- Boost-users - lists.boost.org

extraction invariants from intensive-loop based on template parameters
by Hicham Mouline 19 Feb '09

19 Feb '09

Hello, I have an algorithm like: template <typename Upper, typename Lower> class X { static void f() { .... for (size_t n=0; n<=N; ++n) { // process for j=0 at every n if Lower::IsConstant is false for (size_t j=1; j<=J-1; ++j) { } // process for j=J at every n if Upper::IsConstant is false } .... } So I extracted the double for loop in a separate wrapper class and I verbosely wrote 4 specializations with template args begin bool : <true, true> <false, true> <true, false> <false, false> for e.g. for <true, true> // process for j=0 once for (size_t n=0; n<=N; ++n) { for (size_t j=1; j<=J-1; ++j) { } } // process for j=J once Any advice? Is there a better way than writing explicitly the tmpl specializations? Boost.PP maybe? rds,

2 1

compiling 32 bit boost libraries on a 64 bit system? (gcc, Linux)
by Ákos Maróy 19 Feb '09

19 Feb '09

Hi, I'm trying to compile boost into 32 bit libraries on a 64 bit Linux box. with gcc, with would simply mean to add the flag -m32 to gcc. but I can't seem to make bjam do this. I've tried a number of approaches: - export and set the environment variables CFLAGS, CXXFLAGS and LDFLAGS to -m32 - set the following in tools/build/v2/user-config.jam: cflags=-m32 cxxflags=-m32 linkflags=-m32 - set the following in tools/build/v2/user-config.jam: using gcc : 4.1.2 : g++-4.1.2 : -m32 ; (based on the documentation found at http://beta.boost.org/doc/tools/build/doc/html/bbv2/reference/tools.html#bb… ) but to no avail - the -m32 flag is not honored, and the compiled boost libraries are 64 bit :( what am I doing wrong? Akos

2 2

A need for a complex pool type container
by Robert Dailey 19 Feb '09

19 Feb '09

My requirements are a bit complex and involved, but I'll try to be as brief as possible. We have this "scheduler" type system where for each unit of time (think of 1 unit as 1 ms) there is a "bucket" of events that shall be processed. This is relative. Suppose the current time is 103. We will have a record of events to be processed at time 103, 104, 105, ..., 203. When all of the events in the "bucket" for time stamp 103 are processed, we "pop_front" to remove the bucket for 103. Everything is essentially shifted forward. Hopefully this does not require physically shifting everything forward, and simply is just a matter of using a lesser index to access the same element later on. I insert new events into existing buckets anywhere in the container, it does not necessarily have to be from the back. I always remove items from the front since they are processed linearly. One can kind of think of this as a std::deque< std::vector<> >, but the containing deque would need to be given a fixed size of 100 (as in the example above). The tricky part is shifting all of the "buckets" (the buckets in this case would be the inner-vector objects) forward when something from the front is removed. The key here is that I do not want to call pop_front() each time a bucket is processed because I want to avoid constant new/delete operations since insertion and removal will be happening quite frequently. I'm not sure what the best approach to solving this problem is. I'm hoping someone in the community will be able to suggest a way in which boost can easily solve this problem. Thanks to everyone for taking the time to read this.

2 6

[serialization] problem exporting derived classes
by Jens Weggemann 19 Feb '09

19 Feb '09

Hello, I've been trying out boost::serialization and ran into a problem trying to serialize objects of derived classes. I set up this simple class hierarchy: class BaseNode { std::vector<BaseNode*> children_; ... friend class boost::serialization::access; template<class Archive> void serialize(Archive &ar, unsigned version) { ar & children_; } }; class DerivedNode : public BaseNode { ... friend class boost::serialization::access; template<class Archive> void serialize(Archive &ar, unsigned version) { ar & boost::serialization::base_object<BaseNode>(*this); } }; All works fine as long as I register the types with the archives: arch.register_type(static_cast<BaseNode*>(NULL)); arch.register_type(static_cast<DerivedNode*>(NULL)); Now the documentation suggests exporting the classes as an alternative to that. But when I instead put BOOST_CLASS_EXPORT(BaseNode) BOOST_CLASS_EXPORT(DerivedNode) the polymorphism is not handled correctly and e.g. a tree of 3 DerivedNodes is restored as a single BaseNode object. What am I missing here? The full code for this test is at http://noonti.de/tree_serialization.cpp.txt Thanks, Jens

2 1

Graph lib: read_dimacs_max_flow segment fault
by Xian Chen 18 Feb '09

18 Feb '09

Hi All, I compile read_dimacs_max_flow example ( http://www.boost.org/doc/libs/1_38_0/libs/graph/example/read_write_dimacs-e…) with these warnnings. g++ -I/home/xianchen/opt/include/boost-1_38 -Wno-deprecated -c read.cpp -o objects/simu.o /home/xianchen/opt/include/boost-1_38/boost/graph/read_dimacs.hpp: In function 'int boost::read_dimacs_max_flow(Graph&, CapacityMap, ReverseEdgeMap, typename boost::graph_traits<G>::vertex_descriptor&, typename boost::graph_traits<G>::vertex_descriptor&, std::istream&) [with Graph = main()::Graph, CapacityMap = boost::adj_list_edge_property_map<boost::directed_tag, long int, long int&, unsigned int, boost::property<boost::edge_capacity_t, long int, boost::property<boost::edge_reverse_t, boost::detail::edge_desc_impl<boost::directed_tag, unsigned int>, boost::no_property> >, boost::edge_capacity_t>, ReverseEdgeMap = boost::adj_list_edge_property_map<boost::directed_tag, boost::detail::edge_desc_impl<boost::directed_tag, unsigned int>, boost::detail::edge_desc_impl<boost::directed_tag, unsigned int>&, unsigned int, boost::property<boost::edge_capacity_t, long int, boost::property<boost::edge_reverse_t, boost::detail::edge_desc_impl<boost::directed_tag, unsigned int>, boost::no_property> >, boost::edge_reverse_t>]': read.cpp:86: instantiated from here /home/xianchen/opt/include/boost-1_38/boost/graph/read_dimacs.hpp:108: warning: deprecated conversion from string constant to 'char*' /home/xianchen/opt/include/boost-1_38/boost/graph/read_dimacs.hpp:108: warning: deprecated conversion from string constant to 'char*' ...... My GCC is: Using built-in specs. Target: i686-pc-linux-gnu Configured with: ../gcc-4.2.4/configure --prefix=/home/xianchen/opt --enable-shared --enable-threads=posix --enable-languages=c,c++ --enable-checking=release --enable-mpfr=/home/xianchen/opt/ --with-system-zlib Thread model: posix gcc version 4.2.4 I run this program using data "boost/libs/graph/example/max_flow.dat" But get segment fault: *** glibc detected *** free(): invalid pointer: 0x08053ff4 *** Aborted Can any one tell me the why this happen? The program aborts under GCC.4.3.3 either. Which version of GCC is the most boost friendly? Thanks in advance, Xian

1 0

[serialization] export.hpp changes from 1.37 to 1.38
by Chard 18 Feb '09

18 Feb '09

The BOOST_CLASS_EXPORT.. macros have been moved into a detail namespace (1.38) from a (public) anonymous namespace (1.37). I'm guessing that the reasoning for this was in order to avoid many definitions of the class singletons. However, this appears to have enforced that all of the guids (for all archive types) now have to reside in a single compilation unit. The attached code illustrates the limitation; this compiles under 1.37 but results in duplicate symbol definitions in 1.38 (with MSVC7.1). It worked with 1.37 because the anonymous namespace ensured no symbol clashes, but now the symbols are all in the same namespace. The example shows a case where two archive types (xml and binary) might be in use. The obvious solution to this (simple) case would be to just use a single compilation unit, which brings in the guids for both binary and xml. However, turning a blind eye to the 'untidiness', this fix does not help with the project we're working on. In our project, there is not just a simple type being serialized; there are a lot more classes and a bit of mpl thrown in. This actually resulted in breaching the COFF symbol limit (approx. 65k) with MSVC7.1, so we were forced to split the serialization. In fact, we had to split both the save and load parts of the binary serialization. Our current workaround is to use the archive.register_type<...> method, and ignore the export macros, but this does mean ensuring all types are covered (as described in the documentation).

4 9

Boost and Real Time
by Brian Wrenn 18 Feb '09

18 Feb '09

Hi, I'm an experienced C++ programmer, but very new to Boost. I'm looking into making use of it for my next project, which will run on LynxOS, a real time system. Does anyone have any experience, knowledge of pitfalls, or words of wisdom to share about building and running with Boost, or more likely a subset of Boost's libraries, for a real time system? Thanks for any help. -Brian

2 1

[asio] io_service use in dll leads to service_registry crash
by joshr 18 Feb '09

18 Feb '09

Configuration: 1) Boost 1.37 2) Visual Studio 2008 3) Windows executable using boost's multi-threaded static library, using a static linked multi-threaded run time. 4) A dll I have as part of the project is compiled with the same settings. 5) MFC is not used. The executable creates an io_service (io) and pool of worker threads, each calling io.run(). Afterwards, the executable begins loading DLLs I've written to add specific functionality (i.e plugins). Each of these plugins are passed the executable's io_service so that they can perform asynchronous work against the thread pool. One of the DLLs creates a boost::asio::ip::tcp::acceptor object, passing the io_service to its constructor. This, in turn, eventually calls use_service, which creates a new service and associates it with the service_registry. This is all happening in the DLL. At shut down, I begin unloading the DLLs, giving each a chance to shut down everything (which they do). I then call FreeLibrary() on the DLL's HMODULE handle. After my application completes shutdown, all the destructors hit, as expected. When the destructor of the service_registry hits, I believe its linked list of services is still pointing to a service located in the memory space of the DLL. This results in a crash. I believe this is the case as I can track the pointer to the acceptor_service in a Watch Window. The pointers and the object remain valid until I call "FreeLibrary", at which point they become worthless. I think this is because the DLL and the executable have their own CRT memory space, due to the way I'm linking in the CRT. The service for the acceptor is being created in the DLL's space. When FreeLibrary is called, this space is released to the OS, making the pointer to the object worthless. If that's the case, I believe I'll need to supply my own service creation routine for the DLLs so that the services are created in my executable's memory space. Is this possible? Is there a way to get the service_registry to release a service when it's no longer in use anywhere, or to specify my own new/delete routines the use_service calls should use? Or, is using ASIO in a DLL like this a really really bad idea? If that's the case, how can I tell what other objects in boost could cause these sorts of problems? I'll try to get a project together that reproduces the issue as soon as I can. I can't use the code I have at the moment as it is proprietary. Thanks for any input anyone can provide. -- View this message in context: http://www.nabble.com/-asio--io_service-use-in-dll-leads-to-service_registr… Sent from the Boost - Users mailing list archive at Nabble.com.

2 3

[ASIO] Newbie; How can I access remote_endpoint data in an SSL server
by May James 18 Feb '09

18 Feb '09

Can anybody help a boost::newbie I have generated an ASIO SSL server based on the sample code "server.cpp" How can I determine the Client Address from the ssl_socket which is defined as typedef boost::asio::ssl::stream<boost::asio::ip::tcp::socket> ssl_socket; relevant code is: acceptor_.async_accept(new_session->socket(), boost::bind(&server::handle_accept, this, new_session, boost::asio::placeholders::error)); Thank You

2 1

[assign] Nested initialization of std containers
by Joseph Lisee 18 Feb '09

18 Feb '09

I am trying to use list_of and map_list_of to initialize a std::vector containing std::maps. I am using Boost 1.37 and Mac OS X 10.4 with Apple's G++ 4.0.1. Here is a complete example: #include <vector> #include <map> #include <string> #include <boost/assign/list_of.hpp> using namespace boost::assign; typedef std::map<int, std::string> dict; typedef std::vector<dict> list; int main() { // Works dict d = map_list_of(1,"B")(2,"A"); list l1 = list_of(d); // Does not compile list l2 = list_of(map_list_of(1,"B")(2,"A")); return 0; } Is there some way to get this to work?

2 1