- Boost-users - lists.boost.org

Boost.MPI compilation problems with MPICH2
by Aydın Buluç 22 Oct '07

22 Oct '07

Hi all, I was a happy user of Boost.MPI that was working together with the underlying OpenMPI. However, MPI_THREAD_MULTIPLE support of OpenMPI is quite clumsy and it hangs frequently (OpenMPI developers call this as being "lightly tested"). So, I installed MPICH2 that claims to supports thread-safe MPI. At the same time, I didn't want to remove my working OpenMPI implementation and my old Boost libraries built on top of that. Therefore, I rebuilt Boost (--with-mpi) using the new mpicxx of MPICH (so that it finds the correct MPICH2 directories, etc) on a different location . Everything went ok during compilation and installation. However, I have compilation errors when trying to run any mpi program. For example, with the test suite provided by Boost.MPI itself, I get errors like (this is for all_gather.cpp specifically but I get similar ones with every other test): ------------------------------------- /tmp/ccbyPLBK.o: In function `void boost::mpi::detail::all_gather_impl<int>(boost::mpi::communicator const&, int const*, int, int*, mpl_::bool_<true>)':all_gather_test.cpp:(.gnu.linkonce.t._ZN5boost3mpi6detail15all_gather_implIiEEvRKNS0_12communicatorEPKT_iPS6_N4mpl_5bool_ILb1EEE[void boost::mpi::detail::all_gather_impl<int>(boost::mpi::communicator const&, int const*, int, int*, mpl_::bool_<true>)]+0x1c): undefined reference to `boost::mpi::communicator::operator int() const' /tmp/ccbyPLBK.o: In function `void boost::mpi::detail::broadcast_impl<std::basic_string<char, std::char_traits<char>, std::allocator<char> > >(boost::mpi::communicator const&, std::basic_string<char, std::char_traits<char>, std::allocator<char> >*, int, int, mpl_::bool_<false>)':all_gather_test.cpp:(.gnu.linkonce.t._ZN5boost3mpi6detail14broadcast_implISsEEvRKNS0_12communicatorEPT_iiN4mpl_5bool_ILb0EEE[void boost::mpi::detail::broadcast_impl<std::basic_string<char, std::char_traits<char>, std::allocator<char> > >(boost::mpi::communicator const&, std::basic_string<char, std::char_traits<char>, std::allocator<char> >*, int, int, mpl_::bool_<false>)]+0x27): undefined reference to `boost::mpi::communicator::operator int() const' ------------------------------------- I "quadruple-checked" to see that the versions of mpicxx/libraries/include directories all belong to the new installation of MPICH2 and not remainders of old OpenMPI installation. Thus, I don't think that is a binary incompatibility issue. MPICH2 implementation works OK alone (without boost) and the old Boost+OpenMPI installation libraries also still work. But the new Boost+MPICH2 libraries... ihhh :( Any suggestions and pointers will be appreciated. Thanks, -- Aydin

2 1

Re: [Boost-users] [mpl] at_c type compiler error
by Robbie Morrison 22 Oct '07

22 Oct '07

------------------------------------------------------------ Subject: Boost-users Digest, Vol 1423, Issue 3 Date: Thu, 18 Oct 2007 12:00:06 -0400 Message-ID: <mailman.11.1192723206.19439.boost-users(a)lists.boost.org> ------------------------------------------------------------ Message: 3 Date: Thu, 18 Oct 2007 17:03:25 +0200 From: Vinzenz 'evilissimo' Feenstra <evilissimo(a)web.de> Subject: Re: [Boost-users] [mpl] at_c type compiler error To: boost-users(a)lists.boost.org Message-ID: <471775BD.3010403(a)web.de> ------------------------------------------------------------ Chris Weed schrieb: >>> >>> { >>> >>> return t.get<0>(); >>> >>> } >> >> { >> >> return boost::get<0>(t); >> >> } Hi all -- Just to say I encountered what looks to be the same problem. Here is the note I left myself: // can later access the statistics as follows: // // std::tr1::tuple<int, double, double, double, double, double> stats; // stats = myTs.getStats(); // int count; // count = stats.get<0>(); // member function usage (could be buggy) // count = std::tr1::get<0>(stats); // free function usage Also: #include <tr1/tuple> // n-tuples I relealize this is the TR1 version and not the Boost version. Please let me know if some stripped down example code would be useful. My development environment for the record: boost : 1.34.1 (via ./configure; make all; make install) gcc : 4.1.2 20060928 (prerelease) (Ubuntu 4.1.1-13ubuntu5) os : Ubuntu 6.10, Linux 2.6.17-12-generic (7.04 is current) hardware : Toshiba Tecra A2 330 laptop (purchased Aug-2004) cheers Robbie --- Robbie Morrison PhD student -- policy-oriented energy system simulation Institute for Energy Engineering (IET) Technical University of Berlin (TU-Berlin), Germany [from IMAP client]

5 5

Adjacency matrix problem
by abhishek.v＠tcs.com 22 Oct '07

22 Oct '07

Hi I m trying to implement adjacency matrix. typedef boost::adjacency_matrix<boost::undirectedS> Graph; Graph(N); Now here can the N be a variable as what i want is create graph by varying the value of N which taken as an input from user. Following is the snippet of my code... int main(int argc, char *argv[]) { int SIZE; cout<<"\n Enter the matrix size (nxn) :-"; cin>>SIZE; Graph g(SIZE); .................... ..................... return 0; } And the runtime error is as follows :- Assertion failed: false, file BoostLib/boost_1_34_1/boost/graph/adjacency_matrix.hpp, line 941 Same code works fine in case of adjacency_list but in case of adjacency_matrix it is giving run time error. What could be the reason for this... Thanks Abhishek Vyas Tata Consultancy Services Mailto: abhishek.v(a)tcs.com Website: http://www.tcs.com ____________________________________________ Experience certainty. IT Services Business Solutions Outsourcing ____________________________________________ =====-----=====-----===== Notice: The information contained in this e-mail message and/or attachments to it may contain confidential or privileged information. If you are not the intended recipient, any dissemination, use, review, distribution, printing or copying of the information contained in this e-mail message and/or attachments to it are strictly prohibited. If you have received this communication in error, please notify us by reply e-mail or telephone and immediately and permanently delete the message and any attachments. Thank you

2 1

problem integrating lambda and gsl
by Luke Shulenburger 22 Oct '07

22 Oct '07

I am writing some code that involves several uses of the gnu scientific library that I am trying to streamline using boost lambda. I am having some newbie problems with some rather lengthy compiler errors. Here's a fairly minimal set description of my problem. First I have a functor class that does some relatively complicated processing: class functor : public std::unary_function<double,double> { public: double operator()(double x) const { // do some work here return something; } } For use with the gsl, I have created a wrapper to make this linto something that can be handled by the gsl routines: template<class func> double gslfunc(double x, void* p) { func* f = reinterpret_cast<functor*>(p); return (*f)(p); } I now have the following function that allows me to use gsl routines: template<class func> double myRoutine(func& f) { gsl_function F; F.function = &gslfunc<functor>; F.params = reinterpret_cast<void*>(&f); // call some gsl routines with the gsl_function return something; } I can now use gsl routines on the above functor with no difficulty. What I would like to be able to do is use lambda expressions to make slight modifications to the underlying functor, for example take the following code snippet: functor f; myRoutine(bind(f,_1)+1); However, this dies a horrible death upon compilation complaining that there is no available conversion from the type that is generated by bind(f,_1)+1 to void*. I suspect that there is no way to get a pointer to a lambda'd function and that this is causing my problem. Am I correct? I would happy to provide any more details about what I am doing or the gsl bindings if that would be helpful. Thanks for your consideration. Luke Shulenburger (shulenbu(a)uiuc.edu)

1 0

boost::format: get expected number of arguments
by Ruben Lopez 22 Oct '07

22 Oct '07

Hi, I am using boost::format in my project, and I would need to know what is the exact number of arguments expected, because the format is written by the end-user, and the program has no way to know it. I have seen that the format class actually has some attributes with this value, but I couldn't see any public methods to get them. Is there any way of getting the expected number of arguments? Thanks in advance.

1 0

which libs to link to for serialization?
by Mark Wyszomierski 21 Oct '07

21 Oct '07

Hi, I have an installer for win32 which installs the boost libraries. I'm trying to link to the serialization libraries, and several copies have been created in the boost/lib folder: libboost_serialization-vc80-mt-1_33_1.lib libboost_serialization-vc80-mt-gd-1_33_1.lib libboost_serialization-vc80-mt-s-1_33_1.lib libboost_serialization-vc80-mt-sgd-1_33_1.lib Which version should be linked to when? Right now I'm using a trial and error method to see which one my win32 apps can link to without linking errors. Thanks, Mark

2 1

Parag is your Yaar! :)
by Parag 21 Oct '07

21 Oct '07

1 0

[date_time] Error including posix_time headers using VC80
by Wilson Barbosa 20 Oct '07

20 Oct '07

Hi there, I've got the following error messages when including posix_time.h or posix_time_types.h on a project: ------ Build started: Project: ics99200, Configuration: Debug Win32 ------ Compiling... controle_proc_ora.cpp c:\arquivos de programas\microsoft visual studio 8\vc\platformsdk\include\rpcndr.h(147) : error C2632: 'char' followed by 'int' is illegal c:\arquivos de programas\microsoft visual studio 8\vc\platformsdk\include\rpcndr.h(147) : warning C4091: 'typedef ' : ignored on left of 'unsigned char' when no variable is declared c:\arquivos de programas\microsoft visual studio 8\vc\platformsdk\include\wtypes.h(1075) : error C2371: 'BOOLEAN' : redefinition; different basic types c:\arquivos de programas\microsoft visual studio 8\vc\platformsdk\include\winnt.h(770) : see declaration of 'BOOLEAN' I'm using Visual C++ 8.0 compiler (VS2005 Pro), MS-Windows XP Pro SP2 and the default options on the Boost build. Does anybody have a clue of what is (exactly) happening here and about any solution/workaround for this? TIA, WB::

1 0

[Graph] edge list cache
by Krishna Roskin 20 Oct '07

20 Oct '07

Hello all, I'm working on implicitly defined graphs using the graph boost library. The out_edges function (for example) generates the list of out edges on the fly and returns a shared_container_iterator. Creating this list takes a non-trivial amount of time and so I'd like to cache the result of out_edges so that future calls don't duplicate. But I'm not sure how to do this and still fit within the "BGL way." My first instinct was to add the cache as a member of the graph. But this means that out_edges has to take a non-const reference so that it can update the graph and this causes problems for filtered_graph and reverse_graph. Right now I'm toying with making my cache global and keyed to the address of the graph but this seems inelegant. Anyone have any suggestions? -krish

2 2

[general] bash script to filter compiler output
by Robbie Morrison 20 Oct '07

20 Oct '07

Hi all Postings here often comment on the amount of compiler output generated when using Boost libraries. I found the following script quite useful when trying to track down both bad practice and real errors in my own code. It works something like this: $ callfilter -6 make This script filters the compiler output based on the options -0 thru -7. Option -- has no effect and is provided for typing convenience. Options -5 and -6 are typically the most useful. Some comments at the start of the script provide more information. I find g++ -Weffc++ often rather helpful, but its use definitely requires some form of output screening. Here is the usage message for this script: usage: callfilter [-01234567] <call> modified call with output filtering callfilter -- <call> straight call (provided for convenience) callfilter --help|-h display usage message and quit purpose: append selected filters to build calls to suppress Boost library output additive: 0 2>&1 1 | grep -v 'boost::' | grep -v 'std::tr1::' 2 | grep -v '/include/boost-1_34_1/' | grep -v '/include/c++/4.1.2/tr1/' 4 | grep -v 'instantiated from' examples: $ callfilter -- make $ callfilter -7 g++ -Wall -Weffc++ usingboost.cc -o usingboost If anyone has improvements, let me know. I will endeavor to incorporate these and then repost in a couple of weeks. cheers Robbie Morrison --- [from IMAP client] #! /bin/bash # Copyright: (c) 2007 Robbie Morrison. All rights reserved. # License: Unrestricted reuse in any form, for any purpose. # Version: $Id: callfilter,v 1.7 2007/10/19 11:26:41 robbie Exp $ # Purpose: take a command-line build call and append some # selected 'grep' filters to reduce the amount of output # displayed. A build call could be just 'make' or perhaps # g++ with a string of arguments. The individual filters # are intended to suppress the screenfulls of warnings # generated by Boost library headers, while leaving the # compiler messages that relate more directly to your code. # The ability to mix different combinations of filters is # often useful. The UNIX 'grep' utility does the filtering # and -v is the 'invert-match' option. This script can be # particularly valuable in combination with the '-Weffc++' # option for g++. In fact, the script was developed # because of a comment suggesting such an approach on the # g++ manpage. # # See also: for a more sophisticated approach, # refer to: http://www.bdsoft.com/tools/stlfilt.html # define some suitable filters -- please modify to suit your needs FILTER_0="2>&1" # useful when capturing output FILTER_1="| grep -v 'boost::' | grep -v 'std::tr1::'" FILTER_2="| grep -v '/include/boost-1_34_1/' | grep -v '/include/c++/4.1.2/tr1/'" FILTER_4="| grep -v 'instantiated from'" # preamble SCRIPT=$(basename "$0") # for console reporting # script exit codes E_SUCCESS=0 E_USAGE=2 E_NO_GREP=3 # functions function display_help # using simple echo statements { echo echo " usage: $SCRIPT [-01234567] <call> modified call with output filtering" echo " $SCRIPT -- <call> straight call (provided for convenience)" echo " $SCRIPT --help|-h display usage message and quit" echo " purpose: append selected filters to build calls to suppress Boost library output" echo " additive: 0 $FILTER_0" echo " 1 $FILTER_1" echo " 2 $FILTER_2" echo " 4 $FILTER_4" echo " examples: \$ $SCRIPT -- make" echo " \$ $SCRIPT -7 g++ -Wall -Weffc++ usingboost.cc -o usingboost" echo } function make_filter # with conventional rather than bitwise arithmetic { local base=4 # option range is from 0 to (2 ** base - 1) local num=${opt#-} # strip leading - while [ $base -gt 0 ] do if [ $num -ge $base ] then let "num = num % base" # remainder operator case "$base" in 1) filter="$FILTER_1 $filter" ;; 2) filter="$FILTER_2 $filter" ;; 4) filter="$FILTER_4 $filter" ;; esac fi let "base = base / 2" # divide by two done filter="$FILTER_0 $filter" } # process our part of the command-line case "$1" in --help|--hel|--he|--h|-help|-hel|-he|-h|-\?) display_help exit $E_SUCCESS ;; -[0-7]) opt="$1" shift make_filter ;; --) shift ;; "") echo "$SCRIPT: try --help for usage" exit $E_USAGE ;; esac # confirm that grep is present on the system if [ ! $(which grep) ] then echo "$SCRIPT: required utility not found: grep" echo "$SCRIPT: unable to continue" exit $E_NO_GREP fi # active code echo "$SCRIPT: call : $*" echo "$SCRIPT: filter : $filter" eval "$* $filter" # actual call # housekeeping exit $E_SUCCESS # end of file

2 1