June 2008 - Boost - lists.boost.org

Re: [boost] [circular_buffer] why is there no reserve(size_type)?
by Jan Gaspar 11 Jun '08

11 Jun '08

because: 1) it's behaviour is slightly different from the standard - you can increase but also decrease the capaciry of the container; decreasing of the capacity is not supported by reserve() 2) the word capacity is used all over the documentation of the circular buffer. I think set_capacity expresses better what the method does. Jan ----- Original Message ---- From: Steven Watanabe <watanabesj(a)gmail.com> To: boost(a)lists.boost.org Sent: Wednesday, 11 June, 2008 3:49:21 PM Subject: Re: [boost] [circular_buffer] why is there no reserve(size_type)? AMDG Jan Gaspar wrote: > it is called set_capacity(capacity_type) > Why is this different from the standard library name? In Christ, Steven Watanabe _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost __________________________________________________________ Sent from Yahoo! Mail. A Smarter Email http://uk.docs.yahoo.com/nowyoucan.html

2 1

Building boost 1.35 with boost.serialization 1.36
by Chris Hamilton 11 Jun '08

11 Jun '08

I require a few of the new features introduced in Boost.Serialization 1.36, namely the support of serialization across multiple shared libraries and multi-threaded support. The Serialization library itself doesn't build cleanly, but adding a missing #include in one of the headers solves that for me. However, I now get errors when building the MPI library: gcc.compile.c++ bin.v2/libs/mpi/build/gcc-3.4.6/release/threading-multi/packed_iarchive.o In file included from libs/mpi/src/packed_iarchive.cpp:11: ./boost/archive/impl/archive_pointer_oserializer.ipp:27: error: redefinition of `class boost::archive::detail::<unnamed>::serializer_map<Archive>' ./boost/archive/impl/archive_pointer_iserializer.ipp:27: error: previous definition of `class boost::archive::detail::<unnamed>::serializer_map<Archive>' "g++" -ftemplate-depth-128 -O3 -finline-functions -Wno-inline -Wall -fPIC -pthread -DBOOST_ALL_NO_LIB=1 -DBOOST_MPI_DYN_LINK=1 -DBOOST_MPI_SOURCE=1 -DNDEBUG -I"." -I"/opt/lam/gnu/include" -c -o "bin.v2/libs/mpi/build/gcc-3.4.6/release/threading-multi/packed_iarchive.o" "libs/mpi/src/packed_iarchive.cpp" ...failed gcc.compile.c++ bin.v2/libs/mpi/build/gcc-3.4.6/release/threading-multi/packed_iarchive.o... ...removing bin.v2/libs/mpi/build/gcc-3.4.6/release/threading-multi/packed_iarchive.o gcc.compile.c++ bin.v2/libs/mpi/build/gcc-3.4.6/release/threading-multi/packed_oarchive.o gcc.compile.c++ bin.v2/libs/mpi/build/gcc-3.4.6/release/threading-multi/packed_skeleton_iarchive.o In file included from libs/mpi/src/packed_skeleton_iarchive.cpp:11: ./boost/archive/impl/archive_pointer_oserializer.ipp:27: error: redefinition of `class boost::archive::detail::<unnamed>::serializer_map<Archive>' ./boost/archive/impl/archive_pointer_iserializer.ipp:27: error: previous definition of `class boost::archive::detail::<unnamed>::serializer_map<Archive>' "g++" -ftemplate-depth-128 -O3 -finline-functions -Wno-inline -Wall -fPIC -pthread -DBOOST_ALL_NO_LIB=1 -DBOOST_MPI_DYN_LINK=1 -DBOOST_MPI_SOURCE=1 -DNDEBUG -I"." -I"/opt/lam/gnu/include" -c -o "bin.v2/libs/mpi/build/gcc-3.4.6/release/threading-multi/packed_skeleton_iarchive.o" "libs/mpi/src/packed_skeleton_iarchive.cpp" ...failed gcc.compile.c++ bin.v2/libs/mpi/build/gcc-3.4.6/release/threading-multi/packed_skeleton_iarchive.o... ...removing bin.v2/libs/mpi/build/gcc-3.4.6/release/threading-multi/packed_skeleton_iarchive.o gcc.compile.c++ bin.v2/libs/mpi/build/gcc-3.4.6/release/threading-multi/packed_skeleton_oarchive.o ...skipped <pbin.v2/libs/mpi/build/gcc-3.4.6/release/threading-multi>libboost_mpi-gcc34-mt-1_35.so.1.35.0 for lack of <pbin.v2/libs/mpi/build/gcc-3.4.6/release/threading-multi>packed_iarchive.o... Any idea on how to solve these? It seems that Boost.MPI is rather 'chummy' with the internals of Boost.Serialization, and that things have changed rather significantly there in 1.36. Regards, Chris Hamilton

2 1

[serialization] ambiguous symbol if include "binary_iarchive.hpp"
by Wang Yun 11 Jun '08

11 Jun '08

Hi, I meet a compile error of boost::function only when I add this line #include <boost/archive/binary_iarchive.hpp> I'm using VC2005, boost 1.35.0 error message: error C2872: 'function' : ambiguous symbol 1> could be 'e:\libraries\boost_1_35_0\boost\function\function_base.hpp(106) : boost::function' 1> or 'boost::detail::function' 1> or 'e:\libraries\boost_1_35_0\boost\function\function_base.hpp(106) : boost::function' // ======= the code =============================== #include "stdafx.h" #include <boost/function.hpp> using boost::function; #include <boost/archive/binary_iarchive.hpp> //////////////// when add this line, error will happen int _tmain(int argc, _TCHAR* argv[]) { function<void()> fn; return 0; } //================================================== Thanks, Wang Yun

3 2

[python] can I stop iteration without exception being thrown?
by Gennadiy Rozental 11 Jun '08

11 Jun '08

Hi, I an using vector_indexing_suite for both standard containers export and for our in-house ones. I've noticed that boost::python::error_already_set is bwing thrown every time iteration in python over container instances is about to finish. The exception is ignored somewhere inside boost::python but it's still quite inconvenient for debugging: I want to catch actual error, but can't skip 100s of fake ones. Is there any reason why it was designed this way? Is there any way to avoid it? Gennadiy

6 14

Re: [boost] [circular_buffer] why is there no reserve(size_type)?
by Jan Gaspar 11 Jun '08

11 Jun '08

it is called set_capacity(capacity_type) ----- Original Message ---- From: Thorsten Ottosen <thorsten.ottosen(a)dezide.com> To: "boost(a)lists.boost.org" <boost(a)lists.boost.org> Sent: Wednesday, 11 June, 2008 2:05:14 PM Subject: [boost] [circular_buffer] why is there no reserve(size_type)? Hi Jan, The headline pretty much says it all. -Thorsten _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost __________________________________________________________ Sent from Yahoo! Mail. A Smarter Email http://uk.docs.yahoo.com/nowyoucan.html

2 1

[circular_buffer] why is there no reserve(size_type)?
by Thorsten Ottosen 11 Jun '08

11 Jun '08

Hi Jan, The headline pretty much says it all. -Thorsten

1 0

Bug in iterator_range when NDEBUG not defined but boost asserts are enabled
by John Reid 11 Jun '08

11 Jun '08

iterator_range.hpp defines is_singular() if NDEBUG is not defined but it calls is_singular through the BOOST_ASSERT macro. Unfortunately it is possible to enable BOOST_ASSERT and have NDEBUG defined which leads to a compilation error. To see this compile the iterator_range.cpp example included with the library with the following defines: gcc -DBOOST_ENABLE_ASSERT_HANDLER -I/usr/local/src/boost-trunk/ -DNDEBUG -c iterator_range.cpp John.

4 3

Using Bind & Lambda
by Robert Jones 11 Jun '08

11 Jun '08

This is my first time posting to this list, so I hope you'll forgive me it my etiquette is awry. I'm new to the Boost libraries, and just learning my way around, guided mainly by Bjorn Karlson's book - any other book recommendations would be most welcome btw, and there a few things I just don't quite get! First - is "boost/lambda/bind.hpp" equivalent to "boost/bind.hpp"? Should it ever be necessary to include both? Second - Is it true that a lambda function is syntactically and semantically the same as a bind finction, eg Given:- int foo( int i ) { return i + 3; } Are these two expressions the same? bind( &foo, _1 ); _1 + 3; Finally (for now!) - why is this wrong? struct A { string identity( string ) const; } a; vector<string> v; for_each( v.begin(), v.end(), cout << bind( &A::identity, a, _1 ) << '\n' ); which is intended to output the return value of A::identity() for each element of v, separated by newlines. Thanks in advance for any guidance! Rob. -- ACCU - Professionalism in programming - http://www.accu.org

5 4

[python] working with raw const pointers
by Gennadiy Rozental 11 Jun '08

11 Jun '08

Hi, In our system there are a lot of pointers to const objects flying around. I've got several problems in this regard. They can be exhibited by the following code snippet: a.cpp : class A{}; void boo1( A const* ) {} void boo2( A* ) {} A const* make_const_a() { return (A const*)0; } A* make_non_const_a() { return (A*)0; } std::vector<A const*> make_vec() { return std::vector<A const*>( 1 ); } .... class_<A,boost::noncopyable>( "A", bp::no_init ); class_<std::vector<A const*> >( "AVec", bp::no_init ) .def( vector_indexing_suite<std::vector<A const*> >() ) ; def( "boo", &boo1 ); def( "boo", &boo2 ); def( "make_const_a", &make_const_a, return_value_policy<reference_existing_object>() ); def( "make_non_const_a", &make_non_const_a, return_value_policy<reference_existing_object>() ); def( "make_vec", &make_v ec ); Problem I. Access to mutable func Following python statement mymodule.boo( mymodule.make_non_const_a() ) seems to call incorrect function boo Problem II. Vector fails completely Following python statements vec = pytoolkit.make_vec(); for e in vec: pass produce TypeError: No to_python (by-value) converter found for C++ type: class A const * Any advise? Gennadiy

1 0

Threaded version
by albert meyburgh 11 Jun '08

11 Jun '08

Is there a threaded version of some of the boost components? For example, calling find on a container should be easily threaded. I realize the user could also easily thread this, but it seems like a trivial thing to just add support for in boost itself.

3 2