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

Failed compiling luabind using boost on AIX (xlC): configuration problem?
by Christoph Jaeschke 20 Jan '05

20 Jan '05

Hi all, I failed compiling luabind using boost 1.3.2 on AIX 5.1 (xlC) xlC -c class_rep.cpp -o class_rep.o ... -I.. -I/some_dir/include/boost-1_32 "/some_dir/include/boost-1_32/boost/mpl/aux_/O1_size_impl.hpp", line 33.38: 1540-0827 (S) The constant expression on the #if directive contains a syntax error at "BOOST_TESTED_AT". "/some_dir/include/boost-1_32/boost/mpl/aux_/include_preprocessed.hpp", line 31.13: 1540-0836 (S) The #include file "boost/mpl/aux_/preprocessed/AUX_PREPROCESSED_HEADER" is not found. "/some_dir/include/boost-1_32/boost/mpl/aux_/include_preprocessed.hpp", line 31.13: 1540-0836 (S) The #include file "boost/mpl/aux_/preprocessed/AUX_PREPROCESSED_HEADER" is not found. -- <snipped> -- To me it looks like a configuration problem, but bjam nicely installed the boost-base as specified without any errors. Any help is very appreciated! Christoph

1 0

Problems with std::min when I switched from 1.31 to 1.32
by Scott Moore 20 Jan '05

20 Jan '05

I just switched from boost 1.31 to 1.32 and now any code that uses std::min() or std::max() gives me this error with Visual Studio 2003: ReportResults.cpp(791) : error C2589: '(' : illegal token on right side of '::' ReportResults.cpp(791) : error C2059: syntax error : '::' I even tried something simple like this: int i = std::min(1, 2); And that fails as well. If I switch back to boost 1.31, no problems. I'm sure there must be a simple solution. Any help would be appreciated! Thanks, Scott

5 6

boost::bind and methods of managed classes
by Vasco Lohrenscheit 19 Jan '05

19 Jan '05

Hi, is there a easy way/workaround to use methods of managed classes with boost::bind? boost::signal<void (int value)> testSignal; public __gc class TestManagedClass { public: TestManagedClass() {testSignal.connect(boost:bind(???)); } void test(int i); }; how can I connect the TestManagedClass::test method of a instance of a TestManagedClass to the signal ? It seem already not be allowed to get the adress of the method &OpenGlTestGUI::testMethod (compile error c2843, vc++7.1). Are there any work arounds? Has anybody alread used boost::bind and boost::signal with managed classes? best regards, Vasco Lohrenscheit

3 4

BOOST_FOREACH is not working under VC7.0
by Yaroslav Yanovsky 19 Jan '05

19 Jan '05

Hello boost-users, I know, BOOST_FOREACH is not distributed with boost, but I don't know where else should I write about it so I decided to post it here. BOOST_FOREACH is not compiling under VC 7.0 Here I've wrote small example that represents problem: ----------------------- #include <iostream> #include <list> #include <boost/foreach.hpp> using std::list; class BNamespace { public: list<int> contents; int Find(int what) const; }; inline int BNamespace::Find(int what) const { BOOST_FOREACH (int i, this->contents) //BOOST_FOREACH (const int& i, this->contents) // will fail too if (i==what) return i; return 0; } int main() { BNamespace lst; lst.contents.push_back(4); lst.contents.push_back(5); lst.contents.push_back(6); int i=lst.Find(6); std::cout<<i<<std::endl; return 0; } ----------------------- Compilation of this code on VC 7.0 leads to following errors: a.cpp(15) : error C2667: 'boost::for_each::wrap' : none of 2 overloads have a best conversion f:\Libs\boost\boost\foreach.hpp(138): could be 'boost::for_each::container<T,boost::mpl::true_> boost::for_ each::wrap(const T &)' f:\Libs\boost\boost\foreach.hpp(132): or 'boost::for_each::container<T,boost::mpl::false_> boost::for _each::wrap(T &)' while trying to match the argument list '(const std::list<_Ty,_Ax>)' with [ _Ty=int, _Ax=std::allocator<int> ] a.cpp(15) : error C2668: 'boost::for_each::wrap' : ambiguous call to overloaded function f:\Libs\boost\boost\foreach.hpp(138): could be 'boost::for_each::container<T,boost::mpl::true_> boost::for_ each::wrap(const T &)' f:\Libs\boost\boost\foreach.hpp(132): or 'boost::for_each::container<T,boost::mpl::false_> boost::for _each::wrap(T &)' while trying to match the argument list '(const std::list<_Ty,_Ax>)' with [ _Ty=int, _Ax=std::allocator<int> ] ----------------------- I've tried to compile same code with MinGW and it succeded. Can anyone help me with this problem? Thanks! -- Best regards, Yaroslav mailto:yaroslav@enkord.com

2 2

Re: [Boost-users] Re: (de)serializing vector<const A*>
by Zoltan Szatmary 19 Jan '05

19 Jan '05

Dear Robert Thank you for answering. I am sorry I was not very specific (though please don't think I would put the question without RTFM first ;) ) The part of documentation you pointed to told me to do const_cast manually. It is fine in case of const members of a class, but this time the const type is contained by a vector. I could manually write the serialization code for my vector<const A*> type but I hoped I could use the built-in vector serialization templates in vector.hpp To be clear I copy here the piece of code I am talking about #include <fstream> #include <boost/archive/text_iarchive.hpp> #include <boost/archive/text_oarchive.hpp> #include <boost/serialization/vector.hpp> using namespace std; class A { friend class boost::serialization::access; template<class Archive> void serialize (Archive &ar, const unsigned int version) { } }; /* NEITHER OF THESE HELPS namespace boost { namespace serialization { template <class Archive> inline void serialize (Archive &ar, const A *p, const unsigned int version) { ar & const_cast<A*>(p); } template <class Archive> inline void serialize (Archive &ar, const A &p, const unsigned int version) { ar & const_cast<A&>(p); } }} */ int main() { A a; vector<const A*> v(1, &a); ofstream ofs ("vectorofconstptr"); boost::archive::text_oarchive oa(ofs); oa << v; ofs.close(); // UNTIL THIS POINT EVERYTHING IS FINE // BUT DESERIALIZATION FAILS TO COMPILE ifstream ifs("vectorofconstptr"); boost::archive::text_iarchive ia(ifs); vector<const A*> p; ia >> p; ifs.close(); } Please tell me if I am doing something wrong. I tried to look into the documentation, I have even followed the serialization library code for a while, but eventually decided to ask. Thank you in advance Zoltan >Look in the documentation under Reference/Class Serialization/const members >Robert Ramey >Zoltan Szatmary wrote: >> Hello >> >> could you help please? what is the easiest way to (de)serialize a >> member of type vector<const A*> (A being a base class)? Serializing >> is fine, but deserialization fails to compile since the vector >> deserialization code tries to deserialize a const A* pointer without >> const_cast-ing it (I guess). >> >> Thank you in advance >> >> Zoltan

2 1

long + bigint
by Benjamin Schmeling 19 Jan '05

19 Jan '05

Hi, I am wrapping a class "bigint" from c++ to python. I have overloaded the add method to do something like this on python: >>>n=bigint(5); >>>n+5L 5 This works well, but if I try to do it vice versa my overloaded add method isn't called by python: >>>n=bigint(5); >>>5L+n Causes error. My add methods look like this: bigint add_bigint_l(bigint const& b, long_ const& l) bigint add_l_bigint(long_ const& l,bigint const& b) and are defined by .def("__add__", add_bigint_l) .def("__add__", add_l_bigint) Can someone tell me howto realize it? Benjamin Schmeling -- GMX im TV ... Die Gedanken sind frei ... Schon gesehen? Jetzt Spot online ansehen: http://www.gmx.net/de/go/tv-spot

2 1

re: strange filesystem dll behaviour
by Paul 19 Jan '05

19 Jan '05

Bugger, I forgot to mention: I'm using boost 1.32 and GCC 3.4.2 on a Windows 2000 computer. Thanks Paul Find local movie times and trailers on Yahoo! Movies. http://au.movies.yahoo.com

1 0

RE: [Boost-users] boost::bind and methods of managed classes
by Bjorn.Karlsson＠readsoft.com 19 Jan '05

19 Jan '05

Hello Vasco, > From: Vasco Lohrenscheit [mailto:valoh@krautgames.de] > > is there a easy way/workaround to use methods of managed classes with > boost::bind? It's not exactly easy, but I think it can be done. First of all, you won't be able to take the address of TestManagedClass::test because it's a managed class. Second, the calling conventions of invoking the signal (default: __stdcall) and the function test (__clrcall) are different. Here's a workaround that should work: * Create a nested __nogc class in the __gc class (technically, it doesn't need to be nested, just __nogc). * In the nested class, store a pointer (using gcroot<>, as it's a managed class) to the outer class. * In the constructor of the outer class, create an instance of the nested class. * Define a callback member in both classes. In the inner class, simply forward to the corresponding function in the outer class. * Create the bind expression using the address of the nested class' function. * Invoke the signal and keep your fingers crossed. Here's an example that demonstrates the technique: #include "boost/bind.hpp" #include "boost/signals.hpp" #include "vcclr.h" // For gcroot<> boost::signal<void (int)> sig; __gc class managed_class { public: managed_class() : inner_(new unmanaged_class(this)) { // Connect to the signal sig.connect(boost::bind(&managed_class::unmanaged_class::callback,inner_,_1) ); } void callback(int i) { // This is the managed "callback" } private: // Here's the nested class __nogc class unmanaged_class { public: unmanaged_class(managed_class* outer) : outer_(outer) {} void callback(int i) { outer_->callback(i); } private: gcroot<managed_class*> outer_; }; unmanaged_class* inner_; }; void main() { managed_class* p=new managed_class(); sig(42); // Invokes the signal, // which invokes the binder, // which invokes unmanaged_class::callback(), // which invokes managed_class::callback(). } Bjorn Karlsson

1 0

[Boost-users] Compile on OSX hangs, huge memory usage, xml_grammar.o
by Stian Soiland 19 Jan '05

19 Jan '05

Tried to build boost_1_32_0 on OSX today, but the build process brought the machine to trashing, as cc1plus ended up using all available memory. The problem seems related to libs/serialization - removing it made the rest of boost build almost OK. Here are the details: Mac OS X 10.3.7, with gcc version 3.3 20030304 (Apple Computer, Inc. build 1666) Lots of programs from fink installed to /sw Lots of programs from opendarwin installed to /opt 1. port install boost-jam (3.1.10) 2. Downloaded boost_1_32_0.tar.bz2 from boost.org and unpacked 3. bjam -sTOOLS=darwin stage 4. Start top in another terminal 5. Wait some hours until your hard drive starts to make crazy noises 6. Last bjam output will be: darwin-C++-action bin/boost/libs/serialization/build/libboost_serialization.a/darwin/ release/xml_grammar.o Top will show something like: Processes: 88 total, 2 running, 1 stuck, 85 sleeping... 223 threads 17:46:32 Load Avg: 1.72, 1.89, 1.45 CPU usage: 9.5% user, 30.2% sys, 60.3% idle SharedLibs: num = 131, resident = 7.25M code, 404K data, 136K LinkEdit MemRegions: num = 28961, resident = 896M + 2.28M private, 16.7M shared PhysMem: 111M wired, 601M active, 300M inactive, 1013M used, 10.6M free VM: 7.65G + 88.6M 1116924(79) pageins, 1865004(459) pageouts PID COMMAND %CPU TIME #TH #PRTS #MREGS RPRVT RSHRD RSIZE VSIZE 28631 cc1plus 11.5% 1:51.55 1 12 16184 842M- 804K 543M- 1.09G+ 28630 g++-3.3 0.0% 0:00.03 1 11 18 0K 264K 76K 21.8M 28628 bjam 0.0% 0:07.44 1 12 52 4K 264K 92K 31.2M rm -rf libs/serialization to avoid problem. But I guess some people actually need those xml-grammars.. -- Stian Søiland You can't say civilization don't Trondheim, Norway advance, however, for in every war http://www.soiland.no/ they kill you in a new way. [Rogers] =/\=

2 1

[stringalgo] Tokenizer with string delimiters?
by Stefan Slapeta 19 Jan '05

19 Jan '05

Hi, please could anybody tell if it's possible to use split() to tokenize a string where the token delimiters are strings of - n characters, e.g. "\r\n" - m..n characters, e.g. "\n" _or_ "\r\n" If not, are there alternatives? Thanks, Stefan

2 1