May 2008 - Boost-users - lists.boost.org

[Fwd: is_pod newbie question]
by Elli Barasch 23 May '08

23 May '08

With some wrangling, I figured out how to use the is_pod function, but I discovered it has little value to me unless I can get the status of a class/struct. Has there been any progress in this area? Thanks.

4 6

compile errors with ublas
by Sebastian Weber 23 May '08

23 May '08

Hi! I want to initialize an ublas vector with zeros and always get compile errors in the debug mode which I do not understand, since this is supposed to be the right way as of the ublas FAQ: std::size_t vecSize = 50; bnu::vector<std::size_t> vec( ( bnu::zero_vector<std::size_t>(vecSize) ) ); What is wrong here? My versions are boost 1.35.0, gcc 4.3.0 on ubuntu hardy. BTW, is there some quasi-official code to serialize ublas vectors with the serialization library? I found some code on this mailing list, but is there some code floating around in the svn which I possibly missed? Thanks a lot in advance. Sebastian

1 0

BOOST_MPL_ASSERT at struct scope
by Hansi 23 May '08

23 May '08

Hello, I want to use the BOOST_MPL_ASSERT Macro inside a struct (compiler is MSVC2005) #include <boost/mpl/assert.hpp> #include <boost/type_traits.hpp> template<typename Value> struct Getter { typedef std::wstring result_type; BOOST_MPL_ASSERT(( boost::is_fundamental< Value > )); std::wstring operator()(const Value* val)const { return boost::lexical_cast<std::wstring>(*val); } std::wstring operator()(typename boost::call_traits<Value>::const_reference val)const { return boost::lexical_cast<std::wstring>(val); } }; Here I get allways the following error: Error 1 error C2664: 'boost::mpl::assertion_failed' : cannot convert parameter 1 from 'boost::mpl::failed ************boost::is_fundamental<T>::* ***********' to 'boost::mpl::assert<false>::type' Utils.h 32 Error 2 error C2866: 'Microtec::Viewer::Getter<Value>::mpl_assertion_in_line_32' : a const static data member of a managed type must be initialized at the point of declaration Utils.h 32 Inside the function it works fine.. Has anyone an idea how I can solve that? Best regards Hansjörg

1 0

Boost 1.35 threads library question
by Chris Ross 23 May '08

23 May '08

Hi there. This is an introductory question as I've just started to play with boost libraries. If this would be better on the threads-devel list, let me know, but I thought starting here would be good. I was playing with the threads library in 1.35, and I noticed that I can't (with gcc 4.3 on Linux) put boost::thread objects into an STL container. I think I was trying with a std::vector, in case that matters. I was reading in some of the C++0x documents that it's expected that the standard library thread implementation is expected to be able to have this work. The boost documentation doesn't explicitly say whether it should work or not. Is being non-copyable but movable enough to make this work? Or, would an object have to be copyable to be put into an STL container? I thought that was not the case, but. Thanks... - Chris

4 4

Re: [Boost-users] Boost-users Digest, Vol 1640, Issue 4
by Daniel Lord 23 May '08

23 May '08

From: Vladimir Prus <vladimir(a)codesourcery.com> > Daniel Lord wrote: > > > Several problems with configurations not properly propagating through the > > long build chain it seems.That is if I am setting this up > correctly--there > > is always the possibility this is pilot error in the setup. > > > > 1) configure architecture=combined should be the default and provide a > > universal 32-bit architecture file: > > What do you mean "configure"? If you're using the "configure" script, then > stop, > and invoke bjam directly. > Thanks for your help. I studied Boost-build a bit and the darwin.jam file and came up with a command line for bjam And it mostly worked with one hitch: The command: "bjam release debug toolset=darwin threading=multi architecture=combined address-model=32" works to build both debug and release version of fat-32 (multi-arch i386 and ppc) dynamic libs. But I cannot seem to get static libs to build. the command: "bjam release debug runtime-link=static toolset=darwin threading=multi architecture=combined address-model=32" results in no build at all as shown in this copy of the shell output: $bjam release debug runtime-link=static toolset=darwin threading=multi architecture=combined address-model=32 warning: Graph library does not contain optional GraphML reader. note: to enable GraphML support, set EXPAT_INCLUDE and EXPAT_LIBPATH to the note: directories containing the Expat headers and libraries, respectively. warning: skipping optional Message Passing Interface (MPI) library. note: to enable MPI support, add "using mpi ;" to user-config.jam. note: to suppress this message, pass "--without-mpi" to bjam. note: otherwise, you can safely ignore this message. Building Boost.Regex with the optional Unicode/ICU support disabled. Please refer to the Boost.Regex documentation for more information (don't panic: this is a strictly optional feature). Skipping build of: ./headers <build>no in common properties Skipping build of: ./build_all <build>no in common properties Skipping build of: ./headers <build>no in common properties Skipping build of: ./build_all <build>no in common properties ...found 1 target... So I am obviously close but still misunderstanding something. Any hints?

1 0

Re: [Boost-users] Boost-users Digest, Vol 1640, Issue 4
by Richard L. Aurbach 22 May '08

22 May '08

> Anthony Williams <anthony.ajw(a)gmail.com> writes: > > Chris Ross <cross+boost(a)distal.com> writes: > >> Hi there. This is an introductory question as I've just started to >> play with boost libraries. If this would be better on the threads- >> devel >> list, let me know, but I thought starting here would be good. > > Here is fine. > >> I was playing with the threads library in 1.35, and I noticed that I >> can't (with gcc 4.3 on Linux) put boost::thread objects into an STL >> container. I think I was trying with a std::vector, in case that >> matters. > > No, you can't do that. You need a container that can handle > movable-but-not-copyable objects. Not only that, but it needs to > handle the thread move emulation. > >> I was reading in some of the C++0x documents that it's expected that >> the standard library thread implementation is expected to be able to >> have this work. The boost documentation doesn't explicitly say >> whether >> it should work or not. Is being non-copyable but movable enough to >> make >> this work? Or, would an object have to be copyable to be put into an >> STL container? I thought that was not the case, but. > > Non-copyable but movable is indeed enough for a class to be used in a > C++0x STL container. I don't know whether the containers with gcc 4.3 > have that support in -std=c++0x mode or not. The current trunk has > some rvalue reference support in boost::thread, so it may well work, > but I haven't tried it. Wouldn't another option be to create a vector of shared_ptr's of boost::thread objects?? The shared_ptr's work in vectors just fine, plus they provide a convenient way to pass around a non-copyable object like a thread, as well as other thread-pool functions. Cheers, Rick Aurbach, Ph.D. Aurbach & Associates, Inc.

2 1

tuple type question
by Hansi 22 May '08

22 May '08

Hello, at the moment I want to make a getter function for a tuple type. The tuple is internally hidden in a class. For that I want to make a function which returns the values inside the tuple. The best solution would be if I can make a enum which defines the position inside the tuple and a template function which returns the value for this tuple. I have tested the following, but it doesn't work: typedef struct Members { enum Member { name = 0, value = 1, }; }Members; template<enum Member member> element<0, Properties::Property>::type name()(const boost::tuples::tuple<std::wstring, boost::any>& prop) { return boost::tuples::get<member>(prop); } Can anyone give my an hint? I'm not very familiar with the metaprogramming things... Best regards Hansjörg

7 23

[serialization] Handling of references into a std::map
by Bernhard Maeder 22 May '08

22 May '08

Hello all I'm trying to serialize a std::map along with some pointers that are referencing onto some of the Value elements of that map. What I see now, is that those references are not loaded correctly. Here's what I do: #include <sstream> #include <boost/archive/text_oarchive.hpp> #include <boost/archive/text_iarchive.hpp> #include <boost/serialization/map.hpp> struct dummy { template<typename Archive> void serialize(Archive & ar, const unsigned int version) { } }; struct map_ref_test { std::map<std::size_t, dummy> m; dummy * ref; template<typename Archive> void serialize(Archive & ar, const unsigned int version) { ar & m; ar & ref; } }; int main() { std::ostringstream os; map_ref_test m; m.m[0] = dummy(); m.m[1] = dummy(); m.m[2] = dummy(); m.ref = &(m.m[1]); // serialize boost::archive::text_oarchive out(os); out << const_cast<map_ref_test const &>(m); // De-serialize map_ref_test m2; std::string ser = os.str(); std::istringstream is(ser); boost::archive::text_iarchive in(is); in >> m2; std::cout << "0: " << &m.m[0] << std::endl; std::cout << "1: " << &m.m[1] << std::endl; std::cout << "2: " << &m.m[2] << std::endl; std::cout << "ref: " << m.ref << std::endl; std::cout << "0: " << &m2.m[0] << std::endl; std::cout << "1: " << &m2.m[1] << std::endl; std::cout << "2: " << &m2.m[2] << std::endl; std::cout << "ref: " << m2.ref << std::endl; } This creates the following output: 0: 0x805bfc4 1: 0x805bfe4 2: 0x805c004 ref: 0x805bfe4 0: 0x805ca04 1: 0x805ca24 2: 0x805ca44 ref: 0xbfacad5c I'd expect the last line to read: ref: 0x805ca24 So, what am I doing wrong? Should this even work? BTW: I'm using Boost 1.33.1 Thanks a lot! Bernhard

4 8

[MPI] Compiliation problem while building boost
by Adam Stephens 22 May '08

22 May '08

Hi, all, I'm trying to build Boost on linux using bjam. Everything seems to go smoothly except the compiliation of one file in Boost.MPI. The error is lengthy, but the first part reads: intel-linux.compile.c++ bin.v2/libs/mpi/build/intel-linux/release/threading-multi/datatypes.o ./boost/mpi/python/serialize.hpp(321): error: no operator "<<" matches these operands operand types are: boost::mpi::packed_oarchive << boost::mpl::if_<boost::is_fundamental<long>, long, long &>::type ar << extract<Type>(obj)(); ^ detected during: instantiation of "void boost::python::detail::direct_serialization_table<IArchiver, OArchiver>::default_saver<T>::operator()(OArchiver &, const boost::python::api::object &, unsigned int) [with IArchiver=boost::mpi::packed_iarchive, OArchiver=boost::mpi::packed_oarchive, T=long]" at line 158 of "./boost/function/function_template.hpp" My compiler is Intel/9.1 and the implementation of MPI is openmpi 1.2.4. I've posted my user-config.jam, the first several lines of output from bjam --debug-configuration, and the section of the bjam output with all the errors at http://rafb.net/p/qLNQ6j95.html Thanks

1 0

is_pod newbie question
by Elli Barasch 22 May '08

22 May '08

Can someone point me to a concrete example of using is_pod? I'm confused by the docs. I'm trying to avoid performance-heavy serialization on an object. I can short-cut the serialization if I can determine if my object is POD. Thanks, Elli

1 0