March 2010 - Boost-users - lists.boost.org

How to use Boost Log
by Jean-Sebastien Stoezel 30 Mar '10

30 Mar '10

Hello, Well I know Boost log (http://torjo.com/) is not part of boost (yet ?) but I haven't been able to figure it out on my own... So I was hoping somebody here could help me. I've installed Boost using Boost Pro. The install is very easy though it does not install Boost Log. I got Boost Log separately but it does not come with libraries directly linkable with VS2009. How can I generate linkable libraries for Boost Pro, for VS2009? Thanks! Jean

4 7

boost file system, program_option no longer build on HPUX/aCC/1.42
by Avi Bahra 30 Mar '10

30 Mar '10

For 1.42 can no longer build boost on HP-UX. File-system, program-options and serialisation all show a similar problem + bjam --build-dir=./tmpBuildDir toolset=acc stage link=static --with-filesystem variant=debug ...patience... ....... acc.compile.c++ tmpBuildDir/boost/bin.v2/libs/ filesystem/build/acc/debug/link-static/threading-multi/utf8_codecvt_facet.o "./libs/detail/utf8_codecvt_facet.cpp", line 255: error #2014-D: extra text after expected end of preprocessing directive #elif WCHAR_MAX > 0x10000 The actual file <boost_root>/libs/detail/utf8_codecvt_facet.cpp appears to have been changed in boost 1.40 for the function get_cont_octet_out_count_impl FROM: // note the following code will generate on some platforms where // wchar_t is defined as UCS2. The warnings are superfluous as // the specialization is never instantitiated with such compilers. template<> int get_cont_octet_out_count_impl<4>(wchar_t word){ if (word < 0x80) { return 0; } if (word < 0x800) { return 1; } if (word < 0x10000) { return 2; } if (word < 0x200000) { return 3; } if (word < 0x4000000) { return 4; } return 5; } TO: template<> int get_cont_octet_out_count_impl<4>(wchar_t word){ if (word < 0x80) { return 0; } if (word < 0x800) { return 1; } // Note that the following code will generate warnings on some platforms // where wchar_t is defined as UCS2. The warnings are superfluous as the // specialization is never instantitiated with such compilers, but this // can cause problems if warnings are being treated as errors, so we guard // against that. Including <boost/detail/utf8_codecvt_facet.hpp> as we do // should be enough to get WCHAR_MAX defined. #if !defined(WCHAR_MAX) # error WCHAR_MAX not defined! #endif // cope with VC++ 7.1 or earlier having invalid WCHAR_MAX #if defined(_MSC_VER) && _MSC_VER <= 1310 // 7.1 or earlier return 2; #elif WCHAR_MAX > 0x10000 if (word < 0x10000) { return 2; } if (word < 0x200000) { return 3; } if (word < 0x4000000) { return 4; } return 5; #else return 2; #endif } Does any one know of a work around ? Any help appreciated. Best regards, Ta, Avi

1 0

Boost Graph some newbie questions
by Michał Nowotka 30 Mar '10

30 Mar '10

Hello, Although some time ago I used boost graph in some of my projects I found this library too generic for me :) I still have some questions: 1. How to check if my graph is a tree? 2. How to generate random graph? (using generate_random_graph) - I can't find any examples of random graph generation so it should be very very easy and I need the simplest possible example. 3. How to use layout functions like random_graph_layout? - as in the question above I can't find any examples. 4. Can I perform all this actions on graph of type boost::adjacency_matrix<boost::directedS> ? -- Regards Michał Nowotka

3 4

[Thread]: code snippet simulating future
by Boost lzw 30 Mar '10

30 Mar '10

Hi folks, To better understand the "future" concept, I am looking at the following code and have several questions: (1) I got a compiling warning: 'this' : used in base member initializer list. How can I get rid of it? (2) how can I bind class's local data x_ into the new thread? I tried: t_(boost::bind(&future<T>::run, this, x_) ), but it does not work. (3) There seems a race condition in the overloaded function call operator()(). What wrong did I do? BTW: if the future is constructed as ctor using: future(boost::function<T ()> const& f); the code snippet works with minor modification. Thanks in advance, Cheers, Robert ====== code snippet ===== #include <boost/bind.hpp> #include <boost/thread/mutex.hpp> #include <boost/thread/thread.hpp> #include <boost/function.hpp> template<typename T> class future : boost::noncopyable { public: future(boost::function<T (int)> const& f, int); T operator()(); private: int x_; bool joined_; T v_; boost::function<T (int)> const& f_; boost::mutex m_; void run(); boost::thread t_; }; template<typename T> future<T>::future(boost::function<T (int)>const& f, int x) : x_(x), joined_(false), v_(T()), f_(f), t_(boost::bind(&future<T>::run, this) ) { } template<typename T> T future<T>::operator()() { mutex::scoped_lock lock(m_); if (!joined_) { t_.join(); joined_ = true; } return v_; } template<typename T> void future<T>::run() { v_ = f_(x_); }

1 0

[filesystem] exists() throws if no disk in CD-ROM drive (Boost version 1.41)
by PB 30 Mar '10

30 Mar '10

Hello, I was looking to remove a wrapper function I wrote a while ago when I was using Boost 1.35.0. This wrapper caught exceptions from boost::filesystem::exists and returned false: bool PathExists(const boost::filesystem::path& path) { try { return boost::filesystem::exists(path); } catch(const boost::filesystem::filesystem_error&) { return false; } } Now our organisation has upgraded to 1.41.0 I wanted to get rid of the wrapper as I can see that ticket 2725 (https://svn.boost.org/trac/boost/ticket/2725) has been fixed in 1.37.0. A quick test shows that I'm still getting exceptions from boost::filesystem::exists() when referring to a CD-ROM drive without any media in it: #include <iostream> #include <boost/filesystem/operations.hpp> int main() { std::cout << boost::filesystem::exists("d:\\afile.txt") << std::endl; } Could someone take a look at this please? Many thanks, Pete

1 0

Program_option: problems parsing multiple commands with multiple arguments
by Luca Luisa 30 Mar '10

30 Mar '10

Hello, this is my first post in Boost. If this post is not in the right mailing list please let me know, many thanks. I have problems using the "program_options" library while parsing a command line with multiple commands with multiple arguments each. The test.cpp is in attachment. I would like to pars such a command: ./test --firstCommand=arg1 arg2 arg3 --secondCommand=arg1 arg2 arg3 and having as result: ************************ First command detected List of parameters passed: arg1, arg2, arg3 Second command detected List of parameters passed: arg1, arg2, arg3 ************************ But the first command bring the second command as part of its list of arguments with the result: *********************** First command detected List of parameters passed: arg1, arg2, arg3, --secondCommand=arg1, arg2, arg3, ********************** Any hints would be appreciate. Regards, Luca

3 3

compiler error on using threads with boost
by Iskandar Prins 30 Mar '10

30 Mar '10

Hi, I get this compiler error : boost/thread/future.hpp", line 414: Error: Formal argument other of type boost::unique_lock<boost::mutex> in call to boost::unique_lock<boost::mutex>::operator=(boost::unique_lock<boost::mutex>) has an inaccessible copy constructor. Any idea why this is? I'm using the sun compiler to compile it with Regards, iskandar

4 9

Accessing locale information
by Carla Strembicke 30 Mar '10

30 Mar '10

I need a call that allows me to specifiy the value locale information to retrieve. Does one exist in Boost. I have been swimming through the documentation but find that I am drowning. Background -> trying to replace windows GetLocaleInfo( ....) to become cross platform compliant. Thanks CONFIDENTIAL INFORMATION NOTICE: The information contained in this e-mail is privileged, confidential and intended solely for the use of the addressee named above. If the reader of this e-mail is not the intended recipient, you are hereby notified that any dissemination, distribution or copying of this e-mail is strictly prohibited. If you received this in error, please contact the sender and destroy all copies of this e-mail. Thank you.

3 3

Boost Graph: add_edge causes compiler error
by Trevor Harmon 30 Mar '10

30 Mar '10

Hi, Could someone please help me understand why the simple program below won't compile? On the call to add_edge, I get: error: invalid conversion from ‘Foo*’ to ‘long unsigned int’ ...followed by two "initializing argument" errors. I've read all the docs I could find on add_edge, but I still don't understand why the code below won't work. Thanks for any help, Trevor #include "boost/graph/adjacency_list.hpp" using namespace boost; typedef struct FooType { } Foo; int main() { typedef adjacency_list<vecS, vecS, directedS, Foo*> Graph; Graph g; Foo foo1, foo2; add_vertex(&foo1, g); add_vertex(&foo2, g); add_edge(g[0], g[1], g); return 0; }

2 5

Boost Graph: preventing duplicate vertices when constructing graph incrementally
by Trevor Harmon 29 Mar '10

29 Mar '10

Hi, The vast majority of Boost Graph's sample code has pre-defined source data. For example, from city_visitor.cpp: enum { SanJose, SanFran, LA, SanDiego, Fresno, LasVegas, Reno, Sacramento, SaltLake, Phoenix, N }; E edge_array[] = { E(Sacramento, Reno), E(Sacramento, SanFran), E(Reno, SaltLake), E(SanFran, SanJose), E(SanJose, Fresno), E(SanJose, LA), E(LA, LasVegas), E(LA, SanDiego), E(LasVegas, Phoenix) }; Unfortunately, static source data with enums as the vertices isn't very typical in real-world programs. It's much more common to construct the graph incrementally from a streaming data set, such as a file containing a list of edge pairs. But this presents a problem: As the list of edges is read in, a vertex may be encountered more than once, so add_vertex can't simply be called on every vertex found because that will result in redundant vertices. An obvious fix is to check whether a vertex exists before calling add_vertex. If it is found, it can be used for the edge; otherwise add_vertex can be called and the returned value used for the edge. The only question is, How do I check whether a vertex exists? I could iterate through all vertices in the graph, searching for one that equals the one I'm trying to add, but this approach doesn't scale well. It has a time complexity of O(V). A faster solution to this problem appears to be in the boost_web_graph.cpp sample. It deals with the same issue: It reads in graph data from a file called boost_web.dat, and this file contains the same vertex value in multiple places. Rather than call add_vertex multiple times (creating redundancies), the code puts every vertex it finds into a map. The map associates the vertex's unique value (a string in this case) with its corresponding vertex descriptor. Then, before calling add_vertex, the code checks the map for the unique value; if found, it uses the corresponding vertex descriptor; otherwise it calls add_vertex and adds the new ID/descriptor pair to the map. I assume that this is faster (compared to my naive approach of searching linearly through the vertices) because std::map's insert function is supposed to have logarithmic time complexity. It probably achieves this with a red-black tree, so as long as the keys are sortable, O(log(V)) complexity is guaranteed. Strings (used in the sample code) are certainly sortable, although in my program I'm using pointers as the vertex property. But I guess pointers are sortable too, since they're just numbers really. My initial tests show that this approach works properly with pointers as the values, though I've not done any performance testing, only correctness testing. Any feedback on this would be appreciated. Thanks, Trevor

1 0