July 2003 - Boost-users - lists.boost.org

profiling ublas code
by cerenoc 15 Jul '03

15 Jul '03

Hi, I am trying to optimize my ublas-containing code. The profile from gprof showed these two functions as the top 2 time-consuming ones. Could somebody explain to me a little bit about what these mean? % cumulative self self total time seconds seconds calls ms/call ms/call name 23.76 93.11 93.11 137604968 0.00 0.00 vector<int, allocator<int> > * __uninitialized_copy_aux<vector<int, allocator<int> > *, vector<int, allocator<int> > *>(vector<int, allocator<int> > *, vector<int, allocator<int> > *, vector<int, allocator<int> > *, __false_type) 21.49 177.31 84.20 41208000 0.00 0.00 vector<int, allocator<int> >::operator=(vector<int, allocator<int> > const &) Thanks, Julian

2 1

Assertion tripped in boost::uniform_int
by mvogt＠juptech.com 15 Jul '03

15 Jul '03

Hello. I just experienced an assertion failure using the boost::minstd_rand generator: /home/packages/boost_1_29_0/boost/random/uniform_int.hpp:50: boost::uniform_int<UniformRandomNumberGenerator, IntType>::uniform_int (UniformRandomNumberGenerator &, IntType, IntType) [with UniformRandomNumberGenerator = boost::random::linear_congruential<int32_t, 48271, 0, 2147483647, 399268537>, IntType = int]: Assertion `min < max' failed. I use the following code to generate random numbers: typedef boost::minstd_rand RngType; extern RngType pseudo_rng; template <class T> T random(T min, T max) { boost::random_number_generator<RngType, T> rng(pseudo_rng); return rng(max - min + 1) + min; } Am I doing something wrong, or is there a fault in the uniform_int algorithm? Thanks, Matt. ********************************************************************** This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify mail.administrator(a)jupiters.com.au ********************************************************************** [Non-text portions of this message have been removed]

4 4

Spirit v1.7.0 and v1.6.1 Released
by Joel de Guzman 13 Jul '03

13 Jul '03

Hello, Spirit v1.7.0 and v1.6.1 have been released. Get it here: Spirit v1.7.0: http://tinyurl.com/gssn Spirit v1.6.1: http://tinyurl.com/gsss Take note that by convention, odd minor version releases (e.g. 1.7.0) are developmental while even minor version releases (e.g. 1.6.1) are stable. What's new: http://spirit.sourceforge.net/distrib/change_log.html Have fun!!! Cheers, -- Joel de Guzman joel at boost-consulting.com http://www.boost-consulting.com http://spirit.sf.net

1 0

Logging library ?
by drkm 13 Jul '03

13 Jul '03

Hello. Is there any logging fonctionality in the Boost libraries ? I found nothing. If there is nothing, is there a plan to add a such library to Boost ? --drkm

2 1

MPL::lambda issue: what am I doing wrong?
by Brian Simpson 11 Jul '03

11 Jul '03

I want to express a "type-mapping" scheme. My implementation idea is this: have the user define an MPL sequence comprised exclusively of std::pair derivations. He can then instantiate my template class with his sequence as the argument. My template then derives two MPL sequences, one containing the "first" types, the other containing the "second" types. Here is the attempt: <code> namespace ls { using namespace boost::mpl; using namespace placeholders; template <class TypePairSequence> struct list_splitter { template <class STDPAIR> struct first_type { typedef typename STDPAIR::first_type type; }; template <class STDPAIR> struct second_type { typedef typename STDPAIR::second_type type; }; typedef typename transform < TypePairSequence , first_type<_1> > ::type first_types; typedef typename transform < TypePairSequence , second_type<_1> >::type second_types; }; }//end namespace ls </code> Here is an attempt to instantiate: 1> typedef boost::mpl::list< std::pair<long, short>, std::pair<char, double> > associated_types; 2> typedef ls::list_splitter<associated_types> split_associated_types; 3> typedef split_associated_types::first_types first_types; 4> typedef split_associated_types::second_types second_types; 5> typedef boost::mpl::at_c<first_types, 0>::type first_first_t; Code lines 1-2 compile fine. When I add line 3, I get the following error: "no type named `first_type' in `struct boost::mpl::arg<1>'". If you need more: <error> temp.h: In instantiation of `ls::list_splitter<associated_types>::first_type<boost::mpl::arg<1> >': boost/mpl/transform.hpp:41: instantiated from `boost::mpl::apply1<ls::list_splitter<associated_types>::first_type<boost::m pl::arg<1> >, std::pair<char, double> >' boost/mpl/transform.hpp:41: instantiated from `boost::mpl::aux::transform_op<ls::list_splitter<associated_types>::first_ty pe<boost::mpl::arg<1> > >::apply<boost::mpl::null_node, std::pair<char, double> >' boost/mpl/aux_/preprocessed/gcc/fold_backward_impl.hpp:74: instantiated from `boost::mpl::aux::fold_backward_impl<2, boost::mpl::list_iterator<boost::mpl::list2<std::pair<long int, short int>, std::pair<char, double> > >, boost::mpl::list_iterator<boost::mpl::null_node>, boost::mpl::null_node, boost::mpl::protect<boost::mpl::aux::transform_op<ls::list_splitter<associat ed_types>::first_type<boost::mpl::arg<1> > > >, boost::mpl::arg<1> >' boost/mpl/fold_backward.hpp:45: instantiated from `boost::mpl::fold_backward<associated_types, boost::mpl::null_node, boost::mpl::protect<boost::mpl::aux::transform_op<ls::list_splitter<associat ed_types>::first_type<boost::mpl::arg<1> > > >, boost::mpl::arg<1> >' boost/mpl/transform.hpp:64: instantiated from `boost::mpl::algo_::transform<associated_types, ls::list_splitter<associated_types>::first_type<boost::mpl::arg<1> > >' temp.h:35: instantiated from `ls::list_splitter<associated_types>' temp.h:53: instantiated from here temp.h:19: no type named `first_type' in `struct boost::mpl::arg<1>' </error> I'm sure I'm overlooking something simple. Could someone point me in the right direction? I'm using gcc3.2.2 on Solaris 9. TIA, Brian

2 4

Defining tuples and variants based on mpl::sequences
by Brian Simpson 11 Jul '03

11 Jul '03

Here is the problem: A tuple or variant needs to be defined, whose types are derived from an mpl::list (or other model of MPL Sequence). How is it to be done? I have considered that I might be able to define the tuple/variant with the types and then generate the sequence from there, but that still leaves me wondering how to define another tuple/variant with a sub- or super-set of the types in the generated sequence. Consider that I am supplied with a sequence of types, and that I want to define either 1)a tuple of pointers to them, or 2)a variant based on those types as well as pointers to them. I have considered that it would be a good thing to provide sequence-based tuples/variants and still provide the current variadic template (<--is that the correct term?) versions--which could be implemented in terms of the sequence-based ones (in my uneducated opinion). But I wonder if there is another solution already available which I have not considered. TIA for your consideration, Brian

1 0

Boost.Python interface to Boost Graph Library (BGL)
by BP 11 Jul '03

11 Jul '03

I would like to use the Boost Graph Library (BGL), but I'm not much of a C++ programmer. I'd feel much more comfortable working in Python. Can I download the Boost Graph Library wrapped in Boost.Python? I'd rather not wrap myself if it already exists. Thanks. - BP __________________________________ Do you Yahoo!? SBC Yahoo! DSL - Now only $29.95 per month! http://sbc.yahoo.com

3 2

templated 'handle' in pimpl code? (Slightly off topic)
by Kevin Wheatley 11 Jul '03

11 Jul '03

Hi, I wondered if there are any examples of a templated handle class within the boost code base I could look at? I'm trying to hide both the implementation of the body class and that of the template type, something like that below but my compiler is complaining about partial specialization in the declaration of the implementation class in handle.cpp I know its not strictly boost related, but I don't get any of the comp.lang.* groups. Also before anybody suggests I should be using a smart pointer for the impl variable I'd like to avoid having to introduce the boost library to the user of the two .h files. My goal was that the users would only need the *.h files and I could provide fred.o fred2.o fred3.o etc. as new options for "fred" come along. I know that using the template doesn't appear to be getting me anywhere towads my original goal (generally needing the implementation to get the templated code to work!) but I'd like to see if I could at least get the code to compile by studying anything similar as it may have some use in other areas. Thanks Kevin // -------------- fred.h class fred { public: void somefunction(); }; // -------------- handle.h template <typename T> class handle { public: handle(); ~handle(); void callsomefunctionviaimpl(); private: template <typename U> class handleimpl; handleimpl<T> *impl; }; // -------------- fred.cpp #include <iostream> using namespace std; #include "fred.h" void fred::somefunction() { cout << "fred::somefunction()" << endl; } // -------------- handle.cpp #include <iostream> using namespace std; #include "handle.h" template <class T, class U> class handle<T>::handleimpl<U> { public: void implfunctiononfred() { myfred.somefunction(); } private: static T myfred; }; template <class T> handle<T>::handle() : impl(new handle<T>::handleimpl<T>) { } template <class T> handle<T>::~handle() { delete impl; } template <class T> void handle<T>::callsomefunctionviaimpl() { cout << "handle calling" << endl; impl->implfunctiononfred(); } // ----------- Main code ---------------- #include "fred.h" #include "handle.h" int main(int argc, char** argv) { handle<fred> foo; foo.callsomefunctionviaimpl(); return 0; } -- | Kevin Wheatley | These are the opinions of | | Senior Do-er of Technical Things | nobody and are not shared | | Cinesite (Europe) Ltd | by my employers |

1 1

multi-array versus valarray
by lformaggia 11 Jul '03

11 Jul '03

Hi, I am a new user of boost and I am interest in the multi-array library. However it is not clear to me which are the advantages versus the std::valarray and std::slice classes. There are quite a few numeric functions provided by C++ that work on valarrays (the pow() for example), is there any equivalent for multi-arrays )it would be very nice to have a norm() function at least for two-dimensional arrays!) or you have to write your own. Sorry if this question may seem trivial, yet I have to choose which basic multiarray classes to implemet for my code (it is numerical code based on the finite element method) and so far my choice is between multi-array or valarray + slice. Maybe somebody of you have some experience on this matter and my help in taking an appropriate decision. Thank you Luca

1 0

Having problems with boost::signal and boost::function
by subsample 11 Jul '03

11 Jul '03

I'm having a problem (which was really hard to decipher) when connecting a function object to a signal. Thing is I default construct (no argument) a function object and connect it to the signal (using boost::ref), then at a later point in time I actually assign something to the function object (using boost::bind). But firing the signal after this causes a bad_function_call to be thrown. On the other hand; if I construct the function object with some dummy object before connecting it to the signal, and later do the assignment, the signal does its thing. If someone could enlighten me as to whats going on here, I'd be grateful .. : )

2 1