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

win32.hpp has a using std::max, min
by Charles Schwartz 08 Sep '03

08 Sep '03

win32.hpp has using std::min and using std::max directives. (win32.hpp in fact overrides these functions). These directives hinder use of the max function elsewhere in a client program, for example, by #included c libraries, which are expecting to use the global (non overridden) max function. Indeed, Sutter's More Exceptional C++ #40 (p. 236) asserts that using directives/declerations should never appear in any headers (certainly not at file scope!). Will this be removed in future releases?

2 1

Building Boost with Intel compiler 7.0 and STLPort on windows
by Andre du Toit 08 Sep '03

08 Sep '03

Hello. Does anyone know whether it is currently possible to use Boost with the Intel Compiler v7.0 and the latest STLport, in windows? It seems I still have to hack files in date_time/posix_time (specifically time_formatters.hpp), to reduce my compile errors. Like defining BOOST_MSVC and then undeffing it, before line 43 of this file is processed. Otherwise I get ambigious operator definitions. Also, is it possible to actually build the date_time library with the Intel compiler and STLport at the same time? Why is there a msvc-stlport toolset but no intel-stlport toolset? O; another thing. Has anyone perhaps built boost in MSYS before? If MSVC + Intel setup is giving me too much hassle, I am going to try and build boost in MSYS. Greetings, Andre

3 2

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