November 2005 - Boost-users

Boost.Serialization vs STL containers of 'double'
by Adam M 12 Nov '05

12 Nov '05

Hi, I am using Boost.Serialization to serialize a std::map<std::string, double>. It works fine, except that the precision on the double is not enough for my application. Is it possible to manipulate the precision used when serializing/deserializing double's within an STL container? I tried using setprecision(), but it doesn't understand the "&" stream operator that the Serialization library uses. Thanks, Adam

2 2

MultiArrays of Std::Lists: why const iterators?
by Joshua Schpok 12 Nov '05

12 Nov '05

I want to make a 3D array of std::lists: boost::multi_array<std::list<CSomething>,3> listbin; Then I'd like to actually use those lists: /// Walk through list at x,y,z for(std::list<CSomething>::iterator iter = listbin[x][y][z].begin(); reti != iter[x][y][z].end(); iter++) { iter->m_foo += 42; } But multi_array is giving me back *const* list iterators, producing the error: error: conversion from `std::_List_const_iterator<CSomething>' to non-scalar type `std::_List_iterator<CSomething>' requested Why is this so? I can trick the compiler by doing a cast like: std::list<CSomething>::iterator iter = ((std::list<CSomething>)listbin[x][y][z]).begin(); but this seems to poop out at runtime. Is there a proper way to do this? Joshua Schpok

3 2

[bind]How to bind to a template class member function?
by Wang Yun 12 Nov '05

12 Nov '05

Hi, I tried to bind to a class member function template, but can't pass compile, what's the correct syntax or this is impossible? class A { public: void test() { function<void()> func; func = boost::bind( &A::ttt<int>, this ); } template<typename T> T ttt() { return 1; } }; int _tmain(int argc, _TCHAR* argv[]) { return 0; } Thanks, Wang Yun

2 2

[serialization] Incorrect assertion in basic_binary_oprimitive?
by Austin Bingham 12 Nov '05

12 Nov '05

I've run into a situation where I'm triggering an assertion at basic_binary_oprimitive::save_binary(). This is on basic_binary_oprimitive.hpp, line 115 in version 1.32. This assertion seems to be claiming that an output stream will always be in a "good" state after a call to write(). On it's face, I think there's no way to safely make that claim, especially about a stream whose type is a template parameter. In fact, as I understand it, good()'s whole purpose is to let you know that the stream is busted. Am I misunderstanding something, or should that assert be changed to an exception? I'm trying to get an example program that demonstrates this problem. Currently, the code that exercises it is big and hairy, and I haven't pared down the precise conditions that cause it to happen. I can provide a stack trace, however, which may be of some use. -- Austin Bingham "If I were creating the world I wouldn't mess about with butterflies and daffodils. I would have started with lasers, eight o'clock, Day One!" Evil

2 2

When FSM library is released
by Oliver.Kowalke＠infineon.com 11 Nov '05

11 Nov '05

Hello, could you tell me please when the next release of boost with fsm (finite state machine) library will be released? thx, Oliver

3 2

Namespaces for Intrusive Serialization?
by Merrill Cornish 11 Nov '05

11 Nov '05

In the Serialization tutorial, and example is given of the "intrusive" method that looks like this: private: friend class boost::serialization::access; template<class Archive> void serialize(Archive& ar, const unsigned int version) { ar& var1; ar& var2; ar& var3; } There is full namespace qualification on "access" but nothing on "serialize()". It appears that serialize() would be in whatever the prevailing application namespace is (which might not be std). Later in the tutorial there is an example of the "non-intrusive" method that looks like this" namespace boost { namespace serialization { template<class Archive> void serialize(Archive& ar, const unsigned int version) { ar& var1; ar& var2; ar& var3; } }//serialization }//boost which implies that serialize() *DOES* need to be namespace qualified. So, what's really needed? Merrill

3 2

Deploying Boost App: Best Practices?
by Merrill Cornish 11 Nov '05

11 Nov '05

I am building an open source C++ application using various Boost libraries. I am using the libraries as-is without any tweaks, additions, or wrappers to the source or headers. What is the accepted way (if there is one) of deploying a Boost application when the application includes the source under an open source license? Should I distribute only *my* source along with instructions to users who want to compile it to download the Boost libraries? Or, should I include a boost/ subdirectory in my source tree with the Boost headers and libraries my app uses? Merrill

3 2

filesystem::equivalent does not work for UNC paths on Windows
by Daniel Krügler 11 Nov '05

11 Nov '05

Hello! Its seems as if the Windows implementation of filesystem::equivalent can't handle UNC paths. While Windows is usually quite tolerant concerning the path separator / versus \, CreateFile* seems not to understand an UNC path containing // instead of \\. My proposal to fix that is to replace the call of the string() member function by the native_file_string() member function. What do you think? Greetings from Bremen, Daniel

1 1

[filesystem] path::native_file_string() heap issue and
by Brendon Costa 11 Nov '05

11 Nov '05

Hi all, I have been using the boost filesystem library for a little while under NetBSD and all is fine however I have just been porting our application over to windows and have come across a problem with the boost filesystem library (Boost 1.33.0). I am compiling with MinGW (g++ 3.4.2) on a Windows XP system. When debugging our application in gdb i get a SIGTRAP with an Invalid Address specified to RtlFreeHeap. This is when using the path::native_file_string() method. I also have another problem where a SEGFAULT ocurrs when using the boost::filesystem::exists() method, however i think that may be linked to the native_file_string() error. I wrote a small test program as shown below to reproduce the error when using path::native_file_string() #include <iostream> #include <boost/filesystem/operations.hpp> int main() { std::string filename = "test.dat"; boost::filesystem::path boost_filename(filename); std::cout << "Filename: " << filename << std::endl; std::cout << "Boost string(): " << boost_filename.string() << std::endl; std::cout << "Boost native_file_string(): " << boost_filename.native_file_string() << std::endl; return 0; } This compiles fine, and it also will run fine outside of GDB. However if i run it inside GDB i get errors on the line that uses boost_native_file_string() that are shown below: Starting program: X:\msys\Utils_test/./a.exe Program received signal SIGTRAP, Trace/breakpoint trap. 0x7c901231 in _libkernel32_a_iname () (gdb) warning: HEAP[a.exe]: warning: Invalid Address specified to RtlFreeHeap( 003D0000, 003D5138 ) bt #0 0x7c901231 in _libkernel32_a_iname () #1 0x7c96c943 in _libkernel32_a_iname () #2 0x7c96cd80 in _libkernel32_a_iname () #3 0x7c96df66 in _libkernel32_a_iname () #4 0x7c94a5d0 in _libkernel32_a_iname () #5 0x7c9268ad in _libkernel32_a_iname () #6 0x77c2c2de in _libkernel32_a_iname () #7 0x00409855 in operator delete(void*) () at boost_test.cpp:59 #8 0x0042e2dc in std::string::_Rep::_M_destroy(std::allocator<char> const&) () at c:/Boost/include/boost-1_33/boost/filesystem/path.hpp:57 #9 0x00430d68 in std::string::~string() () at c:/Boost/include/boost-1_33/boost/filesystem/path.hpp:57 #10 0x004015d6 in main () at boost_test.cpp:12 (gdb) I also have a larger program where part of its code uses the boost filesystem library that looks like: static std::string GetUnusedFileName(const char* suggestion = "test_file") { int count = 0; std::string base = suggestion; std::string filename(base + ".dat"); while(boost::filesystem::exists(boost::filesystem::path(filename))) { std::ostringstream stream; stream << base << "_" << count << ".dat"; filename = stream.str(); count++; } return filename; } This will usually cause a seg fault when calling the filesystem::exists() method. I currently do not have a stack trace for that and was unable to reproduce one but i think the two errors may be linked. Am i using the system correctly? The code works fine in NetBSD so i assume so. Has this bug been identified in the past and possibly fixed for the 1_33_1_beta release? I started compiling that but there are a few changes i need to make to the build system to go to 1_33_1 that may be time consuming for a problem that may not yet have been addressed anyway. Thanks for any help in advance, Brendon Costa.

1 0

[BGL] Edge descriptors for parallel edges
by Slava 11 Nov '05

11 Nov '05

Say, i have the following structure for edge bundle properties struct EdgeProperty { int edge_type; } and i can add parallel edges, that is between vertices u and v cab be several edges with bundle property EdgeProperty with different edge_types. the first question how can i use edge(u,v,graph) function if it returns std::pair <edge_descriptor,bool>, only one descriptor? but i have several descriptors between two vertices. second, how to change the edge_type provided that i got edge_descriptor from edge(u,v,graph) Thanks.

2 1