- Boost-users - lists.boost.org

A question about boost::graph iterators
by Jeffrey Holle 03 Nov '05

03 Nov '05

I'm using gcc 4.0.2 with boost 1.33.0. I'm getting a compiler warning about taking the address of a temporary with the following code: Edge* TransshipmentPivotingRule::enteringEdge(const FeasibleSpanningTree* pFST) { graph_traits<AntiTree>::edge_iterator ei,ei_out; tie(ei,ei_out)=edges(*m_pAntiTree); return (find_if(ei,ei_out, ProfitableEdge(*m_pAntiTree,m_c,m_y,*pFST))!=ei_out) ? &*ei : NULL; } Note the "&*ei". I know that the iterator is tempory, but what the iterator is pointing to isn't. Is this a legit warning that I need to be concerned about? In case it matters, AntiTree is a filtered_graph.

2 2

Thread Pools
by Brian Cox 03 Nov '05

03 Nov '05

Hello, I read a few years ago that Bill Kempf was going to develop a thread pool for Boost.Threads. Does anyone know the status of that? Does anyone have any good examples of a thread pool or worker queue developed using Boost.Threads? Thanks for any info. -Brian-

1 0

[Boost][serialization] link problems with MSVC.NET/no bjam
by Topi Mäenpää 03 Nov '05

03 Nov '05

Hi, I have struggled to get the boost libraries link into my application with no success so far. I was able to compile most of the demo programs with bjam, but it seems that only static linking is used there. I'm using the dynamic versions. The libraries compiled fine with MSVC.NET 7.1. However, whenever I try to link them to an application, every symbol seems to be undefined. I turned on the diagnostics to see that the automatic linking is happening (with the nonstandard preprocessor stuff). I also added the libraries to my project manually without success. The funny (?) thing is that the lib file has a definition for the "missing" functions, except for one minor difference. This is the mangled symbol the linker complains about: __imp_?save@?$xml_oarchive_impl@Vxml_oarchive@archive@boost@@@archive@boost@@QAEXABV?$basic_string@GU?$char_traits@G@std@@V?$allocator@G@2@@std@@@Z (Omitting namespaces this one decrypts to xml_oarchive_impl<xml_oarchive>::save(basic_string& const)) This is what seems to be in the library: __imp_?save@?$xml_oarchive_impl@Vxml_oarchive@archive@boost@@@archive@boost@@QAEXABV?$basic_string@DU?$char_traits@D@std@@V?$allocator@D@2@@std@@@Z The only difference is that the library symbol contains a "D" and the linker is looking for "G" in a couple of places. I have found no way of going any further. Since the demos are built in debug mode and with static libraries, I found no useful information by inspecting the commands bjam uses in building them. Everything is just like I have in my own project. I have defined BOOST_SERIALIZATION_DYN_LINK, if you happen to think about that. I even managed to link my app with mingw, but got strange run-time errors. Everything works fine with gcc on Linux. Any advice? -Topi-

3 4

[Boost][serialization] Serialization and DLLs
by Topi Mäenpää 03 Nov '05

03 Nov '05

Hi, Some time ago, we had a discussion about serialization of dynamically loaded classes. I got my code work on Linux, but still cannot do the same on Windows. I'd like someone to elaborate the stuff below a bit: Robert Ramey wrote: > Another post has me convinced that the problem is that there are multiple > copies of a base class extended_type_info record which I believe > shouldn't occur. I can confirm that this indeed happens. > If you trap your debugger the serialization.dll (shared_lib) at line 58 of > extened_type_info.cpp you should trap every time an extended_type_info > record is "self_registered". Examination of the stack should reveal > where this is being called from. I expect that the base class is being > self registered from both the DLL as well as the main executable. Whenever I register a class from a DLL, its base classes get registered as well, although they already have been registered. If I load multiple DLLs that all contain derived classes, the base classes are registered many times, but only once for each DLL. > My current thinking is that this behavior should be prohibited and > detected with an exception. Other users have addressed this > by tweaking the code in extended_type_info.cpp to permit > multiple "self registration" of the same extended_type_info type. > This eliminated their symptoms and everything seems to work. > But from my understanding of the library, I believe this just > hides a deeper problem. Can someone point out the details? How should this be done? Any other suggestions? Best regards, Topi

2 1

Newbie help needed
by Ricardo Kirkner 03 Nov '05

03 Nov '05

Hi: I am trying to write python bindings for an already existant c++ library. I started using pyste, because I felt this was the easiest way to do it. However I keep getting errors (mostly I think because of string and vector arguments conversion). I just wanted to ask if I am on the right track here. Maybe someone can give me some pointers on how to make this task easier? (I was tempted to switch to pyrex, swig, etc, but everything seems to have its own problems.) Should I use pyste, or should I rather write boost.python code directly? Thank you, Ricardo

2 1

enable_if, type_traits and Visual C++ 7.0
by Ovanes Markarian 03 Nov '05

03 Nov '05

Hello all, I am trying to get enable if to work with Visual C++ 7.0 and boost 1.33. for the following member function implemenation I get an error: template<> typename boost::enable_if<boost::is_pointer<_ElemFct::result_type>, _ElemFct::result_type>::type operator->() { return get_at(ptr_coll, pos); } template<> typename boost::enable_if<boost::is_reference<_ElemFct::result_type>, boost::add_pointer<_ElemFct::result_type>::type>::type operator->() { return &(get_at(ptr_coll, pos)); } My compiler is in German. I have tried to translate the error messages... error C2504: 'boost::enable_if_does_not_work_on_this_compiler<T>' : Base class undefined with [ T=boost::function2<std::string &,MemberTest *,int,int>::result_type ] see reference to the instance of compiled class template 'boost::enable_if<Cond,T>' with [ Cond=boost::is_pointer<boost::function2<std::string &,MemberTest *,int,int>::result_type>, T=boost::function2<std::string &,MemberTest *,int,int>::result_type ] see reference to the instance of compiled class template 'generic_iterator<_Coll,_Ty,_ElemFct,_LenFct,_Diff,_Ptr,_Ref,_Cat>' with [ _Coll=MemberTest, _Ty=std::string, _ElemFct=element_retrieval, _LenFct=length_retrieval, _Diff=int, _Ptr=std::string *, _Ref=std::string &, _Cat=std::bidirectional_iterator_tag ] error C2039: 'type' : Is not an element of 'boost::enable_if<Cond,T>' with [ Cond=boost::is_pointer<boost::function2<std::string &,MemberTest *,int,int>::result_type>, T=boost::function2<std::string &,MemberTest *,int,int>::result_type ] With Kind Regards, Ovanes Markarian

2 1

[date-time] posix timezone string
by Stering Wolfram 03 Nov '05

03 Nov '05

Hi, I am implementing a year_based_date-derived date generator to calculate the easter sunday for a given year. Now, this is the easy part, and it is done already. But I wonder how to generate a meaningful posix timezone string that is expected to result from calling the generator's to_string() member. The documentation at the (not well formatted, b.t.w.) "Description" section of class posix_time_zone didn't give any hints for Easter Sunday either. Any ideas? regards, -wolfi -- DI Wolfram Stering (Entwicklung) HALE electronic GmbH., Salzburg, Austria Tel: +43 (662) 439011 550 Fax: +43 (662) 439011 9 Email: wolfram.stering(a)hale.at

2 1

How to apply -Winline with Bjam v2
by Jeffrey Holle 03 Nov '05

03 Nov '05

I am using gcc 4.0.2 and Boost.Jam Version 3.1.11. I've been told that the way to determine if an inline method is actually inline by the compiler is to compile with the "-Winline" option. When this is done, a warning is emmitted when a sugguested inline method wasn't inlined. I'm using bjam v.2. I have not found the equivalent to "features.jam" in the v2 documentation, so I tried "<cxxflags>-Wonline", which I got from /tools/build/v1/features.jam. This doesn't work. What is the way to do it in Bjam v2?

1 0

design question for boost::program_options and dynamic options
by boost＠nogga.de 02 Nov '05

02 Nov '05

Hello, I'm new to the boost::program_options library and tried to use it for a command pattern based software. The program is launched with the following commandline: > program command [Options] <Files> The program can use different output formatters with --formatter={XML|binary|text}. Every command and formatter specify their own set of options. I first tried to collect all options form the commands and formatters and parse the command line only once. This lead to the problem, that due to the different positional options information the commandline parser always threw an exception due to wrong command line formatting. Since I couldn't find a way to add mutliple optional option groups to the parser, I moved the parsing code into my command base class and query the derived commands for the different required information, like: class Command { public: vritual options_description GetOptionsDescription (); virtual options_description GetHiddenOptionsDescription (); virtual positional_program_options GetPositionalProgramOptions (); virtual Run (variables_map const & vm) = 0; int Execute (std::vector<std:.string> args) { options_description desc; descr.add (GetOptionsDescription ()).add (GetHiddenOptionsDescription ()); variables_map vm; store(command_line_parser(args) .options(desc). .positional(GetPositionalProgramOptions ()) .run(), vm); notify(vm); Run (vm); } } The execute function is called form the program main after the "command" has been stripped from the commandline and the concrete command was created by the factory. So far so good. This works for the different commands, but not for the formatters. How can I parse the command line dynamically? E.g. if I encounter the "--formatter=xml" argument, a new XMLFormatter is created and the options description from this formatter are added to list of valid options? I know, that there is the notify mechanism, but this only works after the commandline has been successfully parsed. And also the custom parsers do not solve my problem. Any idea how to solve this issue? Best regards Dirk

2 2

Tokenizer: how to eliminate unwanted null tokens?
by CN 02 Nov '05

02 Nov '05

Hi! The following code uses '\0' and ';' as seperators and produces the result: <X><Y><><Z><> string str=string("X;Y\0\0Z\0",7); typedef tokenizer<boost::escaped_list_separator<char> > Tok; escaped_list_separator<char> sep(string(),string("\0;",2),string()); Tok tokens(str, sep); for(Tok::iterator i = tokens.begin();i != tokens.end();++i) cout << "<" << *i << ">"; How to revise it in order to get this result? <X><Y><Z> Thank you in advance! CN -- http://www.fastmail.fm - Does exactly what it says on the tin

2 1