February 2008 - Boost-users - lists.preview.boost.org

Are there some regression testings for the boost
by Wang, Ronnie 13 Feb '08

13 Feb '08

Hi All, Are there some regression testing for boost? Best Regards, -Ronnie

2 1

[boost-users] [Mixing a dll boost library with a static runtime is a really bad idea]
by chun ping wang 13 Feb '08

13 Feb '08

1.) how would I build a library in *msvc* through *nmake* so that I could get autolink to work My process is like this. a.) build my own lib file. b.) build boost lib file (succeeded) c.) build the main file and auto link to boost file (failure). 2.) What are the suggestions and requirement. 3.) Is their a way to autolink to static runtime library. Thanks.

1 0

[graph] edge_descriptor invalidation
by Sebastian Weber 13 Feb '08

13 Feb '08

Hello everyone! I'm asking myself, why edge-descriptors become invalid if I remove the edge and reinsert it? In my case the graph is represented by an adjacency list with vecS storage for vertices and edges. If I'm not mistaken, then the edge_descriptor's are just represented by pairs of vertex indices, right? So why are edge_descriptors invalid if I delete these edges, reinsert the edges and then want to use the old edge_descriptors to address those recreated edges? Most interestingly, a very cumbersome edge(source(old_descriptor, graph), target(old_descriptor, graph), graph).first gives me perfectly valid edge_descriptors again after removal and insertion. I'm confused. Thanks for any help in advance. Sebastian Weber

2 3

[bjam] using bjam as a build tool
by Ovanes Markarian 13 Feb '08

13 Feb '08

Hello bjam developers! I currently read through the docs for bjam and try to build my first project using. We currently need an automated build system in the current project to run directly from the version control system. The doc which I read currently leaves many items open which I do not understand, despite the fact I see bjam as a very cool thing! far superior as make;) These questions are open: 1. From the userman.pdf I learned that there is a possibility to use modules. Is there any location where I can what the modules do or how they work. Since such examples are not very well to understand: userman.pdf page 47-48 ------- "Accessing environment variables Many users would like to use environment variables in Jamfiles, for example, to control location of external libraries. In many cases you better declare those external libraries in the site-config.jam file, as documented in the recipes section. However, if the users already have the environment variables set up, it's not convenient to ask them to set up site-config.jam files as well, and using environment variables might be reasonable. In Boost.Build V2, each Jamfile is a separate namespace, and the variables defined in environment is imported into the global namespace. Therefore, to access environment variable from Jamfile, you'd need the following code: import modules ; local SOME_LIBRARY_PATH = [ modules.peek : SOME_LIBRARY_PATH ] ; exe a : a.cpp : <include>$(SOME_LIBRARY_PATH) ;" ------- 2. As I understood I can use user-config.jam to make default configurations for my project. I copied the user-config.jam into the project root, but it is not read, at least bjam issues warnings that toolset was not set and msvc is the default one. How can I specify in user-config.jam a default toolset? Or is the user-config-jam a file to declare available toolset, but toolspecification must be somewhere else? My user-config.jam contains: import toolset : using ; using msvc ; 3. How is it possible to specify the source location to 3d party lib (libpath)? I can't build a project with bjam v2 which uses boost libs. I don't want to enumerate each lib which my project depends on, but submit to the linker the lib path. My file content is: import modules ; local BOOST_STAGE_INCLUDE = [ modules.peek : BOOST_STAGE_INCLUDE ] ; local BOOST_STAGE_LIBS = [ modules.peek : BOOST_STAGE_LIBS ] ; exe MTSentry : main.cpp <source>$(BOOST_STAGE_LIBS) : <include>$(BOOST_STAGE_INCLUDE) <threading>multi ; The resulting error is: bjam.exe --v2 --toolset=msvc error: at Jamroot:7 error: properties found in the 'sources' parameter for ./MTSentry Removing the <source> option will build the app, but the linking fails, since boost/thread lib is not found. How can I specify a source directory for libs? With Kind Regards, Ovanes

2 3

OSI Approval
by KSpam 12 Feb '08

12 Feb '08

In perusing the list of OSI approved licenses today, I noticed that Boost is finally listed: http://www.opensource.org/licenses/alphabetical http://www.opensource.org/licenses/bsl1.0.html Although the approval process seems to be incredibly long and painful, it is great to see the Boost Software License officially recognized by OSI!

1 0

XML Serialization and the Generation of a W3C Schema
by Joseph Fradley 12 Feb '08

12 Feb '08

Hello, I've seen a few postings on how a XSD generating archiver would be written and a posting for a google code project on it. But I can't seem to find any implementations. Does anybody know of any implementations? Joe

2 1

[Regex] "Truly" incremental search?
by Eugene M. Kim 12 Feb '08

12 Feb '08

Hello, The current "incremental search" code needs to resume search where the a possible match starts, and for certain expressions the worst-case time (i.e. input bytes come in one at a time) complexity becomes O(n^2) where n is the number of characters that match the expression: Regex: omg.*wtf Pass 1 input: o Pass 1 input to matching: o Pass 1 verdict: Partial, resume at offset 0. Pass 2 input: m Pass 2 input to matching: om Pass 2 verdict: Partial, resume at offset 0. Pass 3 input: g Pass 3 input to matching: omg Pass 3 verdict: Partial, resume at offset 0. ... Pass n-1 input: t Pass n-1 input to matching: omg...wt (length n-1) Pass n-1 verdict: Partial, resume at offset 0. Pass n input: f Pass n input to matching: omg...wtf (length n) Pass n verdict: Match found and starts at offset 0. Total number of byte matches = n*(n+1)/2, i.e. O(n^2) I am looking for a way to feed the input bytes only once, in order to reduce the time complexity to approximately O(n). Of course I will lose the ability to recover submatches, but the ability is unnecessary for this particular application that I am considering. I cannot break the regex into two parts ("omg" and "wtf") either because it is specified as an input to the program. Eugene

5 6

bounded_matrix initialization
by Barberi 12 Feb '08

12 Feb '08

A boost newbie asks, how can I initialize a bounded_matrix? I tried various things like this: bounded_matrix<double, 2, 2> rot = list_of( 1.) (2.) (3.) (4.); and got the error message 'no overloaded function takes 2 arguments'. So far, I'm reduced to using copy assignment from a c array, but there's got to be a better way. What am I missing? BTW, I'm using MSVC++ 2008, with the MSVC++ 8 boost libraries.

1 0

[asio] Synchronization & infinite wait
by Sergei Politov 12 Feb '08

12 Feb '08

I have two questions about Boost.Asio library: 1) Is it safe to invoke deadline_timer::expires_from_now in thread different from thread where io_service is running? For instance: =========================================== void print(const boost::system::error_code& e) { std::cout << "Hello, world:" << e << "!\n"; } void execute(boost::asio::deadline_timer & timer) { timer.expires_from_now(boost::posix_time::seconds(0)); } int main() { boost::asio::io_service io; boost::asio::deadline_timer t(io, boost::posix_time::seconds(1)); t.async_wait(print); boost::thread th(boost::bind(&execute, ref(timer))); io.run(); return 0; } ============================================ 2) Are there any type of objects similar to events from WINAPI? I.e. these objects is not signalled (handler is not invoked) until they are signalled manually. Obviously such objects could be emulated using deadline_timer. But it would be great to have such objects itself.

1 0

Re: [Boost-users] Unresolved symbol in boost::filesystem lib with MSVC++ 2005 Pro
by Peter Jansson 12 Feb '08

12 Feb '08

Hi, I have the same issue as described in this thread and I am very much interested to hear if anyone have any suggestion on how to solve it. Best regards, Peter Jansson Sweden

1 0

Static build on darwin
by Thomas Hauk 12 Feb '08

12 Feb '08

Hi all, I'm having trouble with using Boost.Thread (1.34.1) on Mac OS X (10.4.11/ppc/Xcode 2.3). I have an Xcode project ("BSD command line tool") that links to what I believe is the static build of Boost.Thread (libboost_thread-mt- d-1_34_1.a). However, when Xcode goes to run the executable, it gives the error: > Running… > dyld: Library not loaded: bin.v2/libs/thread/build/darwin/debug/ > threading-multi/libboost_thread-mt-d-1_34_1.dylib > Referenced from: /path/to/my/executable/executablename > Reason: image not found > Program received signal: "SIGTRAP". If I run the executable inside an X11 shell, I get a similar error. If I add the path to the .dylib to the $DYLD_LIBRARY_PATH environment variable, I can run the executable. Which makes me wonder if somehow this .a file is really a .dylib. What's also strange is that although the build directories for all the other libraries have the same structure (e.g. filesystem/build/ darwin/debug/link-static/libboost_filesystem-d-1_34_1.a), the thread directory has a threading-multi directory under the link-static directory (meaning the file I am trying to link to is in thread/build/ darwin/debug/link-static/threading-multi/libboost_thread-mt-d-1_34_1.a). I'm building all the libraries with bjam 3.1.15 and the command line "bjam --toolset=darwin". FWIW, I'm also using the Boost filesystem library with this executable, without any compile, link or run-time errors. Any help/ideas? Also: is there any plan to distribute the Boost libraries for Mac OS X in a similar way to how they are distributed for Windows (i.e. installer that has pre-built library binaries)? T -- Some people, when confronted with a problem, think "I know, I'll use regular expressions." Now they have two problems. --Jamie Zawinski, in comp.lang.emacs

1 0

[serialization] exceptions are not caught
by Krishanu Debnath 12 Feb '08

12 Feb '08

Hello, This is a platform specific problem, platform is RHEL4. We use boost::serialization code, and linked with their static library. And the problem we are facing that all of our exception handler codes are not working any more i.e. catch handler is not called if an exception is thrown. We are not even calling the boost::serializer code. If we remove serializer code everything works fine. Serializer code works pretty fine in suse platform. Now my question is how come exception handler codes are not executing just because we are linking with serializer library. Any insight will be helpful. Thanks, Krishanu

1 0

[filesystem] bug in wpath with utf8 locales
by Filip Konvička 12 Feb '08

12 Feb '08

Hi, there is a problem with wpath when the system locale is UTF-8. My setup is g++ 4.2.3 --enable-clocale=gnu on a 2.6.18-5-686 GNU/Linux, boost 1.34.1, locale "cs_CZ.UTF-8". The problem seems to be in translating wstring into native strings inside Filesystem. The error I'm getting (and others too, see for example "Problem with exists() and wpath" and "Bug in filesystem library?" on this list) is boost::filesystem::wpath::to_external conversion error I've seen some discussions on the boost-dev list dealing with this issue, and to me this looks like a serious bug in wpath implementation. Until this is resolved, I think that we need to #ifdef this and use char-based paths on Linux. Are there some plans to fix this? I'm willing to help if that is needed. Thanks, Filip

1 0

Boost threads.
by Paul J R 12 Feb '08

12 Feb '08

Hi all, I had a question about boost threads as im using them currently in a project. I've written a very simple implementation of what im trying to do and the result I get, but im having trouble understanding why: mytest.cpp: class A { public: A() {} ~A() {} virtual void worker() { std::cout << "I am A worker" << std::endl; } virtual A *factory() { std::cout << "Factory for A class called" << std::endl; return new A; } virtual void operator()() { std::cout << "Calling A operator()" << std::endl; worker(); } void threadcontroller() { boost::thread_group thre; for(;;) { std::cout << "I am in the A.threadcontroller() method" << std::endl; A *mea = factory(); mea->worker(); thre.create_thread(*mea); exit(0); } } }; class B: public A { public: B() {} ~B() {} virtual void worker() { std::cout << "I am B worker" << std::endl; } virtual void operator()() { std::cout << "Calling B operator()" << std::endl; worker(); } B *factory() { std::cout << "Factory for B class called" << std::endl; return new B; } }; int main() { B meb; meb.worker(); meb.threadcontroller(); } Output: $ ./mytest I am B worker I am in the A.threadcontroller() method Factory for B class called I am B worker Calling A operator() I am A worker The first output is from the meb.worker() call in main(), the second is from meb.threadcontroller. the thrid is the factory call from within threadcontroller, the forth is mea.worker() in thread controller, then the next 2 are from after the thread_group.create_thread() call. Once its in the thread it seems to revert to class A's methods. Now, i know if i overload the threadcontroller class in B it'd solve all my problems, but the real threadcontroller method is alot more complex and class A has about 6 sub classes so replicating the threadcontroller in each of the classes would be a little tedious. Have i stumbled across a bug, or is that how its supposed to work? Thanks in advance

5 7

[python] Integration without a DLL
by Robert Dailey 11 Feb '08

11 Feb '08

Hi, I have a game that currently integrates the python interpreter using the Python C API. I would like to expose various C++ interfaces & functions to python, but I want to do so without having to create a separate DLL, as state needs to be modified in the actual core engine of the game as the user calls C++ functions from Python. For example, say I have a C++ function called CompleteQuest(). In Python, we would script a quest for my game and at some point call CompleteQuest() from python. This should map directly to a function in the executable (game). Is there a way to do this? Thanks.

4 9

[thread] is this a bug?
by Hector C 11 Feb '08

11 Feb '08

Hello fellow boost fans, Could anyone explain to me why the destructor (in the program listed below) seems to be called 10 times but the constructor only one time? The output of the program is also shown below. Cheers, Hector C. --- BEGIN PROGRAM --- #include <boost/thread/thread.hpp> #include <iostream> using namespace std; struct ThreadBug { ThreadBug() { cout << "Begin sending ... " << endl; } ~ThreadBug() { cout << "End sending ... " << endl; } void operator()() { cout << "Do something ... " << endl; } } w_ThreadBug; int main() { cout << "New Sequence -----------" << endl; boost::thread thrd(w_ThreadBug); thrd.join(); cout << "End New Sequence -----------" << endl; return 0; } ---END PROGRAM --- Begin sending ... New Sequence ----------- End sending ... End sending ... End sending ... End sending ... End sending ... End sending ... End sending ... End sending ... End sending ... Do something ... End sending ... End New Sequence ----------- End sending ...

2 1

[DateTime] posix_time_zone constructor crashes with bad format string
by Richard Buckle 11 Feb '08

11 Feb '08

Hello, I'm using Boost 1.34.1 with Visual Studio 2005 SP1. I've found that if the posix_time_zone constructor is invoked with certain types of invalid time zone string, the application crashes with an attempt to dereference an invalid iterator. Examples of strings that do this are strings with no digits, e.g. "UTC" and strings with a daylight saving time named but no changeover rules, such as "GMT0BST". This is a problem for us as in our application the time zone string is input by the user, and we need to fail gracefully if the user naively supplies an invalid string such as the two above. I think this is easily fixed by testing the iterators against end(), although I'm not quite sure which exception to throw in this case. Any ideas? TIA, Richard.

1 0

Linking to boost_regex library giving unresolved symbols
by shouvik ganguly 11 Feb '08

11 Feb '08

Hi, I am trying to access the function regex_search() by linking to the boost_regex library. However I am getting the following unresolved symols on GNU/Linux OS 1.undefined symbol: _ZN5boost14reg_expressionIcNS_12regex_traitsIcEEN4_STL9allocatorIcEEE14set_expressionEPKcS8_j(../libAspPatternFindReplace.so) 2. undefined symbol: _ZN5boost9re_detail21raise_regex_exceptionERKN4_STL12basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEE (../libAspPatternFindReplace.so) 3. undefined symbol: _ZN5boost14c_regex_traitsIcE17transform_primaryERN4_STL12basic_stringIcNS2_11char_traitsIcEENS2_9allocatorIcEEEERKS8_ (../libAspPatternFindReplace.so) 4. undefined symbol: _ZN5boost14c_regex_traitsIcE9transformERN4_STL12basic_stringIcNS2_11char_traitsIcEENS2_9allocatorIcEEEERKS8_ (../libAspPatternFindReplace.so)The first and third unresolved symbol complain about the functions set_expression() and transform_primary(). However I am not able to get a clue as to the 1st and 4rth link error. How can I resolve the link errors. Thanks _________________________________________________________________ Express yourself instantly with MSN Messenger! Download today it's FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/

2 1

Re: [Boost-users] Review of Logging library is in progress
by David Abrahams 11 Feb '08

11 Feb '08

on Mon Feb 11 2008, "Gennadiy Rozental" <rogeeff-AT-gmail.com> wrote: > Hi, > > To those who managed to miss this very exciting submission, I would like to > encourage to find some time to invest into reviewing it. Admitedly this is > very open-bound problem domain and many people have different opinions on > design requirements and possible interfaces. It makes it even more important > to get large number of reviewers representing wide spectrum of users base. > Please let me know about any time constrains, I might try to arange for the > review period extension. You left out a very important piece of information: when is the review deadline? -- Dave Abrahams Boost Consulting http://boost-consulting.com

1 0

[program_options] exceptions
by Filip Konvička 11 Feb '08

11 Feb '08

Hi, I started using the program_options library and I have a question about using the exceptions thrown by the library. I'd like to localize the errors reported by the library, but I can't see why the members of the exception classes are private, or why are there no getters to retrieve the properties needed to re-construct the messages in another language. Also, I can't make default wstring values work...my attempt is: #include <boost/program_options.hpp> #include <string> using namespace std; using namespace boost::program_options; int main() { wstring lang; options_description lang_opt_desc; variables_map vm; lang_opt_desc.add_options() ("lang", wvalue<wstring>(&lang)->default_value(L"en_US"), "Language ID.") return 0; } fails with both gcc and msvc, saying that boost/lexical_cast.hpp(174) : error C2679: binary '=' : no operator found which takes a right-hand operand of type 'std::basic_string<_Elem,_Traits,_Ax>' (or there is no acceptable conversion) (I'm using Boost 1.34.1 with MSVC 9.0 and g++ 4.2.3). I looked in the code and it looks that I could use ... ("lang", wvalue<wstring>(&lang)->default_value(L"en_US", "en_US"), "Language ID.") ... which works fine...I just wanted to know whether this is the intended approach. Thanks, Filip

1 0

error:void* is not a pointer-to-object type
by BomBielFil 11 Feb '08

11 Feb '08

Hi I tried to compile this example: typedef boost::adjacency_list <boost::listS,boost::listS,boost::undirectedS,vertex_descriptor > Graph; typedef boost::graph_traits<Graph>::vertex_descriptor v_des; typedef boost::graph_traits<Graph>::vertex_iterator v_iter; typedef boost::graph_traits<Graph>::edge_descriptor e_des; typedef boost::graph_traits<Graph>::edge_iterator e_iter; Graph newg; ......// insert data in newg v_iter vit, ve; int index = 0; for(boost::tie(vit,ve)=boost::vertices(newg); vit!=ve; ++vit ){ v_des vd = *vit; vd->id()= index++; } but the reply is: "void*" is not a pointer-to-object type at the last line. What kind of error it is?? How can I solve this problem?? Thanks Best regards -Bombielfil-

2 2

Re: [Boost-users] Boost threads.
by dariomt＠gmail.com 11 Feb '08

11 Feb '08

> Once its in the thread it seems to revert to class A's methods. Slicing occurs here: > thre.create_thread(*mea); The B object is being copied as an A object. I think it's when the temporary boost::function0<void> object passed to create_thread() is constructed. Try passing a refefence to A. e.g: // warning: untested template <typename T> struct callable_ref { T& m_ref; explicit callable_ref(T& ref) : m_ref(ref) {} void operator()() {m_ref();} }; // no copy -> no slicing // reference to A -> virtual B::operator() is called thre.create_thread( callable_ref<A>(*mea) ); By the way, I think you can't use boost::ref, as boost::reference_wrapper does not have operator() thre.create_thread( boost::ref(*mea) ); // doesn't work? Maybe TR1 can do it as std::tr1::reference_wrapper should have an appropriate operator() thre.create_thread( std::tr1::ref(*mea) ); // probably works

1 0

Regex Library
by Srinivas Chamarthi 11 Feb '08

11 Feb '08

Hi I have built boost regex library and trying to use with stlport and I encountered the following error. I dont seem to have this lib but have libs with 'gd' but not 'gdp'. any one can help ? - Error 1 fatal error LNK1104: cannot open file 'boost_regex-vc80-mt-gdp-1_34_1.lib' thanks & regards Srinivas Chamarthi

2 2

Regex Library
by Srinivas Chamarthi 11 Feb '08

11 Feb '08

Hi, I am new to C++ development and using boost regex library. I have downloaded 1_34_1 version. I have built regex library and it created boost_regex-vc80-mt-*1_34*.lib & dll files for me. But when I included headers and compiled my application but its giving me the following linker error. - *Error 1 fatal error LNK1104: cannot open file 'boost_regex-vc80-mt-gd-1_34_1.lib' ExpEval * * *so I defined *BOOST_REGEX_NO_LIB *and ran the application this time I got the following error - *This application has failed to start because MSVCP80.dll was not found. Reinstalling the application may fix this problem.* Is any one faced the similar problem ? Please help me if you know whats wrong here. Appreciate your help in advance. thanks Srinivas

2 6

Exception handler of RHEL4 isn't working after the integration of Serialization code !!!
by Arunava Saha 11 Feb '08

11 Feb '08

Hi experts, After the integration of boost serialization code, the exception handler of RHEL4 isn't working at all. Do anyone of you encountered this? Or can you point to any potential mistake for which this can happen? This is quite urgent. Thanks, Arunava.

1 0