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

lambda and smart pointers
by Rodolfo Lima 08 Sep '03

08 Sep '03

What's the proper way to use lambda with smart pointers? I'm trying to compile the following: class Dados { public: Dados(int _a) : a(_a) {}; int Pega() const { return a; } private: int a; }; int main() { shared_ptr<Dados> pDados(new Dados(5)); cout << (bind(&Dados::Pega, _1)(pDados)==5) << endl; return 0; } I've already tried with *_1 with no success. In this case, the compiler (VC7.1) returns "error C2039: 'iterator_category' : is not a member of 'boost::shared_ptr<T>'". Any hope? Thanks, rod

1 0

Compile errors boost 1.30.2
by Andre du Toit 08 Sep '03

08 Sep '03

Hi. I get the following error when compiling the latest version of boost, using Intel compiler 7.0 in visual c++: ..\..\..\boost\boost/type_traits/is_integral.hpp(39): error: class "boost::is_integral<unsigned short>" has already been defined BOOST_TT_AUX_BOOL_TRAIT_CV_SPEC1(is_integral,wchar_t,true) Now the offending line sits between a "#ifndef BOOST_NO_INTRINSIC_WCHAR_T". The simple fix of defining BOOST_NO_INTRINSIC_WCHAR_T solves this problem, but causes other problems in other locations. All I want is to compile and link against boost::date_time library, using my compiler setup. Thats all. Any suggestions? Greetings, Andre

1 0

RE: [Boost-users] pyste and incomplete types
by Jason.Sibthorpe＠aculab.com 05 Sep '03

05 Sep '03

I have just checked out the 'latest' version from cvs. I notice that the 'problem' has been fixed. Thanks -J -----Original Message----- From: Jason.Sibthorpe(a)aculab.com [mailto:Jason.Sibthorpe@aculab.com] Sent: 05 September 2003 17:21 To: boost-users(a)lists.boost.org Subject: [Boost-users] pyste and incomplete types Hi, I have encountered a problem with pyste and incomplete types. From earlier posts and the doco I think that pyste should generate BOOST_PYTHON_OPAQUE_SPECIALIZED_TYPE_ID(it) But is does not currently (boost boost-1.30.2). invoking bjam without the BOOST_PYTHON_OPAQUE_SPECIALIZED_TYPE_ID(it) results in the output attached (output1) with BOOST_PYTHON_OPAQUE_SPECIALIZED_TYPE_ID(it) results in the output attached (output2) In either case It does not build. Am I misunderstanding something? Thanks Jason //////////////////////////////////////////////////////////////////////////// /// //// incomplete_type.h //////////////////////////////////////////////////////////////////////////// /// typedef struct incomplete_type it; it *func1(void); //////////////////////////////////////////////////////////////////////////// /// //// incomplete_type.pyste //////////////////////////////////////////////////////////////////////////// /// itype = AllFromHeader("incomplete_type.h") set_policy(itype.func1, return_value_policy('return_opaque_pointer')) // Note the stringified 'return_opaque_pointer' //////////////////////////////////////////////////////////////////////////// /// //// itype.cpp - pyste output: //////////////////////////////////////////////////////////////////////////// /// // Includes =================================================================== #include <boost/python.hpp> #include <incomplete_type.h> // Using ====================================================================== using namespace boost::python; // Module ===================================================================== BOOST_PYTHON_MODULE(itype) { def("func1", &func1, return_value_policy< return_opaque_pointer >()); class_< incomplete_type, boost::noncopyable >("incomplete_type", no_init) ; } Jason Sibthorpe Software Engineer Aculab Tel: +44 (0) 1908 273 869 Fax: +44 (0) 1908 273 801 Email: jason.sibthorpe(a)aculab.com Website: <http://www.aculab.com>

1 0

Re: [Boost-Users] ostream_iterator and std::copy - better solutions?
by Victor A. Wagner, Jr. 05 Sep '03

05 Sep '03

is this what you had in mind?? I've considered this from time to time (first time I've formalized it tho) especially thinking of alternatives to the 3 delimiter arguments (substitute one? "{,}" ??) or what-have-you maybe we should boostify it? template <typename InputItr> ostream& range_streamer(ostream& to, InputItr start, InputItr stop, const char* sdelim, const char* mdelim, const char* edelim) { // output the beginning delimiter to << sdelim; // output the 1st element if it exists if (start != stop) to << *start++; // output the rest of the list (if any) separated by mdelim while (start != stop) { to << mdelim << *start; ++start; } //output the ending delimiter and return return to << edelim; } and of course, if you're going to do it, do it all the way: template <typename Container> inline ostream& container_streamer(ostream& to, Container const& cont, const char* sdelim, const char* mdelim, const char* edelim) {return range_streamer(to, cont.begin(), cont.end(), sdelim, mdelim, edelim);} At Wednesday 2003-09-03 09:07, you wrote: >Before rolling my own, does boost have a facility to stream out an iterator >range without the dangling delimiter at the end? For example > > >std::ostream& operator<<( std::ostream& s, const MyClass& aC ) >{ > const std::string lDelim( "," ); > > std::vector<int> lInts; > > lInts.push_back(1); > lInts.push_back(2); > lInts.push_back(3); > > std::ostream_iterator<int> lItr( s, lDelim.c_str() ); > > std::copy( lInts.begin(), lInts.end(), lItr ); >} > >yields: > >1,2,3, > >I'm thinking something like the following usage would be nice: > > s << range_streamer( lInts.begin(), lInts.end(), "{", "," "}" ); > >yielding: > >{1,2,3} > >Any thoughts? > > >-- > >Jeff Flinn >Applied Dynamics, International Victor A. Wagner Jr. http://rudbek.com The five most dangerous words in the English language: "There oughta be a law"

2 2

pyste and incomplete types
by Jason.Sibthorpe＠aculab.com 05 Sep '03

05 Sep '03

Hi, I have encountered a problem with pyste and incomplete types. From earlier posts and the doco I think that pyste should generate BOOST_PYTHON_OPAQUE_SPECIALIZED_TYPE_ID(it) But is does not currently (boost boost-1.30.2). invoking bjam without the BOOST_PYTHON_OPAQUE_SPECIALIZED_TYPE_ID(it) results in the output attached (output1) with BOOST_PYTHON_OPAQUE_SPECIALIZED_TYPE_ID(it) results in the output attached (output2) In either case It does not build. Am I misunderstanding something? Thanks Jason //////////////////////////////////////////////////////////////////////////// /// //// incomplete_type.h //////////////////////////////////////////////////////////////////////////// /// typedef struct incomplete_type it; it *func1(void); //////////////////////////////////////////////////////////////////////////// /// //// incomplete_type.pyste //////////////////////////////////////////////////////////////////////////// /// itype = AllFromHeader("incomplete_type.h") set_policy(itype.func1, return_value_policy('return_opaque_pointer')) // Note the stringified 'return_opaque_pointer' //////////////////////////////////////////////////////////////////////////// /// //// itype.cpp - pyste output: //////////////////////////////////////////////////////////////////////////// /// // Includes =================================================================== #include <boost/python.hpp> #include <incomplete_type.h> // Using ====================================================================== using namespace boost::python; // Module ===================================================================== BOOST_PYTHON_MODULE(itype) { def("func1", &func1, return_value_policy< return_opaque_pointer >()); class_< incomplete_type, boost::noncopyable >("incomplete_type", no_init) ; } Jason Sibthorpe Software Engineer Aculab Tel: +44 (0) 1908 273 869 Fax: +44 (0) 1908 273 801 Email: jason.sibthorpe(a)aculab.com Website: <http://www.aculab.com>

1 0

Fw: Unable to deliver your message
by Jeff F 05 Sep '03

05 Sep '03

I keep getting the following bounce messages when trying to post to gmane.comp.lib.boost.user via nntp. Something not running right yet? Something not yet set on gmane? Or is this not possible with the new setup? Jeff Flinn Applied Dynamics, International ----- Original Message ----- From: "Yahoo! Groups" <notify(a)yahoogroups.com> To: <yg-boost-users(a)m.gmane.org> Sent: Thursday, September 04, 2003 5:32 PM Subject: Unable to deliver your message > > We are unable to deliver the message from <yg-boost-users(a)m.gmane.org> > to <boost-users(a)yahoogroups.com>. > > Posting to this group has been temporarily disabled. > > For further assistance, please email support(a)yahoogroups.com > or visit http://help.yahoo.com/help/us/groups/ >

3 4

[BGL] adjacency property iterator
by sashan 05 Sep '03

05 Sep '03

Hi Is there an adjancency iterator for property types. You can get an iterator for property types like this: typedef graph_property_iter_range<GraphType, vertex_name_t>::iterator VNameIterator; .... tie(It, ItEnd) = get_property_iter_range(m_Graph, vertex_name); I'd like to know if there is something similar for adjacency iterators. I posted this question a few days ago but got no response so I guess that the answer is no but I want to make sure. Thanks -- sashan http://www.cs.auckland.ac.nz/~sgov008/

1 0

using boost::mem_fn result as a parameter?
by Mark Sizer 04 Sep '03

04 Sep '03

I'm new to boost::bind and boost::mem_fn. I've read the documentation and I think I understand what they do, but I'm not sure how to use them to accomplish my task. I have a class that contains several vectors in a map. I want to put a "foreach" method on the class that will callback an arbitrary member function, of an arbitrary class, for each item in the vector. I'm stuck on the signature of that method. A quick sketch of the structure: class VectorContent { public: void dosomething(); // there are bunches more, which is why I don't // want a foreach for every possible choice. }; class VectorHolder { std::map<std::string, std::vector<VectorContent> >; std::vector<VectorContent>& rvectGet(const std::string& rksName); public: void foreach( const std::string& rksName, <WHAT?> ); }; // the callback Caller::Bar( VectorContent& vc ) { vc.dosomething(); } I see two options for the caller: // option 1 Caller::Foo1() { VectorHolder callee; // in this case, foreach uses boost::mem_fn(), not me callee.foreach( "VectorName", &this::Bar ); } // option 2 Caller::Foo2(); { VectorHolder callee; callee.foreach( "VectorName", boost::mem_fn(&this::Bar) ); } In the case of option 1, all callers must share a common base class, which they do NOT, so that the signature of VectorHolder::foreach can be: typedef void (BaseClass::* MyCallBack)(VectorContent&); Leading to: void VectorHolder::foreach( const std::string& rsName, MyCallBack callback ) { std::vector<VectorContent>& rvect = rvectGet( rsName ); std::for_each( rvect.begin(), rvect.end(), boost::mem_fn(callback) ); } In the case of option 2, I have no idea what the signature should be. But the resulting method would be: void VectorHolder::foreach( std::string& rsName, WHAT? callback ) { std::vector<VectorContent>& rvect = rvectGet( rsName ); std::for_each( rvect.begin(), rvect.end(), callback ); } Grateful for any help, - Mark P.S. If there's an "option 3", I'm up for that, too.

2 1

Fw: [Boost-Users] Pb with threads on Mac OS X Carbon.
by Peter Dimov 04 Sep '03

04 Sep '03

Peter Dimov wrote: > degaine1966 wrote: >> Hi, >> >> I'd like to use shared pointers with CodeWarrior 8.3 in a Carbon >> application. I have some preprocessing errors, the first one being >> relative to a thread Macro being undefined: >> >> lightweight_mutex.hpp line 86 # error Unrecognized threading >> platform >> >> >> I read in boot/config/plateform/macos.hpp that Boost was thread safe >> under Carbon but apparently there is no macro to define the usable >> type of thread... > > Boost.Threads has a Mac Carbon port, but unfortunately > lightweight_mutex does not, and nobody familiar with Carbon has > contributed one.

2 1

List host change
by Jon Kalb 04 Sep '03

04 Sep '03

Boost Users, We are moving the Boost-Users mailing list from Yahoo! Groups to a mailman based system. This means that effective immediately mail for Boost Users should be send to boost-users(a)lists.boost.org. You can manage your own subscriptions at http://lists.boost.org/mailman/listinfo.cgi/boost-users . If you have been subscribed to this list receiving individual mail messages or digest messages, then we have subscribed you to the mailman list with the same settings. If you were subscribed with either "no email" or "special emails only" then you will need to re-subscribe to the list by visiting http://lists.boost.org/mailman/listinfo.cgi/boost-users . Note that in order to send messages to this list you must be subscribed to the list. This does not mean that you must receive email from the list. You can use the "no email" setting and read and reply by newsgroup, but you must be a subscriber or your messages will not be accepted. If you have any questions or problems please contact the list moderators at boost-users-owner(a)lists.boost.org. Please note that questions and problems related to the list are off topic on this list and should only be sent to the list moderators. -- Jon Kalb SmartFriends(TM) U: Languages and Libraries Kalb(a)LibertySoft.com Sept. 26-28 http://SmartFriends.com/U Keynote: STL Creator, Alexander Stepanov

1 0