename Archive, BOOST_PP_ENUM_PARAMS(nargs,typename T) > \ void serialize(Archive & ar, \ boost::tuple< BOOST_PP_ENUM_PARAMS(nargs,T) > & t, \ const unsigned int version) \ { \ BOOST_PP_REPEAT_FROM_TO(0,nargs,GENERATE_ELEMENT_SERIALIZE,~) \ } BOOST_PP_REPEAT_FROM_TO(1,10,GENERATE_TUPLE_SERIALIZE,~); }} It does not handle 0-element tuples (not sure that it should, actually!) Kudos to the insanity that produced Boost.Preprocessor. -- Sohail Somani http://uint32t.blogspot.com

2 6

object relational mapping library
by Pavlo Korzhyk 27 Mar '08

27 Mar '08

Dear List, I use Boost.Serialization in my projects. It's very useful, powerful, and cool. But it doesn't meet some of my data persistence needs. Thus, I present the C++ library for transparent object-relational mapping to SQLite3. It works similar to Java Hibernate, and its API is inspired by Boost.Serialization. http://code.google.com/p/hiberlite/ I hope some of you will find this work useful. Comments on design, usability, proposals, and contributions will be highly appreciated! Yours sincerely, Pavlo Korzhyk skype: pavlo.korzhyk ICQ#155870780

6 6

boost::python PyUnicodeUCS4_FromEncodedObject
by Venkatraman S 27 Mar '08

27 Mar '08

Hi, Am stumbling on a very rudimentary problem and am totally not able to proceed. Googling did not help me much and hence am trying to state the problem in its entirety. Any quick comments/hints would be highly helpful. Am getting the following error when i try to import the module (class) into python : $python >>> import myClass *Traceback (most recent call last): File "<stdin>", line 1, in ? ImportError: ./myClass.so: undefined symbol: PyUnicodeUCS4_FromEncodedObject * *Environment:* g++ (GCC) 4.0.0 20050519 (Red Hat 4.0.0-8) Python 2.4.3 *Other Information:* $ldd myClass.so libstdc++.so.6 => /usr/lib/libstdc++.so.6 (0x00db8000) libm.so.6 => /lib/tls/libm.so.6 (0x002b3000) libgcc_s.so.1 => /lib/libgcc_s.so.1 (0x00eb7000) libpthread.so.0 => /lib/tls/libpthread.so.0 (0x0057a000) libc.so.6 => /lib/tls/libc.so.6 (0x00b6f000) /lib/ld-linux.so.2 (0x00712000) * Files Used :* *myClass.cpp* ------------- #include <boost/python.hpp> #include <boost/python/def.hpp> #include <boost/python/tuple.hpp> #include <boost/python/dict.hpp> #include <boost/python/module.hpp> #include <boost/python/raw_function.hpp> #include <string> class Var { public: int value; std::string name; Var(std::string n) { name=n; value=10; } }; std::string processObj(boost::python::object x) { Var obj=boost::python::extract<Var>(x); return obj.name; } using namespace boost::python; BOOST_PYTHON_MODULE(myClass) { class_<Var>("Var", init<std::string>()) .def_readonly("name", &Var::name) .def_readwrite("value", &Var::value); def("processObj", processObj); } *setup.py* --------------- from distutils.core import setup from distutils.extension import Extension setup(name="myvaarg",ext_modules=[Extension("myvaarg",["sample.cpp "],libraries=["boost_python"])]) $python setup.py build -- Venkat

2 2