- Boost-users - lists.boost.org

Cross compiling boost
by Peter Poulsen 30 May '07

30 May '07

I'm trying to build boost for an arm processor, but all the hidden bjam stuff is simply to much, so I have to ask for help :) All my build tools are in ~/arm-tools/. I'm on a Linux x86 and the cross compiler I'm trying to use is a gcc version 3.4.4. I have tried various build commands. The latest being: ~/boost_1_34_0 $ bjam --build-dir=/tmp/build-boost --toolset-gcc-3.4.4 --with-regex -sGXX=arm-wrs-linux-gnueabi-g++ stage But I still get: ~/boost_1_34_0 $ file stage/lib/libboost_regex-gcc34.so stage/lib/libboost_regex-gcc34.so: ELF 32-bit LSB shared object, Intel 80386, version 1 (SYSV), stripped Can anybody explain to me what I must do to compile boost-regex for arm? btw I have found others trying to do the same but they seem to end up hacking the bjam files, and I have no idea of how to that. /peter -- A: Because it messes up the order in which people normally read text. Q: Why is it such a bad thing? A: Top-posting. Q: What is the most annoying thing on usenet and in e-mail?

2 1

[mpl] call functor for each type in a type list
by Ares Lagae 30 May '07

30 May '07

I want to call a functor for each type in a type list. Consider the following program. #include <iostream> #include <boost/mpl/apply.hpp> #include <boost/mpl/empty_base.hpp> #include <boost/mpl/fold.hpp> #include <boost/mpl/identity.hpp> #include <boost/mpl/inherit.hpp> #include <boost/mpl/joint_view.hpp> #include <boost/mpl/lambda.hpp> #include <boost/mpl/pair.hpp> #include <boost/mpl/transform.hpp> #include <boost/mpl/vector.hpp> template <typename T, typename Functor> class functor_list_member { public: functor_list_member() { Functor f; f.template operator()<T>(); } }; template<typename Sequence, typename Functor> class functor_list: private boost::mpl::fold< Sequence, boost::mpl::empty_base, boost::mpl::inherit< boost::mpl::_1, functor_list_member<boost::mpl::_2, Functor> > >::type {}; typedef boost::mpl::vector<bool, char, short, int, long, float, double> types; struct functor { template <typename T> void operator()() { std::cout << __PRETTY_FUNCTION__ << std::endl; } }; typedef functor_list<types, functor> my_functor_list; int main(int argc, char* argv) { my_functor_list(); } The program outputs void functor::operator()() [with T = bool] void functor::operator()() [with T = char] void functor::operator()() [with T = short int] void functor::operator()() [with T = int] void functor::operator()() [with T = long int] void functor::operator()() [with T = float] void functor::operator()() [with T = double] However, I want to create a functor object first, and pass it to each functor_list_member, like this: template <typename T, typename Functor> class functor_list_member { public: functor_list_member(Functor f) { f.template operator()<T>(); } }; template<typename Sequence, typename Functor> class functor_list: private boost::mpl::fold< Sequence, boost::mpl::empty_base, boost::mpl::inherit< boost::mpl::_1, functor_list_member<boost::mpl::_2, Functor> > >::type { public: functor_list(Functor f) /* how do I call base(f) ? */ { } }; typedef boost::mpl::vector<bool, char, short, int, long, float, double> types; struct functor { template <typename T> void operator()() { std::cout << __PRETTY_FUNCTION__ << std::endl; } }; typedef functor_list<types, functor> my_functor_list; int main(int argc, char* argv) { functor f; my_functor_list(f); } but I do not know how to call the constructor of the base class in functor_list::functor_list(Functor f). (see comment /* how do I call base(f) ? */). Can someone help me? Best regards, -- Ares Lagae Computer Graphics Research Group, Katholieke Universiteit Leuven http://www.cs.kuleuven.be/~ares/

2 2

help with fusion filter_view
by Christian Holmquist 30 May '07

30 May '07

Hello, I'm trying to create a fusion::filter_view with a predicate that must access, for each element, the element at the same position in an 'external' typelist. Let's say I have two sequences like this typedef mpl::vector<int&> mpl_vec; typedef fusion::vector<int> f_vec; Now, I would like a view into f_vec of, for instances, all references in mpl_vec. typedef fusion::filter_view<f_vec, is_reference<mpl::_> > f_view; This is obviously wrong; is_reference should evaluate with mpl_vec as its source. If I wrap is_reference like this template<class Sequence> struct is_reference_in { template<class T> struct apply : boost::is_reference<T> // Wrong, T should be extracted from Sequence { }; }; and create the view like typedef fusion::filter_view<f_vec, is_reference_in<mpl_vec> > f_view2; I gain access to the mpl_vec in my predicate, but since the predicate is immutable I cannot iterate over the Sequence template parameter to extract the type to send to is_reference. The way i'm solving this now is to store everything in the same vector, like typedef fusion::vector<fusion::pair<int, int&> > f_vec and create predicates that work with fusion::pair instead. This seems clumsy. Could zip_view help me out here? Any other suggestions? Thanks, Christian

1 0

function template as template parameter
by Ares Lagae 30 May '07

30 May '07

consider the following program: template <typename T> void function_template(T value) { std::cout << __PRETTY_FUNCTION__ << std::endl; } struct function_template_wrapper { template <typename T> static void call(T value) { function_template<T>(value); } }; int main(int argc, char* argv[]) { int i; function_template_wrapper::call<int>(i); } Is it possible to somehow parameterize the struct function_template_wrapper such that I can pass function_template as a parameter? E.g. function_template_wrapper<function_template>::call<int>(i); would call function_template<int>(i), and function_template_wrapper<other_function_template>::call<int>(i); would call other_function_template<int>(i) Underlying problem: I want to somehow pass a name of a function template to a class, that will instantiate the function template for a number of types (known at compile time), and store these instantiations in function objects, for later use. Best regards, -- Ares Lagae Computer Graphics Research Group, Katholieke Universiteit Leuven http://www.cs.kuleuven.be/~ares/

3 6

Serialization Performance
by Andrew VanUitert 30 May '07

30 May '07

I am using Boost Serialization as a basic RPC mechanism on an embedded Linux system, however it seems that the serialization library uses a lot of CPU cycles to process some rather simple messages. Each of the messages I am using passes a pointer to an object that is a serializable object representing the needed parameters for each type of message. As each message/command is passed from one system to another, the pointer is filled in with the de-marshaled parameters. So far this has worked splendidly except for the CPU utilization. The embedded processor is an Arm running at 300 MHz, and when I send it repeated messages the CPU utilization pegs. Profiling has revealed that the embedded processor is spending all its time in serialization library. For the most part, the message being sent back and forth are arrays of chars under 255 in size. Are there some general guidelines when considering performance with this library ? For instance, I read that the performance of 1.34 increased substantially with the use of a streambuf and a binary archive for arrays of basic types. However, the constructor that takes a streambuf is only available in the binary archive, and I am using a text archive to avoid ending and alignment issues. Is the use of a pointer for the parameters a place for optimization ? Andy

2 1

[iostreams] Are streams really models of Device?
by Chad Walters 30 May '07

30 May '07

Standard stream and stream buffers are supposed to be models of the Device concept (http://www.boost.org/libs/iostreams/doc/concepts/device.html) However, I am finding that their treatment by various parts of the library doesn't seem to be consistent. For example, streams don't actually seem to be usable everywhere a Device can be used. For example, if I define SinkOrStream as follows: using namespace boost::iostreams; typedef file_sink SinkOrStream; Then the following declarations compile properly: SinkOrStream mySink(mFileName); stream<composite<gzip_compressor, SinkOrStream> > out1(gzip_compressor(), mySink); stream<SinkOrStream> out2(mFileName); However, neither out1 nor out2 compile properly if I define SinkOrStream as: typedef stream<file_sink> SinkOrStream; If I understand the documentation correctly, these should both be legal, no? Does this work for others? I also find it a little strange that streams are supposed to be models of Device and they have a close() method, yet they do not support the closable_tag. Is this by design? Thanks for any help on clarifying these issues... Chad Walters

1 0

Any existing base64 encoding class/function in boost ?
by Raja Cherukuri 29 May '07

29 May '07

Is there a base64 encoding class/function in boost 1.33.x ? Thanks in Advance. - Raja

1 0

[Boost.Test] Embedding unit testing in DLL projects
by Tanner Davis 29 May '07

29 May '07

Hello- What is the best practice for embedding test cases into library (versus executable) projects? I have successfully used the Boost.Test framework to build and create EXE targeted, testing specific (MSVS) projects. However, much of what I need to unit-test is compiled into a Win32 DLLs, and I would much rather embed the test cases directly within that DLL project code (inside define blocks of course), rather than having to define a separate project to produce an EXE to execute the tests for each DLL. When I say "I would rather", it really means my manager demands it. It appears that the boost.test MACROs nicely obfuscate the normal main()-ish entry point for the testing. I have searched for information on overriding or extending the kick-off of whatever calls the init_unit_test_suite() so I could write my own custom entry points for testing the DLL as a post-build step, but have not found what I am looking for. So, my question is this. Is there a provided framework for testing code within a DLL project? And if not, is there a way to kick start the testing from a non-main method (say a custom DLL entry point), and would this strategy work within the confines of Boost.Test? I believe that other well known unit testing framework (which shall remain nameless) supports such a thing, and I would suspect Boost.Test does also. I apologize if this question is often asked or obviously documented elsewhere. Thank you for the assistance.

1 0

[Building 1.34] Linking against 1.33.1 and/or 1.34
by Christoph Duelli 29 May '07

29 May '07

Is it possible to configure Boost so that the library files built (linux, gcc 4.1) are *not* named like My problem: Boost 1.33.1 libs are a.t.m. named like libboost_regex-gcc.so Boost 1.34.0 libs are now named like libboost_regex-gcc41.so Consequently, my makefile that contain lines like -lboost_regex-gcc do not work with 1.34. If I 'fix' them for 1.34 my other machines (and colleagues) will be in trouble til they switch to 1.34 as well. Question: Is it possible to build boost (or write my makefiles) in a way so that my projects build with both Boost 1.33.1, 1.34 and so on (whatever is present)? Is there a standard or best practice for that? Thank you and best regards Christoph

4 3

[Boost.graph] compilation error for dijkstra_shortest_paths
by Nicolas Ray 29 May '07

29 May '07

Hello all, I have developed an application using the dijkstra_shortest_paths function of the Graph library of version 1.31 of Boost. When upgrading to Boost version 1.34, I have the following compilation error under Borland C++ Builder 6: [C++ Error] relaxed_heap.hpp(52): E2247'relaxed_heap<unsigned int, indirect_cmp<float*,less<float>>,vec_adj_list_vertex_id_map<VertexProperty, unsigned int>>::largest_key' is not accessible If I comment the call to the "dijkstra_shortest_paths" function in my code, there is no compilation error anymore. Is it related to how the default vertex_index is defined? If I add and use a custom vertex_index in the vertex property map, I get the same compilation error. I have tried to compile with version 1.33 and add the same error. I defined my graph as follow: struct vertex_altitude_t {typedef vertex_property_tag kind;}; typedef property<vertex_altitude_t, float> VertexProperty; typedef property<edge_weight_t, float > EdgeProperty; typedef adjacency_list<vecS, vecS, undirectedS, VertexProperty, EdgeProperty> Graph; Any help would be greatly appreciated, thank you very much, nicolas

2 2