July 2003 - Boost-users - lists.preview.boost.org

shared_ptr or weak_ptr in general
by Jason Winnebeck 22 Jul '03

22 Jul '03

I just wanted to follow up on my previous thread about my issues I've been discussing on the list with a general question. By default should I try to use shared_ptr or weak_ptr? I had always thought that shared_ptr should always be used except for rare cases where weak_ptr's semantics is either very desirable or needed. Now my pointers experience so far has been limited to explicit memory management (new/delete) and in Java with its references, where any reference keeps an object alive, and weak references exist but I've never used them despite a few years of using Java on the job. Based on some of the discussion (particularly from Mr Dimov), made me think that weak_ptr may be desirable a lot more often than I previously thought. I know that shared_ptr, unlike Java, is suceptable to cycles. And I feel that shared_ptr is better and less bug prone than new/delete but I'm afraid I'll have cycle problems if I just start throwing shared_ptr all over the place. I know that in my code I have back pointers all over the place, which I thought was pretty common, but I have realized that that only occurs when I'm following the generator/listener design scheme, and I've gotten that figured out. So A to B, B to A cycles don't concern me much but I do wonder about long cycles if I use shared_ptr a lot. The rule of thumb I'm using right now is to look for cycles if I "convert" a class to use shared_ptr (static crate method) that contains shared_ptr objects, similar to the rule I use to check for deadlocks with mutexes. But with all of these rules checking for safety I begin to question somewhat the benefits of shared_ptr over normal pointers. I suppose with shared_ptr you can only leak memory whereas with normal pointers you can do lots of nasty uninitialized accesses. I can see an argument that using shared_ptr only when needed and weak_ptr otherwise can help prevent cycles. But using weak_ptr requires the code to handle both cases of the lock operation, and I'm still unsure about any performance issues with lock if there is a mutex involved -- although too slight to not choose weak_ptr, it is enough to ask why not choose shared_ptr over weak_ptr if they both work equally well. Thank you for your comments and patience with my attemps at understanding, Jason Winnebeck

1 0

staticially linked STLport and Boost
by hh10k 20 Jul '03

20 Jul '03

I've compiled STLport for vc6, and can compile Boost with this: bjam "-sTOOLS=msvc-stlport" "-sBUILD=debug <stlport-iostream>on" So far so good, but when I try to compile a simple Boost.Python example (why I got Boost) I get a heap of link errors because Boost has linked with the DLL version of STLport, and I want it statically linked. I thought it's meant to default to static, but I can only guess that STLport is detecting otherwise. So, in order to force Boost to use the static verion, I tried: bjam "-sTOOLS=msvc-stlport" "-sBUILD=debug <stlport-iostream>on <define>_STLP_USE_STATIC_LIB" But now I get errors like: vc-Link libs\python\build\bin\boost_python.dll\msvc-stlport\debug\runtime-link-dynamic\stlport-cstd-namespace-std\stlport-iostream-on\stlport-version-4.5.3\boost_python_debug.dll libs\python\build\bin\boost_python.dll\msvc-stlport\debug\runtime-link-dynamic\stlport-cstd-namespace-std\stlport-iostream-on\stlport-version-4.5.3\boost_python_debug.lib msvcprtd.lib(MSVCP60D.dll) : error LNK2005: "public: __thiscall std::bad_alloc::bad_alloc(char const *)" (??0bad_alloc@std@@QAE@PBD@Z) already defined in stlport_vc6_stldebug_static.lib(dll_main.obj) LIBCMTD.lib(osfinfo.obj) : error LNK2005: __get_osfhandle already defined in MSVCRTD.lib(MSVCRTD.dll) LIBCMTD.lib(dbgheap.obj) : error LNK2005: _malloc already defined in MSVCRTD.lib(MSVCRTD.dll) ...lots more errors... This isn't just a problem with Boost.Python. Anything else that builds a dll or exe has the same errors. The first error seems really strange, because I can't understand how std::bad_alloc should exist in stlport_vc6_stldebug_static.lib(dll_main.obj). STLport does define it's own bad_alloc, but it also #defines std to _STL so there should be no 'std'. When I use bad_alloc in my own programs I don't have any problems, so I can only assume it's Boost. I've wasted so much time trying to work this out, but I don't know much about the Boost build process. What's going wrong?

3 3

Unusal regex behavior.
by sdf sdf 20 Jul '03

20 Jul '03

Hellos everyone! Here is a good one. I am using regex and am having some weird returns. I can assign my results to a std:vector <std:string>. The vector[0] does contain the correct match, but vector[1] does not contain what is in a set of parentheses as it should. In fact I usually get a segment fault when I try to access it. I then found I can access results from the grep by using the boost::Regex.What(n). However What(0) returns the next available match in the string, and What(1) and up will correctly report the substrings defined in this second match correctly. If no second match is available then the What() member will return the original match. How do I fix the vector problem? OR? Does using What() force another search to take place? And if so can I change that behavior? Code snippet: string reg_string ="[0-9]{3}(-[0-9]{2}-)[0-9]{4}"; //yes it is a SSN string search_string = "Big Bird \n123 sesame street \n the bronx NY 11722\n545-65-2343\nmorestuff\nandevenmore\nSnuff E Lufigous \n 234 sesame street \n the bronx NY 11722\n889-99-5527\nmorestuff\nandevenmore\n"; std::vector<std::string> results; boost::regex regsearch = boost::regex(reg_string,false); int num_matchs; num_matchs=regsearch.Grep(results, search_string, boost::match_any); cout << num_matchs << " Number of Matchs\n"; // this is 2 cout << results[0] << " results return string vector[0] \n"; // this is 545-65-2343 cout << regsearch.What(0) << " results from what(0) \n"; // this is 889-99-5527 cout << regsearch.What(1) << "results from what(1) \n";// this is -99- //this line aborts // cout << results[1] << " results return string vector[1] \n"; // this should be -65- HELP. Note I never ran the code above, but I think it is correct. __________________________________ Do you Yahoo!? SBC Yahoo! DSL - Now only $29.95 per month! http://sbc.yahoo.com

4 5

filesystem help
by Christian Gmeiner 20 Jul '03

20 Jul '03

Hello! This is my problem: I have an std::string contains something like C:\Test\Media\Cloud.dds. Now i want to extract the Cloud.dds. This is, what i have tried: fs::path temp(Name, fs::native); fs::directory_iterator dir_itr(temp); Name = dir_itr->leaf(); Where is my error? This code throughs an exception. Thanks, Christian Gmeiner

2 1

Multiple inheritance and enable_shared_from_this
by Jason Winnebeck 20 Jul '03

20 Jul '03

Can you multiply inherit from objects that inherit from enable_shared_from_this? It seems not. My program compiled on MSVC.NET "crashes" due to shared_from_this throwing boost::bad_weak_ptr. Originally when I started using shared_ptr I wanted to have an ability to obtain a shared_ptr to this since my objects sometimes register themselves. The way I do it now is like this: static sptr create() { sptr ret( new Object() ); ret->setThisPtr( ret ); return ret; } However it is very easy to forget which classes you inherit from that force you to call setThisPtr. A good use of the assert statement helps but I'd rather not do this. The problem is that I use multiple inheritance in one place: static sptr create() { sptr ret( new ComplexObject() ); ret->BaseA::setThisPtr( ret ); ret->BaseB::setThisPtr( ret ); return ret; } I thought I'd use enable_shared_from_this but as I said it crashes. The following program as tested on MSVC.NET throws boost::bad_weak_ptr then the line "Base2::sptr b2 = c->Base2::getThis();" is executed. #include <iostream> #include <boost/smart_ptr.hpp> #include <boost/enable_shared_from_this.hpp> using namespace std; template <class T> class SharedThis : public boost::enable_shared_from_this<T> { public: boost::shared_ptr<T> getThis() { return shared_from_this(); } }; class Base1 : public SharedThis<Base1> { public: typedef boost::shared_ptr<Base1> sptr; typedef boost::weak_ptr<Base1> wptr; protected: Base1() { cout << "Base1 ctor" << endl; } public: virtual ~Base1() { cout << "Base1 dtor" << endl; } static sptr create() { return sptr( new Base1() ); } private: Base1( const Base1& o ); Base1& operator = ( const Base1& rhs ); }; //Base2 is identical to Base1 except different name class Base2 : public SharedThis<Base2> { public: typedef boost::shared_ptr<Base2> sptr; typedef boost::weak_ptr<Base2> wptr; protected: Base2() { cout << "Base2 ctor" << endl; } public: virtual ~Base2() { cout << "Base2 dtor" << endl; } static sptr create() { return sptr( new Base2() ); } private: Base2( const Base2& o ); Base2& operator = ( const Base2& rhs ); }; class Child : public Base1, public Base2 { public: typedef boost::shared_ptr<Child> sptr; typedef boost::weak_ptr<Child> wptr; protected: Child() { cout << "Child ctor" << endl; } public: virtual ~Child() { cout << "Child dtor" << endl; } static sptr create() { return sptr( new Child() ); } }; int main( int argc, char* argv[] ) { Child::sptr c = Child::create(); Base1::sptr b1 = c->Base1::getThis(); Base2::sptr b2 = c->Base2::getThis(); /* assert( c.get() == c->Base1::getThis().get() ); assert( c.get() == c->Base2::getThis().get() ); assert( c == c->Base1::getThis() ); assert( c == c->Base2::getThis() ); */ }

3 8

signals through Python
by mat_2775 19 Jul '03

19 Jul '03

Hi, I'm very new to BOOST so please forgive my ignorance. I want to use the signals library of BOOST through Python. My question is whether that is actually possible and if so how? Kind regards Mat

2 1

Bind To Lambda
by Richard Crossley 18 Jul '03

18 Jul '03

Hi, I'm trying merge some code some of which uses a mixture of BB and BLL to just use the BLL under VC7.1. Having changed some 200 expressions without incident if find I can't see the wood for the trees here. The existing code, working with BB, is; typedef boost::function<void(ICollection*)> FECCFunc; void ForEachCollectionInChannels(IAmplifier* pIAmp,FECCFunc const & fec); typedef boost::function<void(IProperty*)> FEPFunc; void ForEachProperty(ICollection* pCollection,FEPFunc const & fep); void CalcPropCurValMaxWidth(IProperty* pProp,CColumnMap& map); ... void Problem() { CColumnMap m_ChanCols; IAmplifier* pIAmp; ForEachCollectionInChannels(pAmp,bind(ForEachProperty,_1,cref(bind(&CalcProp CurValMaxWidth,_1,ref(m_ChanCols)))); //Here } What's the equivalent with using BLL? Thanks, Richard.

1 1

StlPort-4.5.3, ublas and MSVC 6.0 problem
by Bosko Ivanisevic 18 Jul '03

18 Jul '03

Although I found on the site that UBLAS supportes MSVC 6.0 with STLPort I couldn't compile any example with this configuration. It seems that using Boost ublas library with StlPort (with native MS streams) and MSVC 6.0 have some problems and require specific settings. Does anyone have any hint how this problem can be solved? When I try to compile some of the samples from libs\numeric\ublas\doc \samples (for example vecto_binary.cpp) I got following errors: c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(1534) : error C2027: use of undefined type 'const_iterator1' c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(1529) : see declaration of 'const_iterator1' c:\projects\test\boost_matrices\main.cpp(23) : see reference to class template instantiation 'boost::numeric::ublas::hermitian_adaptor <class boost::numeric::ublas::matrix<class std::complex<double>,struct boost::numeric::ublas::row_major,cla ss boost::numeric::ublas::unbounded_array<class std::complex<double> > >,struct boost::numeric::ublas::lower>' being compiled c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(1831) : error C2027: use of undefined type 'iterator1' c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(1826) : see declaration of 'iterator1' c:\projects\test\boost_matrices\main.cpp(23) : see reference to class template instantiation 'boost::numeric::ublas::hermitian_adaptor <class boost::numeric::ublas::matrix<class std::complex<double>,struct boost::numeric::ublas::row_major,cla ss boost::numeric::ublas::unbounded_array<class std::complex<double> > >,struct boost::numeric::ublas::lower>' being compiled c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(1950) : error C2027: use of undefined type 'const_iterator2' c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(1945) : see declaration of 'const_iterator2' c:\projects\test\boost_matrices\main.cpp(23) : see reference to class template instantiation 'boost::numeric::ublas::hermitian_adaptor <class boost::numeric::ublas::matrix<class std::complex<double>,struct boost::numeric::ublas::row_major,cla ss boost::numeric::ublas::unbounded_array<class std::complex<double> > >,struct boost::numeric::ublas::lower>' being compiled c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(2247) : error C2027: use of undefined type 'iterator2' c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(2242) : see declaration of 'iterator2' c:\projects\test\boost_matrices\main.cpp(23) : see reference to class template instantiation 'boost::numeric::ublas::hermitian_adaptor <class boost::numeric::ublas::matrix<class std::complex<double>,struct boost::numeric::ublas::row_major,cla ss boost::numeric::ublas::unbounded_array<class std::complex<double> > >,struct boost::numeric::ublas::lower>' being compiled c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(1534) : error C2027: use of undefined type 'const_iterator1' c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(1529) : see declaration of 'const_iterator1' c:\projects\test\boost_matrices\main.cpp(30) : see reference to class template instantiation 'boost::numeric::ublas::hermitian_adaptor <class boost::numeric::ublas::matrix<class std::complex<double>,struct boost::numeric::ublas::row_major,cla ss boost::numeric::ublas::unbounded_array<class std::complex<double> > >,struct boost::numeric::ublas::upper>' being compiled c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(1831) : error C2027: use of undefined type 'iterator1' c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(1826) : see declaration of 'iterator1' c:\projects\test\boost_matrices\main.cpp(30) : see reference to class template instantiation 'boost::numeric::ublas::hermitian_adaptor <class boost::numeric::ublas::matrix<class std::complex<double>,struct boost::numeric::ublas::row_major,cla ss boost::numeric::ublas::unbounded_array<class std::complex<double> > >,struct boost::numeric::ublas::upper>' being compiled c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(1950) : error C2027: use of undefined type 'const_iterator2' c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(1945) : see declaration of 'const_iterator2' c:\projects\test\boost_matrices\main.cpp(30) : see reference to class template instantiation 'boost::numeric::ublas::hermitian_adaptor <class boost::numeric::ublas::matrix<class std::complex<double>,struct boost::numeric::ublas::row_major,cla ss boost::numeric::ublas::unbounded_array<class std::complex<double> > >,struct boost::numeric::ublas::upper>' being compiled c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(2247) : error C2027: use of undefined type 'iterator2' c:\programming\boost_1_30_0\boost\numeric\ublas\hermitian.hpp(2242) : see declaration of 'iterator2' c:\projects\test\boost_matrices\main.cpp(30) : see reference to class template instantiation 'boost::numeric::ublas::hermitian_adaptor <class boost::numeric::ublas::matrix<class std::complex<double>,struct boost::numeric::ublas::row_major,cla ss boost::numeric::ublas::unbounded_array<class std::complex<double> > >,struct boost::numeric::ublas::upper>' being compiled c:\projects\test\boost_matrices\main.cpp(37) : warning C4508: 'main' : function should return a value; 'void' return type assumed Error executing cl.exe. Does anyone have any hint how this problem could be solved?

1 0

boost and Watcom
by filbico 18 Jul '03

18 Jul '03

Hello, I would like to use boost 1.29 under Watcom C++ complier (11.0C, Powersoft Optima 2.1). I am looking for configuration files. Robert Nowak.

1 0

Replace/Extend default execution_monitor
by aragorn_yes 18 Jul '03

18 Jul '03

Hi all, What is the best way to replace/extend the default execution_monitor (unit_test_monitor) in the test library. Thanks in advance. Manuel.

2 1