- Boost-users - lists.boost.org

Problem with boost 1.36 on RHEL4
by Sunil Thomas 23 May '11

23 May '11

Hi all! I am struggling with some unit tests that fail only on RHEL4 release build. They pass on RHEL4 debug, RHEL5 release/debug, likewise also pass on win32, win64 (using either vc8 or vc9).. I've run the debug RHEL4 mode using valgrind, using a minimal case that reproduces the problem and here's what I see (I used the "full, no-reachable mode" to run valgrind; and I see the same output minus line #s, even running valgrind on RHEL4 release build, since that is the only platform-build combo where the test failure happens): ----------------------------------------------------------------------------------------------------------------------------- ==12285== Memcheck, a memory error detector. ==12285== Copyright (C) 2002-2005, and GNU GPL'd, by Julian Seward et al. ==12285== Using LibVEX rev 1575, a library for dynamic binary translation. ==12285== Copyright (C) 2004-2005, and GNU GPL'd, by OpenWorks LLP. ==12285== Using valgrind-3.1.1, a dynamic binary instrumentation framework. ==12285== Copyright (C) 2000-2005, and GNU GPL'd, by Julian Seward et al. ==12285== For more details, rerun with: -v ==12285== ==12285== My PID = 12285, parent PID = 12284. Prog and args are: ==12285== ./xrestunittest_xconn_rescalerd ==12285== ==12285== Syscall param sigaltstack(ss) points to uninitialised byte(s) ==12285== at 0x3AD682EDA9: sigaltstack (in /lib64/tls/libc-2.3.4.so) ==12285== by 0x6488638: boost::detail::signal_handler::~signal_handler() (in /net/ rpenas1.xcp.chevrontexaco.net/vol/vol2/data/grp2/sypb/devl/xrest/trunk/3rdp… ) ==12285== by 0x648975E: boost::execution_monitor::catch_signals(boost::unit_test::callback0<int> const&) (in /net/ rpenas1.xcp.chevrontexaco.net/vol/vol2/data/grp2/sypb/devl/xrest/trunk/3rdp… ) ==12285== by 0x6489813: boost::execution_monitor::execute(boost::unit_test::callback0<int> const&) (in /net/ rpenas1.xcp.chevrontexaco.net/vol/vol2/data/grp2/sypb/devl/xrest/trunk/3rdp… ) ==12285== by 0x648F2E4: boost::unit_test::framework::run(unsigned long, bool) (in /net/ rpenas1.xcp.chevrontexaco.net/vol/vol2/data/grp2/sypb/devl/xrest/trunk/3rdp… ) ==12285== by 0x649BD02: boost::unit_test::unit_test_main(bool (*)(), int, char**) (in /net/ rpenas1.xcp.chevrontexaco.net/vol/vol2/data/grp2/sypb/devl/xrest/trunk/3rdp… ) ==12285== by 0x4147F0: main (init.cpp:132) ==12285== Address 0x7FEFFF3B0 is on thread 1's stack ==12285== ==12285== ERROR SUMMARY: 6 errors from 1 contexts (suppressed: 4 from 1) ==12285== malloc/free: in use at exit: 190,112 bytes in 1,869 blocks. ==12285== malloc/free: 23,128 allocs, 21,259 frees, 2,520,845 bytes allocated. ==12285== For counts of detected errors, rerun with: -v ==12285== searching for pointers to 1,869 not-freed blocks. ==12285== checked 2,184,272 bytes. ==12285== ==12285== LEAK SUMMARY: ==12285== definitely lost: 0 bytes in 0 blocks. ==12285== possibly lost: 0 bytes in 0 blocks. ==12285== still reachable: 190,112 bytes in 1,869 blocks. ==12285== suppressed: 0 bytes in 0 blocks. ==12285== Reachable blocks (those to which a pointer was found) are not shown. ==12285== To see them, rerun with: --show-reachable=yes --------------------------------------------------------------------------------------------- Any ideas from boost experts here on how to resolve this problem is sincerely appreciated.. Thanks in advance, --Thomas.

1 0

What's the status of log library?
by archie14 22 May '11

22 May '11

When the log library (http://boost-log.sourceforge.net/) will be the part of official boost distribution? If there is no log library currently included in boost then what is the one most in use?

5 5

regex_replace for removing words
by Kraus Philipp 22 May '11

22 May '11

Hello, I have some texts and I would like to remove words. Each word for removing is in a std::vector<std::string> and I use a std::string for separating the words on the text. My first idea is to use the boost:split for splitting words of the text, in the result vector I remove the words, which should be removed, and recreate the text from the vector by concatination the elements. But another idea is to create a regexpr with the removing word and the separators and use regex_replace for removing. Which idea is the better one or is there another way to removing word of a text? Thanks PHil

2 2

Re: [Boost-users] [BGL] Problems with planar_face_traversal
by "Ulrich Küttler" 22 May '11

22 May '11

> >I'd love to help out - can you give a self-contained example that >doesn't use read_graphviz and your dot file (e.g., construct the graph >in the code)? Or, failing that, maybe just describe what's going wrong >and how that differs from what you expect to see? > Hi Aaron, thanks a lot. Here is a self-contained version of my example. The graph is the same. I used boost 1.46.1 on gcc linux to run it. The graph contains 22 vertices and is supposed to look like this: 0--- 4--- 6--- 8---10---12---14---16---18---20--- 1 | | | | | | | | | | | 2--- 5--- 7--- 9---11---13---15---17---19---21--- 3 Thus, there are 10 faces with 4 vertices each and one face with 22 vertices. At least that is what I suppose there should be. However, the face traversal results in the following faces (vertices): 0 12 17 16 12 0 16 18 19 17 1 13 14 15 13 1 15 14 11 9 8 10 2 3 5 7 6 4 3 2 20 18 16 17 19 21 2 4 5 3 21 20 5 4 6 8 9 7 6 7 9 11 10 8 10 11 14 13 19 18 20 21 Maybe it is a misinterpretation on my part, maybe I did make a mistake. I just don't see it. Ulrich PS: My first example lacked the edge index initialization. The above problem, however, occurs with an initialized edge index. ___________________________________________________________ Schon gehört? WEB.DE hat einen genialen Phishing-Filter in die Toolbar eingebaut! http://produkte.web.de/go/toolbar

2 1

[Serialization] serialize from and to an already instantiated object.
by nico 22 May '11

22 May '11

Hello. I use boost.serialization to perform save and load to files. If a class is serializable through the boost mechanism, I'd like to easily add the "savable" feature (through serialization and boost archives). Here how I do, and then I've got 2 questions. I don't post all the #include statements. Let's have a serializable class (I only use text archive for this example) ---------MyClass.h--------- class MyClass { friend class boost::serialization::access; template<class Archive> void serialize(Archive & ar, const unsigned int version); int data_; public: MyClass(); virtual ~MyClass(); }; BOOST_CLASS_EXPORT_KEY(MyClass) ---------MyClass.cpp--------- MyClass::MyClass() : data_(123) {} MyClass::~MyClass() {} template<class Archive> void MyClass::serialize(Archive & ar, const unsigned int version) { ar & data_; } // instanciate the serialize function for text archives #include "boost/archive/text_iarchive.hpp" #include "boost/archive/text_oarchive.hpp" template void MyClass::serialize(boost::archive::text_iarchive & ar, const unsigned int version); template void MyClass::serialize(boost::archive::text_oarchive & ar, const unsigned int version); BOOST_CLASS_EXPORT_IMPLEMENT(MyClass) --------- Now I define a FileSavable class. Its purpose is to add the "save-to-file" feature to a class that can be serialized in a text archive, making the latter inherting from FileSavable, and adding a line in its serialize function, I write it below). ---------FileSavable.h--------- class FileSavable { friend class boost::serialization::access; template<class Archive> void serialize(Archive & ar, const unsigned int version); public: virtual ~FileSavable() {} void SaveToFile(std::string path); void LoadFromFile(std::string path); }; BOOST_CLASS_EXPORT_KEY(FileSavable) ---------FileSavable.cpp--------- template<class Archive> void FileSavable::serialize(Archive & ar, const unsigned int version) {} void FileSavable::SaveToFile(std::string path) { std::ofstream ofs(path.c_str()); assert(ofs.good()); boost::archive::text_oarchive oa(ofs); oa & this; } void FileSavable::LoadFromFile(std::string path) { std::ifstream ifs(path.c_str()); assert(ifs.good()); boost::archive::text_iarchive ia(ifs); FileSavable* p; ia & p; *this = *p; delete p; } // instanciate the serialize function for text archives #include "boost/archive/text_iarchive.hpp" #include "boost/archive/text_oarchive.hpp" template void FileSavable::serialize(boost::archive::text_iarchive & ar, const unsigned int version); template void FileSavable::serialize(boost::archive::text_oarchive & ar, const unsigned int version); BOOST_CLASS_EXPORT_IMPLEMENT(FileSavable) ---------Modification to MyClass--------- It now inherit from FileSavable: class MyClass : public FileSavable The serialize function call the base class: template<class Archive> void MyClass::serialize(Archive & ar, const unsigned int version) { ar & boost::serialization::base_object<FileSavable>(*this); ar & data_; } --------- Thanks if you read me until here, now my questions: 1) I don't care the FileSavable part of MyClass being serialized, but if I don't call it with boost::serialization::base_object, I have a runtime error: terminate called after throwing an instance of 'boost::archive::archive_exception' what(): unregistered void cast 7MyClass<-11FileSavable Do you see a way to avoid having to add this line in MyClass? Ideally, just having to inheriting from FileSavable would be great: a simple way to add a feature. 2) As you can see, my LoadFromFile function first creates a new MyClass object in p, by de-serializing an archive, and then copies this newly created object into "*this", and finally deletes the object created by the archive d-eserialization. I didn't find a way to avoid this temporary and de-serialize directly in this. Moreover, my solution is not exception-safe, but I didn't manage to de-serialize in a temporary auto_ptr nor boost::shared_ptr. Thanks in advance for replies. Nicolas.

1 0

[BGL] Problems with planar_face_traversal
by "Ulrich Küttler" 22 May '11

22 May '11

Hi all, I'm new to BGL and I already like it a lot. Right now I'm searching for faces in a planar graph. I use boost::planar_face_traversal, however the results I get are confusing. The example I tested consists of 10 faces of 4 vertices each. See side-graph.dot: dot side-graph.dot -Tps > side-graph.ps My test code reads that file and calls both boyer_myrvold_planarity_test and planar_face_traversal. The graph itself seems to be fine, the face traversal fails. Is there anything I do wrong here? Thanks a lot for your help. Ulrich ___________________________________________________________ Schon gehört? WEB.DE hat einen genialen Phishing-Filter in die Toolbar eingebaut! http://produkte.web.de/go/toolbar

2 1

[icl] interval_map<> and overlapping intervals at boundaries
by sguazt 22 May '11

22 May '11

Hello, I don't understand the behavior of icl::interval_map<,> on the following scenario: 1. Before adding a new segment: {([0.179494,0.679494)->0.333333)([2.51847,3.01847)->0.333333)([3.24221,3.74221)->0.333333)} 2. Add segment: ([3.74221,4.24221)->0.333333) 3. After adding the segment: {([0.179494,0.679494)->0.333333)([2.51847,3.01847)->0.333333)([3.24221,4.24221)->0.333333)} What I would expect at step #3 is: {([0.179494,0.679494)->0.333333)([2.51847,3.01847)->0.333333)([3.24221,4.24221)->0.666667)} For the sake of clarity I provide a sample code below: --- [code] --- #include <boost/icl/interval.hpp> #include <boost/icl/interval_map.hpp> #include <iostream> #include <utility> namespace icl = ::boost::icl; int main() { typedef double time_type; typedef double value_type; typedef icl::interval<time_type> interval_type; icl::interval_map<time_type,value_type> intervals; intervals += ::std::make_pair(interval_type::right_open(0.1794936069088625, 0.6794936069088625), 0.3333333333333333); intervals += ::std::make_pair(interval_type::right_open(2.518467327901063, 3.018467327901063), 0.3333333333333333); intervals += ::std::make_pair(interval_type::right_open(3.242205294874485, 3.742205294874485), 0.3333333333333333); std::cout << "Intervals:\n" << intervals << std::endl; intervals += ::std::make_pair(interval_type::right_open(3.742205294874485, 4.242205294874486), 0.3333333333333333); std::cout << "Updated Intervals:\n" << intervals << std::endl; } --- [/code] --- What kind of interval container should I use in order to obtain the wanted result? I'm using the SVN trunk version of boost with GCC 4.5.1 (linux x86_64) Thank you very much for helping me! Best, -- Marco

1 1

program_options: pointer being freed was not allocated
by Matan Nassau 21 May '11

21 May '11

Hi! When I run the following program, #include <boost/program_options.hpp> int main() { boost::program_options::positional_options_description positional; return 0; } the program aborts with the error sandbox(1791) malloc: *** error for object 0x7fff70506500: pointer being freed was not allocated *** set a breakpoint in malloc_error_break to debug My platform is Mac OS X 10.6.7. I installed Boost 1.46.1 on my own, like so: ./bootstrap --prefix=/usr/local ./bjam memory-model=64 ./bjam install The toolset bjam chose was darwin, which uses gcc-4.0 if I understood the build correctly. I also tried rebuilding boost with ./bjam memory-model=64 toolset=darwin-4.2 but the issue persisted. Any idea what causes this crash? Matan

1 1

Re: [Boost-users] [boost][msm] std::vector<MyStateMachine> generate stack overflow
by Christophe Henry 20 May '11

20 May '11

> [Christophe Henry] >> Maybe Stephan is aware of this problem? > > I haven't heard of this problem before. (Is a templated constructor involved? They can be squirrely.) There are quite a few ones, yes. But I'd suspect a problem with inheriting. I got quite a few problems with getting a wrong type when inheriting from state_machine. Specifically, the compiler saw the derived type (we call it Derived) sometimes as Derived, sometimes as state_machine<Derived...>. Regards, Christophe > Stephan T. Lavavej > Visual C++ Libraries Developer

1 0

[hash] hash returning fixed width integer instead of size_t?
by Frédéric Bron 20 May '11

20 May '11

I would like this code return the same hash on different machines: #include <iostream> #include <boost/functional/hash.hpp> int main() { boost::hash<std::string> hasher; std::cout<<hasher("abcd"); return 0; } But I get 1244709680 when size_t is 32 bits and 706246339511088 when size_t is 64 bits. Is it possible to have hash work with fixed width uint32_t instead of size_t? Frédéric

2 4