- Boost-users - lists.preview.boost.org

Re: [Boost-users] [Serialization] unable to fix UNREGISTERED CLASSexception
by David Graf 17 Feb '10

17 Feb '10

Hello Robert Sorry to bother you again, but I am unable to use force_include correctly. I guest the makro BOOST_USED must be used, correct? I wrote BOOST_USED into each function header in the example. But it doesn't change anything. David On Feb 17, 2010, at 5:25 PM, Robert Ramey wrote: > This is likely due to the fact that a static linker will only include > modules which have have been explicitly referred to. The library uses > "force_include" to implement this when it can. You'll just have to figure > out what the linker is excluding and make sure it's included. > > > Robert Ramey > 830 Cathedral Vista Lane > Santa Barbara, CA 93110 > (805)569-3793 > www.rrsd.com > -----Original Message----- > From: David Graf [mailto:davidagraf@gmail.com] On Behalf Of David Graf > Sent: Wednesday, February 17, 2010 8:08 AM > To: boost-users(a)lists.boost.org; Robert Ramey > Subject: Re: [Boost-users] [Serialization] unable to fix UNREGISTERED > CLASSexception > > Hello Robert > > I fixed the example. I had to add somewhere a virtual function. Otherwise, > the polymorphism doesn't work. > > Unfortunately, I have a new problem. If I put the code into a static > library, the serialization from the outside doesn't work anymore. Then, I > get a unregistered class exception. It only works when I put the code to > execute the serialization into the cpp file where the classes are registered > with BOOST_CLASS_EXPORT_GUID. > > Again, I attached the example source code. The library is created with > cmake. Therefore, you need to have cmake installed on your machine to test > the example. The commands to build the example are cd > /path/to/unpacked/testfolder mkdir build cd build cmake .. > make > > Funnily enough, the serialization from the outside (from the main function) > works if the serialization in the cpp file is executed in advance. > > The example works if the library is a dynamic library. If you wanna see > this, you need to replace STATIC in CMakeLists.txt with SHARED. > > David > > >

1 0

[exception] Conditional error_info addition
by Adam Badura 17 Feb '10

17 Feb '10

Is there a way to add error_info to an exception object conditionally? In my use case I throw upon error during Windows Registry manipulation. One of the data attached to the exception object is the path to the registry key on which operation failed. The problem however is that sometimes (while the library is still developed...) the failure is due to internal code error: the handle to the key is no longer valid because it was closed (erroneously) previously. In such case obtaining path to the key will yield an empty string. I don't want to attach an empty string. I would prefer not to attach anything. (Now actually in this particular case it might be even useful to attach an empty string which is then a clear indication that the key handle was invalid. However I hope that my point is clear here.) Currently I would have to do something like: ExceptionType e; std::wstring path = getPath( hKey ); if ( !path.empty() ) e << PathInfo( path ); BOOST_THROW_EXCEPTION( e ); This code however involves more copying with little hopes for optimization. And it is obviously long. If this was to be done for two or three data items the throwing code would become quite long... Can anything be done about it? Can we have some form of conditional error_info insertion? Adam Badura

2 4

[Serialization] unable to fix UNREGISTERED CLASS exception
by David Graf 17 Feb '10

17 Feb '10

Hello Everybody I am using boost serialization heavily in my project to serialize C++ objects. Unfortunately, I have a big problem when I when polymorphic pointers and non intrusive serialize functions are involved in the serialization. Therefore, I made a little project which demonstrates my usage of boost functionality (see attachment). Basically, I have four classes which are derived in the following way: Item <- AtomicItem <- BooleanItem <- SpecialItem. SpecialItem contains an intrusive serialized functions. The other three contain non intrusive serialize functions. In the main function (testing.cpp), I create a SpecialItem and the stack and assign it to a pointer to an Item. Then, I serialize that Item. Unfortunately, it doesn't work. Only the serialize function of the class Item is invoked. Can you please tell me what I did wrong in my example? Note: The example can be compiled with the command: g++ atomic_items_serialization.cpp testing.cpp -o testing -lboost_serialization-mt And the expected result is: serializing Item serializing AtomicItem serializing BooleanItem serializing SpecialItem Thanks in advance David

3 3

Property tree custom_data_type.cpp fails to compile in boost 1.42.
by Mauricio Gomes 17 Feb '10

17 Feb '10

The example C:\boost_1_42_0\libs\property_tree\examples\custom_data_type.cpp fails to compile for me. I am using Code::Blocks 8.02 with mingw gcc 4.4.0. I am getting the following error message: ||=== test, Debug ===| C:\projects\test\main2.cpp||In function 'int main()':| C:\projects\test\main2.cpp|59|error: wrong number of template arguments (5, should be 3)| C:\boost_1_42_0\boost\property_tree\ptree_fwd.hpp|29|error: provided for 'template<class Key, class Data, class KeyCompare> class boost::property_tree::basic_ptree'| C:\projects\test\main2.cpp|59|error: invalid type in declaration before ';' token| C:\projects\test\main2.cpp|63|error: request for member 'put' in 'pt', which is of non-class type 'my_ptree'| C:\projects\test\main2.cpp|64|error: request for member 'get' in 'pt', which is of non-class type 'my_ptree'| C:\projects\test\main2.cpp|64|error: expected primary-expression before 'int'| C:\projects\test\main2.cpp|64|error: expected ',' or ';' before 'int'| C:\projects\test\main2.cpp|68|error: request for member 'put' in 'pt', which is of non-class type 'my_ptree'| C:\projects\test\main2.cpp|68|error: expected primary-expression before '>' token| C:\projects\test\main2.cpp|68|warning: left-hand operand of comma has no effect| C:\projects\test\main2.cpp|69|error: request for member 'get' in 'pt', which is of non-class type 'my_ptree'| C:\projects\test\main2.cpp|69|error: expected primary-expression before '>' token| C:\projects\test\main2.cpp|74|error: request for member 'put' in 'pt', which is of non-class type 'my_ptree'| C:\projects\test\main2.cpp|74|error: expected primary-expression before '>' token| C:\projects\test\main2.cpp|74|warning: left-hand operand of comma has no effect| C:\projects\test\main2.cpp|75|error: request for member 'get' in 'pt', which is of non-class type 'my_ptree'| C:\projects\test\main2.cpp|75|error: expected primary-expression before '>' token| ||=== Build finished: 15 errors, 2 warnings ===| The problem seems to be the number of template parameters in line 59: // Property_tree with boost::any as data type // Key comparison: std::less<std::string> // Key type: std::string // Path type: path // Data type: boost::any // Translator type: my_translator typedef basic_ptree<std::less<std::string>, std::string, path, boost::any, my_translator> my_ptree; my_ptree pt; Am I missing something ? Thanks for your attention, -- Mauricio Gomes Pensar Digital 55-11-9698-1683 (Brazil - mobile)

2 1

[Boost.Build]: Suppress compiler version in build path
by Roland Bock 17 Feb '10

17 Feb '10

Hi, using the current version of boost.build, compilation results are generated in paths like this: /example/gcc-4.2.4/debug/threading-multi Earlier versions (e.g. at a customer's site) ignore the compiler version, thus /example/gcc/debug/threading-multi Can I tell the current bjam to omit the compiler version in the path? Thanks and regards, Roland

1 0

info about connect method of boost::asio::ip::tcp::iostream
by Conoscenza Silente 17 Feb '10

17 Feb '10

Hi All Right now I started using boost::asio::ip::tcp::iostream class. If I use the constructor of this class with 2 strings stating IP and PORT everything is fine and I can connect to my server. boost::asio::ip::tcp::iostream sockStream("1.2.3.4","5000"); Instead if I use the connect method using the port as int, connection doesn't happen. boost::asio::ip::tcp::iostream sockStream; sockStream.connect ("1.2.3.4", 5000 ); if( sockStream.fail() ){ std::cout << "I am not connected" << std::endl; } What am I doing wrong? It is something allowed to do or should I use other classes like resolver::query + resolver::iterator? Thanks CS

2 2

Re: [Boost-users] [program_options][1.35] validating string value
by michele.caini＠gmail.com 17 Feb '10

17 Feb '10

If you have problem using boost.regex or another regex framework, don't worry about ask for help, please. :) ----- Reply message ----- Da: "Soumen" <soumen78(a)gmail.com> Data: mer, feb 17, 2010 12:21 Oggetto: [Boost-users] [program_options][1.35] validating string value A: <boost-users(a)lists.boost.org> Thanks Andre and Michele - got the way to accomplish it. -- View this message in context: http://old.nabble.com/-program_options--1.35--validating-string-value-tp276… Sent from the Boost - Users mailing list archive at Nabble.com. _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

1 0

Re: [Boost-users] [program_options][1.35] validating string value
by michele.caini＠gmail.com 17 Feb '10

17 Feb '10

Using regex can help you, maybe? ----- Reply message ----- Da: "Soumen" <soumen78(a)gmail.com> Data: mer, feb 17, 2010 11:40 Oggetto: [Boost-users] [program_options][1.35] validating string value A: <boost-users(a)lists.boost.org> Hi, My option (say, "opt1") can take any of only few possible string type values (say, "value1", "value2", "value3" and "value4"). Is there anyway to specify this such that if any other value, say - "valueXX", is passed to "opt1", it throws an error? If yes, can someone point me to documentation or sample code snippet to show how to do this? Or, explicitly, checking against these possible values is my only option? Regards, ~ Soumen -- View this message in context: http://old.nabble.com/-program_options--1.35--validating-string-value-tp276… Sent from the Boost - Users mailing list archive at Nabble.com. _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

3 3

Re: [Boost-users] [program_options][1.35] validating string value
by michele.caini＠gmail.com 17 Feb '10

17 Feb '10

Something like [on¦off¦error] is enough, for that simple case. ;) ----- Reply message ----- Da: "Soumen" <soumen78(a)gmail.com> Data: mer, feb 17, 2010 11:58 Oggetto: [Boost-users] [program_options][1.35] validating string value A: <boost-users(a)lists.boost.org> Using regex can help you, maybe? Say, my switch name is "semantic_check" and possible values are "on" (turns on the checker, issues warning when something fails), "off" and "error" (changes severity of warning to error). Could you cite an example code snippet me how regex may help? Regards, ~ Soumen -- View this message in context: http://old.nabble.com/-program_options--1.35--validating-string-value-tp276… Sent from the Boost - Users mailing list archive at Nabble.com. _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

1 0

[1.35.0][OSX][iostreams] Exception thrown with utf8 filename
by Christophe Radondy 17 Feb '10

17 Feb '10

Hi, I try to do a cross-platform source code and I have this code. I compiled boost 1.35 in universal binaries on Mac OS 10.5.4 with intel platform (IMAC core 2 duo 3Ghz) The same code works on linux (Ubuntu 8.0.4 and debian) and windows platforms but on Mac OS 10.5 I have an exception in boost's libraries. We compile on universal binaries with -fexceptions options of g++ 4.0.1 (Apple). This is my code: <code> boost::filesystem::wpath unicode(L"abcdef\u0110"); boost::filesystem::ofstream file(unicode.string()); // throws Invalid Argument on Mac file << "abcdef" << endl; file.close(); <endcode/> If I do this, it works: <code> boost::filesystem::path utf8("abcdef\xC4\x90"); // \xC4\x90 == \u0110 boost::filesystem::ofstream file(utf8.string()); // works ok file << "abcdef" << endl; file.close(); <endcode/> When the ofstream is created, this function is executed and crashes when I use Unicode parameters : <code> unicode.external_file_string(); // throws uncaught exception of type boost::filesystem::basic_filesystem_error<boost::filesystem::basic_path<std::basic_string<wchar_t, std::char_traits<wchar_t>, std::allocator<wchar_t> >, boost::filesystem::wpath_traits> > - boost::filesystem::wpath::to_external conversion error: Invalid argument <endcode/> Do you have any idea? Is this a known bug? Did I miss a setting? Thanks for your help. Regards, Christophe.

2 2

error importing hello_ext
by hitesh dhiman 17 Feb '10

17 Feb '10

Hi all, I'm trying to get the boost.python tutorial working. The hello.py example builds successfully. The problem is, when i run my python file, there is an error: Evaluating hello.py Traceback (most recent call last): File "<string>", line 6, in <fragment> ImportError: No module named hello_ext Now, the tutorial directory has the python and c++ file, and a bin directory. the bin directory is as follows : tutorial\bin\msvc-9.0\debug\threading-multi the hello_ext.pyd file is located inside here. Can anyone help? Its a path problem i believe, but i'm not able to pin point whats wrong. -- Regards, Hitesh Dhiman Electrical Engineering National University of Singapore

3 4

Object pools?
by Ryan McConnehey 17 Feb '10

17 Feb '10

I recently had to create a class that managed a pool of objects for the user. Instead of the user creating the object themselves they would ask the class for an object. The class would hand the user a boost::shared_ptr of the object. When the boost::shared_ptr needed to destroy the object it handed it back to the class to reuse. The class is designed for those areas that need a lot of the same object but not all at the same time. This class has been useful for the boost::asio library where the async_receive and async_write need the buffer to be valid until the callback occurs. Other people at my work have shown interest and I'm looking at needing to improve the class to make it more general. Instead of spending time doing the improvements I would like to know if the boost libraries have anything like what I described. I'll looked through the descriptions but haven't seen anything that comes close. Ryan

3 4

Embedding boost in a given project
by joel falcou 17 Feb '10

17 Feb '10

In order to satisfy various users request and to limit potential whining about implicit dependencies on boost, I've started lookign at how to embed boost components into my own tools. What are the usual way to do that ? I pondered working with a given version (say 1.42) in which I could add w/e patches and fixed from the trunk along with my own custom patches. I then use BCP to extract sub boost compoennt to my release package. Now, the final funny thing is that, for various reason, we don't use bjam atm for our own product build. I guess adding a makefile fuel like dependency: bjam --with-xxxx is enough for example ? Any experience to be shared on that topic ? -- ___________________________________________ Joel Falcou - Assistant Professor PARALL Team - LRI - Universite Paris Sud XI Tel : (+33)1 69 15 66 35

6 6

Plugin library boost
by Lloyd 17 Feb '10

17 Feb '10

Hi, Does boost have a plug-in framework to develop a plug-in in applications. If not, does anybody know a good plug-in framework. Thank you very much, Lloyd ______________________________________ Scanned and protected by Email scanner

2 2

random: uniform_01 usage and mixing boost::random with boost::math::normal
by Hicham Mouline 16 Feb '10

16 Feb '10

Hello, I have this code: boost::mt19937 mers; boost::uniform_01<boost::mt19937> uf01(mers); boost::exponential_distribution<> ed(0.1); I call the operator() on uf01 3 million times, as uf01() I don't quite understand the warning in the documentation about the constructor taking the generator by value. This means that my mers doesn't change between calls to uf01(). Is this bad? Will it affect the quality of the randomness of the uniform numbers generated? I also have boost::math::normal nrm; and I use boost::math::cdf(nrm, …) and boost::math::quantile( nrm, … ); Are there any issues with using both boost∷math∷normal and other distributions from the boost∷random library? Regards,

2 1

[ASIO] When the work object is destroyed?
by Daniele Barzotti 16 Feb '10

16 Feb '10

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi, in my library I create a work object on the heap to maintain the io_service running and I assign it to a object member shared_ptr so, when the last object is destroyed, the work ends and the io_service.run() returns. All works but, if when I create an object after the thread is ended, the io_service returns immediately.. ..and I don't know why, any suggestions? Thanks! Daniele. - ------------------------------------------------------------------- // Main io_service static boost::asio::io_service io_service; static boost::scoped_ptr<boost::thread> io_service_thread; static boost::asio::io_service::work* p_work; static BOOL thread_started; void io_worker_thread(void) { #if(WIN32 && _WIN32_DCOM) struct com_init { com_init() { CoInitializeEx(NULL, COINIT_APARTMENTTHREADED); } ~com_init() { CoUninitialize(); } } initializer_object; #endif io_service.run(); thread_started = false; }; EXTERN_C EUROATLIB_API Hndl CALL GetTETRAMotoObj() { if (!thread_started) { try { // create the work object on the heap p_work = new boost::asio::io_service::work(io_service); // run the IO service as a separate thread: // Instead of start the worker thread directly by passing the // io_service object, // I call the io_worker_thread function where I can do some // initialization tasks like calling CoInitializeEx for COM io_service_thread.reset(new boost::thread(io_worker_thread)); thread_started = !thread_started; } catch (boost::thread_resource_error e) { // Failed to create the new thread return 0; } } // create the new object TETRAMoto* pTETRAMoto = new TETRAMoto(io_service); // Assign a reference to io_service work object: // This reference is used to counting the objects created. // When the last TETRAMoto object is deleted the main work object // is destroyed and the io_service thread ends. pTETRAMoto->_io_work_ptr.reset(p_work); // return the object return pTETRAMoto; }; - ------------------------------------------------------------------- -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (MingW32) iEYEARECAAYFAkt6f04ACgkQ/l+kMioSZwjYiwCgleSrLLiKKQlmZjN7HQBSEs75 olUAoKhMEV7/SK6tiI4hIxUoRjQv9qEE =KGSy -----END PGP SIGNATURE-----

2 2

[Iostreams] stream<file_sink> does not throw an exception and not set fail/bad bits when file is wrong
by Mariusz Wojtysiak 16 Feb '10

16 Feb '10

Hello! I'm trying to write some content to a new file using boost::stream<file_sink> class. Unfortunately errors/failures during opening the file are ignored by boost library. Here is a snippet: using namespace boost::iostreams; stream<file_sink> stream( "/wrong.txt" ); std::cout << stream.fail() << ' ' << stream.bad() << std::endl; stream.exceptions( std::ios::badbit | std::ios::failbit ); stream << "test" << std::endl; stream.close(); Program is started as not root user and have not access for file / wrong.txt. It should throw an exception, but instead of this: - console prints: 0 0 - none exception is thrown - file /wrong.txt is not created Similar piece of code using std::ofstream works as I expect: using namespace std; ofstream stream( "/wrong.txt" ); std::cout << stream.fail() << ' ' << stream.bad() << std::endl; stream.exceptions( std::ios::badbit | std::ios::failbit ); stream << "test" << std::endl; stream.close(); - console prints: 1 0 - exception is thrown I cannot believe that boost::iostreams has such obvious bug and just ignore failures! I have to do something wrong. But what? Regards, Mariusz Wojtysiak

2 2

how to realize that a connection is closed by a client
by Conoscenza Silente 16 Feb '10

16 Feb '10

Hi All I am doing first steps with boost::asio and I want to use ip::tcp::stream to send data from a server to a client. Reading the sample in boost docs I did a first sample of server code. Is there any way for the server to check if a client connection is closed or gone at some point? for( ;; ){ boost::asio::ip::tcp::iostream stream; acceptor.accept( *stream.rdbuf() ); for( int i = 0; i < 10; i++ ){ stream << "message" << i << std::endl; // how to realize that the client has gone here? boost::this_thread::sleep( boost::posix_time::milliseconds(1000) ); }

2 2

Problems with boost build 1.42
by Avi Bahra 16 Feb '10

16 Feb '10

I am using bjam to build my application on suse linux. I have defined a environment variable BOOST_ROOT, which allows me to switch between boost builds. I have used this approach for 1.38, 1.39,1.40. I always build the boost libraries I need myself using bjam. bjam --build-dir=./tmpBuildDir toolset=gcc stage link=static --with-system i.e I don't bother with build scripts in $BOOST_ROOT With 1.42 when I build my application I get an error of the form: error: Unable to find file or target named error: '/var/tmp/ma0/boost/boost_1_42_0/stage/lib/libboost_system.a' error: referred from project at error: '' The error is very misleading because the file exists? When I switch back, to 1.39/1.40 there are no problems. This is what my boost-build.jam look like: -------------------------------------------------------------------------------- BOOST_BUILD = [ MATCH --boost-build=(.*) : $(ARGV) ] ; BOOST_BUILD ?= $(BOOST_ROOT)/tools/build/v2 ; boost-build $(BOOST_BUILD) ; Here is jam file for my leaf project: ---------------------------------------------------------------------------- use-project theCore : ../ACore ; exe test : [ glob src/*.cpp ] /site-config//boost_system /site-config//boost_serialization : <include>../SCRATCH/src <variant>debug:<define>DEBUG ; -------------------------------------------------------------------------- Notice that I reference /site-config// to reference the boost libs. for some reason in 1.42 this fails ! My site-config.file which is located in $BOOST_ROOT/tools/build/v2 is of the form: ------------------------------------------------------------------------ import os ; local BOOST_ROOT = [ os.environ BOOST_ROOT ] ; lib boost_serialization : : <file>$(BOOST_ROOT)/stage/lib/libboost_serialization.a ; lib boost_system : : <file>$(BOOST_ROOT)/stage/lib/libboost_system.a ; project site-config : requirements <include>$(BOOST_ROOT) ; using gcc ; using testing ; -------------------------------------------------------------------- My site-config.jam is the same between 1.40 and 1.42. There is probably a silly mistake on my part somewhere, but its preventing me from moving to 1.42. Any help appreciated. Best regards, Ta, Avi

1 0

Server / client Boost.Test
by Christian Auby 16 Feb '10

16 Feb '10

How would I go about writing a unit test that runs first a network server, then connects to it and does some operations with it? I can see issues with the run order, and if there's a problem in the server it might never respond and the test would run forever. regards, Christian

1 0

boost.asio Bus error on g++ 4.5 / Mac OS X 10.5.6
by Orca X 16 Feb '10

16 Feb '10

Hi! For some reason very simple program crashes with Bus error. code is --------------------------- #include <iostream> #include <boost/asio.hpp> int main(int argc, char **argv) { std::cout << "xxx" << std::endl; return 0; } --------------------------- Crash happens before any output is made. I'm using boost.build to build executables. The bug dissapears when building variant=debug. Gdb output: --------------------------- Program received signal EXC_BAD_ACCESS, Could not access memory. Reason: KERN_PROTECTION_FAILURE at address: 0x00000000 0x000019a8 in ?? () (gdb) bt #0 0x000019a8 in ?? () #1 0x00001e90 in boost::asio::detail::service_id<boost::asio::detail::reactive_descriptor_service<boost::asio::detail::kqueue_reactor<false> > >::~service_id () Previous frame inner to this frame (gdb could not unwind past this frame) --------------------------- Situation is that I'm building with g++4.5 installed from macports, with --std=c++0x, but boost itself was built with regular gcc 4.0.1. 4.5 version located at /opt/local/* , boost located at /usr/local/*. How could it happen that asio library destroys something even before my code is reached? No library initialization involved. Thanks in advance!

1 0

Boost asio serial and rs-485
by Christian Auby 15 Feb '10

15 Feb '10

Is it possible to use Boost asio serial with rs485? If yes, is there anything special you need to be aware of? regards, Christian

2 1

Trouble using bjam
by Roland Bock 15 Feb '10

15 Feb '10

Hi, I have a little trouble using bjam: I'd like to tell it to add /my/path to the include paths, by using this command-line: bjam include=/my/path I took this syntax from here: http://www.boost.org/boost-build2/doc/html/bbv2/overview/invocation.html#bb… But bjam answers with /usr/share/boost-build/build/property.jam:258: in validate1 from module property error: Invalid property '<include>': No value specified for feature 'include'. /usr/share/boost-build/build/property.jam:266: in property.validate from module property /usr/share/boost-build/build/build-request.jam:185: in convert-command-line-element from module build-request /usr/share/boost-build/build/build-request.jam:143: in build-request.from-command-line from module build-request /usr/share/boost-build/build-system.jam:164: in load from module build-system /usr/share/boost-build/kernel/modules.jam:261: in import from module modules /usr/share/boost-build/kernel/bootstrap.jam:132: in boost-build from module /usr/share/boost-build/boost-build.jam:1: in module scope from module Any idea? I am using bjam installed by Ubuntu-8.04: Boost.Build V2 (Milestone 11) Boost.Jam 03.1.14 Regards, Roland

3 9

[Boost.Asio] Compiling Boost.Asio with OpenMPI
by vibhor aggarwal 15 Feb '10

15 Feb '10

Hello, I am trying to write a program which runs using OpenMPI on our cluster and communicates the results using TCP/IP to another machine which has a network connection to the cluster. I am using Boost.Asio for the TCP/IP connection. If I remove the MPI code from the executable and compile it with g++, it connects to the remote machine without a problem. But when I add a simple MPI code to it and compile with mpicxx and execute using mpirun, it generates a connection time out in the call to socket.connect(). Any idea what might be going wrong ? Thanks. Your Mail works best with the New Yahoo Optimized IE8. Get it NOW! http://downloads.yahoo.com/in/internetexplorer/

3 6

[Serialization] crash in boost::serialization::extended_type_info::key_unregister ()
by Avi Bahra 15 Feb '10

15 Feb '10

> Hi, > I had the same problem... > > This problem was fixed in the next release (1.40) > Regards, > Damien. Ok I tried this and found: a/ whereas 1.39 serialisation compiles with out _any_ warnings. 1.40 has a _blizzard_ sorry I meant avalanche of warnings on building boost and on my application of the kind: /var/tmp/ma0/boost/boost_1_40_0/boost/archive/basic_binary_oarchive.hpp:88: warning: comparison between signed and unsigned integer expressions This makes it extremely difficult to see my own warnings. Is 1.40 really production quality ? We have a policy of no warnings, so the obvious question now, is what release are these warning's fixed ? b/ Still get the same crash on program exit. Program received signal SIGSEGV, Segmentation fault. 0xb7ce3a08 in strcmp () from /lib/libc.so.6 (gdb) bt #0 0xb7ce3a08 in strcmp () from /lib/libc.so.6 #1 0x083c814d in boost::serialization::extended_type_info::key_unregister () #2 0x083c8407 in boost::serialization::extended_type_info::~extended_type_info () #3 0x081bf8cc in ~extended_type_info_typeid (this=0x8789604) at /var/tmp/ma0/boost/boost_1_40_0/boost/serialization/extended_type_info_typeid.hpp:80 #4 0x081bf98e in ~singleton_wrapper (this=0x8789604) at /var/tmp/ma0/boost/boost_1_40_0/boost/serialization/singleton.hpp:104 #5 0x0819f508 in __tcf_25 () at /var/tmp/ma0/boost/boost_1_40_0/boost/serialization/singleton.hpp:121 #6 0xb7ca3599 in exit () from /lib/libc.so.6 #7 0xb7c8cfe8 in __libc_start_main () from /lib/libc.so.6 #8 0x08053921 in _start () Oh well! I will have to go back placing all the serialisation in one translation unit and stick with 1.39. Best regards, Ta, Avi

2 1