June 2005 - Boost-users - lists.preview.boost.org

static link of boost.thread requires link to static runtime ?
by Seb Martel 14 Jun '05

14 Jun '05

Hi- I am trying to static link to boost.thread 1.32 (winXP) and I see it requires the runtimes to be linked static as well (#define _DLL mustn't be defined). I want to use the dynamic runtimes, but static link boost.thread. Is this possible ? What prevents it if not ? Thanks, -seb

1 1

Serialization - compile warnings with with text_iarchive.hpp and demo.cpp
by Ken Roser 14 Jun '05

14 Jun '05

When compiling demo.cpp I get the warnings: /home/test/boost/include/boost-1_32/boost/archive/detail/common_iarchive.hpp: In member function `void boost::archive::detail::common_iarchive<Archive>::init() [with Archive = boost::archive::text_iarchive]': /home/test/boost/include/boost-1_32/boost/archive/text_iarchive.hpp:69: instantiated from `boost::archive::text_iarchive_impl<Archive>::text_iarchive_impl(std::istream&, unsigned int) [with Archive = boost::archive::text_iarchive]' /home/test/boost/include/boost-1_32/boost/archive/text_iarchive.hpp:82: instantiated from here /home/test/boost/include/boost-1_32/boost/archive/detail/common_iarchive.hpp:75: warning: choosing `boost::archive::version_type::operator unsigned int&()' over ` boost::archive::version_type::operator unsigned int() const' /home/test/boost/include/boost-1_32/boost/archive/detail/common_iarchive.hpp:75: warning: for conversion from `boost::archive::version_type' to `unsigned int' /home/test/boost/include/boost-1_32/boost/archive/detail/common_iarchive.hpp:75: warning: because conversion sequence for the argument is better I'm using this compiler: g++ (GCC) 3.2.3 20030502 (Red Hat Linux 3.2.3-52) I'd like a way to silence these warnings assuming this isn't a problem on my system. I apologize if this was discussed previously, but I couldn't find any means to search the mailing list archive.

2 1

bug on /libs/graph/doc/quick_tour.html
by martin f krafft 14 Jun '05

14 Jun '05

The first bit of sample code in the section "Adding Some Color to your Graph" on http://www.boost.org/libs/graph/doc/quick_tour.html does not compile. Graph G(num_nodes, edges, edges + sizeof(edges) / sizeof(E), weights); --> there seems to be no corresponding constructor. I am still fighting to find the proper solution and will followup when I find something. -- martin; (greetings from the heart of the sun.) \____ echo mailto: !#^."<*>"|tr "<*> mailto:" net@madduck invalid/expired pgp subkeys? use subkeys.pgp.net as keyserver! spamtraps: madduck.bogus(a)madduck.net "you don't sew with a fork, so I see no reason to eat with knitting needles." -- miss piggy, on eating chinese food

2 1

Boost.Thread : catching all exceptions
by Adrian Mawbey 13 Jun '05

13 Jun '05

Hi folks. Apologies if this has been discussed frequently before. I'm puzzled as to why the boost.thread thread_proxy function catches all exceptions and then quietly ends the thread. For example:- class MyThreadFunctor { public: void operator()() const { // Do some processing... throw very_bad_exception(); // ...never get to do some processing. } }; int main(int argc, char * argv[]) { MyThreadFunctor my_functor; boost::thread my_thread(my_functor); // Do lots of processing and then join to my_thread. // ... } In boost 1.32, the exception that is thrown in the created thread is caught in the library's thread_proxy function and then the thread exits quietly. Is there a compelling reason for this behaviour? >From my point of view, it seems more flexible to not catch exceptions in the thread_proxy function. If I wanted to, I could still get the current behaviour by putting a try/catch-all block in the operator() function. I would have expected the normal C++ uncaught exception actions (i.e. calling terminate()) to occur if an exception is thrown out of the user provided thread code. I noticed this when (oh horror!) an error in my code threw an exception out of operator(). I could have identified where the problem was from the core file or Dr Watson log that would have been created if the exception had not been caught by thread_proxy(). Am I missing something obvious here? Thanks, Adrian.

1 0

Tokenizer Question
by Tom Browder 13 Jun '05

13 Jun '05

I have used my own C++ tokenizer in the past, but I would like to use Boost's instead. The predominant use of tokenizing for me is to split on white space, but Boost's default is to use white space AND punctuation. Is there any possibility to have either the default changed, or another TokenizerFunction added such as ws_separator, or something similar? I know I can use boost::char_separator<char> sep(" \n\t"); (but do I need to add "\v" to the char set?) but I would rather have something like boost::ws_separator sep; and, better, make the ws_separator be the default TokenizerFunction for tokenizer. Thanks for listening. Tom Browder

2 2

Text_archives don't support integer types?
by Marcin Tustin 13 Jun '05

13 Jun '05

I'm using the CVS version of Boost from 2005-06-06, and it seems that text archives don't support anything except strings, characters, and the like, specifically not integer types like int. I've grepped the archive includes, and it seems that there are no save methods specified for the integer types (except char) by the text archives, and I get compiler errors like: c:\D2\API\NewClient\EnhancedMessageHandler.h(370) : error C2679: binary '<<' : no operator found which takes a right-hand operand of type 'int' (or there is no acceptable conversion) (with VC7.1) It works when the RHS of << is a variable, but not when it is a function. I take it that this is because with variables, a string is automatically constructed to wrap the int types. This is very confusing, and might well be worth mentioning in the docs. Marcin Tustin IT Innovation Centre 2 Venture Road Chilworth Science Park Southampton, SO16 7NP, UK tel: +44 23 8076 0834 fax: +44 23 8076 0833 mailto:mt@it-innovation.soton.ac.uk http://www.it-innovation.soton.ac.uk

2 1

qualified name error on vc7-stlport
by Johannes Strömberg 13 Jun '05

13 Jun '05

Hi, I am trying to build boost for vc7-stlport (stlport version 4.6.2) and I get most libraries to compile but when compiling the date_time library I get errors like the following: C:\wspace\dev\ssdk\ssdk\stlport\src\stlport\stl\_iterator_base.h(98) : error C2825: '_Iterator::iterator_category': cannot form a qualified name C:\wspace\dev\ssdk\ssdk\boost\src\boost\date_time\date_parsing.hpp(252) : see reference to class template instantiation '_STL::iterator_traits<_Iterator>' being compiled with [ _Iterator=_STL::basic_string<char,_STL::char_traits<char>,_STL::allocator<char>>::iterator ] C:\wspace\dev\ssdk\ssdk\boost\src\boost\date_time\gregorian\parsers.hpp(73) : see reference to function template instantiation 'boost::date_time::period<point_rep,duration_rep> boost::date_time::from_simple_string_type<boost::gregorian::date,char>(const _STL::basic_string<_CharT,_Traits,_Alloc> &)' being compiled with [ point_rep=boost::gregorian::date, duration_rep=boost::gregorian::date_duration, _CharT=char, _Traits=_STL::char_traits<char>, _Alloc=_STL::allocator<char> ] C:\wspace\dev\ssdk\ssdk\stlport\src\stlport\stl\_iterator_base.h(98) : error C2039: 'iterator_category' : is not a member of 'operator``global namespace''' C:\wspace\dev\ssdk\ssdk\stlport\src\stlport\stl\_iterator_base.h(98) : fatal error C1507: previous user errors and subsequent error recovery halt further compilation I know it has to do with a compiler deficiency but how can I fix it? /Johannes

1 0

multi_array operator= and copy constructor have problem
by Yan Zhang 13 Jun '05

13 Jun '05

I'm testing multi_array with a simple example: typedef boost::multi_array<double, 3> array; array A(boost::extents[3][4][2]); A[0][0][0] = 3.14; array B; B = A; When execute the last statement B=A, you encounter assertion: assert(std::equal(other.shape(),other.shape()+this->num_dimensions(), this->shape())); in file multi_array_ref.hpp (line 476). I used VC++ 7.1 with debug build. Please advice.

2 1

Re: [Boost-users] How to inherit from boost::multi_array
by Yan Zhang 13 Jun '05

13 Jun '05

I'm read the posting by Thomas and replay by Ronald on how to inherit from boost::multi_array as following: On Oct 2, 2004, at 10:26 AM, Thomas Mensch wrote: Dear List, I am using boost 1.31.0 on Linux (with gcc 3.3.2) I would like to create a 2D array class, which inherits from boost::multi_array<double, 2> class. However I have problems with the copy constructor of my class Array. Below is the code, which I am trying to run. #include <boost/multi_array.hpp> class Array : public boost::multi_array<double,2> { public: Array (void) : boost::multi_array<double, 2> () {} Array (const int nrow, const int ncol) : boost::multi_array<double, 2> (boost::extents[nrow][ncol]) {} Array (const Array& other) : boost::multi_array<double, 2> (other) {} ~Array (void) {} Array& operator= (const Array& rhs) { if (&rhs != this) { boost::multi_array<double, 2>::operator= (rhs); } return *this; } }; /* class Array */ Ronald's replay: Greetings Thomas, Though you did not supply the error message from your compiler, I suspect that the problem is this: multi_array has a templated constructor of the form: template <class T> multi_array(T const&); that will match your Array type better than the constructor with the signature: multi_array(multi_array<double,2> const&); In short, overload resolution is not selecting the constructor that you desire. You'll need to coerce your Array reference to a multi_array<double,2> reference before passing it to the constructor. Hope this helps. I had the same problem as Thomas'. I don't quit understand Ronald's comment. How to you overload template <class T> multi_array(T const&) instead of multi_array(multi_array<double,2> const&)? I thought Thomas' code is overloading template <class T> multi_array(T const&). Thanks in advance, Yan

2 1

[BGL] edge_index_t
by martin f krafft 13 Jun '05

13 Jun '05

I see the edge_index_t symbol used all over the place, but I cannot find its defintion. So now I am stuck: error: `egde_index_t' undeclared in namespace `boost' What header file do I have to include to be able to access my graph's edge index property? -- martin; (greetings from the heart of the sun.) \____ echo mailto: !#^."<*>"|tr "<*> mailto:" net@madduck invalid/expired pgp subkeys? use subkeys.pgp.net as keyserver! spamtraps: madduck.bogus(a)madduck.net echo '9,J8HD,fDGG8B@?:536FC5=8@I;C5?@H5B0D@5GBIELD54DL>@8L?:5GDEJ8LDG1' |\ sed ss,s50EBsg | tr 0-M 'p.wBt SgiIlxmLhan:o,erDsduv/cyP'

2 2