December 2010 - Boost-users - lists.preview.boost.org

serialization: "derived class not registered or exported" exception , 1.45 boost
by Hicham Mouline 19 Dec '10

19 Dec '10

Hello, I have moved all serialization code in a separate static library that I called serialization. This, in my case, depends on mpi. The headers in this static library: 1. include <boost/serialization/export.hpp> 2. include the definition of the derived class being serialized 3. declare a template function serialize() that takes a reference to the derived class being serialized in the boost::serialization namespace 4. invoke the BOOST_CLASS_EXPORT_KEY(<derivedclass>) macro at the global namespace scope The corresponding source files in this static library: 1. include the header 2. define the function template serialize() 3. explicitly instantiate the serialize() function with all the archives used: mpi::packed_iarchive and mpi::packed_oarchive in the boost::serialization namespace 4. invoke the BOOST_CLASS_EXPORT_IMPLEMENT(<derivedclass>) macro at the global namespace scope The main program links against this static library, and invokes a mpi function (broadcast) that serializes by the base pointer. The base class is polymorphic. All class derivations are public. Debugging, the code reaches : archive/detail/oserializer.hpp boost::archive::detail::save_pointer_type<boost::mpi::packed_oarchive>::poly morphic::save<baseclass>(boost::mpi::packed_oarchive & ar={...}, baseclass& t={...}) Line 400 + 0x38 bytes where NULL==true_type However, looking at the t argument of save(), it shows that is the correct derived object. But the type T is the base class. What am I missing? regards,

2 2

1.45 download sourceforge corrupt
by Hicham Mouline 19 Dec '10

19 Dec '10

hello. has anyone downloaded 1.45 archives from sourceforge recently. I've downloaded the 4 formats and all of them appear corrupt to 7zip and winrar on windows I've tried 2 different mirrors rds,

5 6

mpi: what is the archive used when serializing the objects send by mpi functions?
by Hicham Mouline 19 Dec '10

19 Dec '10

Hello, I may need to instantiate all the serialize() template functions in 1 library in my project. What is the archive type(s) that mpi uses so that I can explicitly instantiate that template function? regards,

2 1

RAD Studio XE and boost library
by Christer Borgqvist 19 Dec '10

19 Dec '10

Hi, Date and Time Formatted Input and Output Try to use a sample like this #include <boost/date_time/gregorian/gregorian.hpp> #include <iostream> #include <locale> int main() { boost::gregorian::date d(2009, 1, 7); boost::gregorian::date_facet *df = new boost::gregorian::date_facet("%A, %d %B %Y"); std::cout.imbue(std::locale(std::cout.getloc(), df)); std::cout << d << std::endl; } When compile I get [BCC32 Error] Unit1.cpp(86): E2316 'date_facet' is not a member of 'boost::gregorian' Appreciate any help. //Christer

2 2

Re: [Boost-users] [Boost.Test] does not always include checkpoint information
by Peter Soetens 19 Dec '10

19 Dec '10

On Thu, Dec 2, 2010 at 8:34 PM, Gennadiy Rozental <rogeeff(a)gmail.com> wrote: > Peter Soetens <peter.soetens <at> gmail.com> writes: > >> >> I'm using boost 1.40. When parsing the output of a Boost.Test xml file, I >> sometimes get this error: >> >> "test-result.xml:cvc-complex-type.2.4.b: The content of element 'Exception' is >> not complete. One of '{LastCheckpoint}' is expected." (output from Hudson Test >> plugin). > > Can't really comment on this output. This is something generated by your tool. > Can you should the Boost.Test xml? The output is: ./testboost --log_level=all --report_level=detailed --output_format=xml <TestLog><TestSuite name="Master Test Suite"><TestSuite name="TestSuite"><TestCase name="testThrow"><Exception>unknown type</Exception><TestingTime>0</TestingTime></TestCase></TestSuite></TestSuite></TestLog><TestResult><TestSuite name="Master Test Suite" result="failed" assertions_passed="0" assertions_failed="1" expected_failures="0" test_cases_passed="0" test_cases_failed="1" test_cases_skipped="0" test_cases_aborted="1"><TestSuite name="TestSuite" result="failed" assertions_passed="0" assertions_failed="1" expected_failures="0" test_cases_passed="0" test_cases_failed="1" test_cases_skipped="0" test_cases_aborted="1"><TestCase name="testThrow" result="aborted" assertions_passed="0" assertions_failed="1" expected_failures="0"></TestCase></TestSuite></TestSuite></TestResult> The code is : struct A {}; BOOST_FIXTURE_TEST_SUITE( TestSuite, A ) BOOST_AUTO_TEST_CASE(testThrow) { throw A(); } BOOST_AUTO_TEST_SUITE_END() > >> It appears that It does not consider the entry point of the unit test function >> as a checkpoint, while the documentation says it does. > > I do not believe so. Can you point me to the relevant page? >From : http://www.boost.org/doc/libs/1_45_0/libs/test/doc/html/utf/user- guide/test-output/test-log.html "In most cases The UTF can't provide an exact location, where system error occurs or uncaught C++ exception is thrown from. To be able to pinpoint it as close as possible the UTF keeps track of checkpoints - the location a test module passed through. ***A test case entrance and exit points, a test tool invocation point the UTF tracks automatically***. Any other checkpoints should be entered by you manually. The test log provides two macros for this purpose: BOOST_TEST_CHECKPOINT - to specify a "named" checkpoint and BOOST_TEST_PASSPOINT - to specify an "unnamed" checkpoint." *** emphasis mine. So I would expect that it appears as a checkpoint in the XML output, with the line number pointing at declaration of the test name. > >> I can 'fix' this by >> inserting check/pass points at the beginning of every function, but shouldn't >> Boost.Test do this automatically ? > > Test log output reports test case enter/exit events, thus you should be able to > figure this out already. I don't get any line number when it's not explicitly present as a checkpoint xml tag for example. The xml data is read by an external tool, that then tries to relate the errors to files and linenumbers. As you can see from the XML output above. No filename nor linenumber is listed, although something definately went wrong. A 'LastCheckPoint' tag would provide that information. Could I add a checkpoint in all my fixtures in order to 'inject' an initial checkpoint in each test ? Peter

2 2

[MPL] Applying a boost::mpl::list to the template parameter of a type.
by Andrew Hundt 18 Dec '10

18 Dec '10

Hello, I have a class that requires a boost::variant containing shared pointers to various types as follows: template <typename ToySharedPtrVariant, typename ColorSharedPtrVariant> class ToyPicker { typedef std::pair< ToySharedPtrVariant, ColorSharedPtrVariant > toyAndColorPair; typedef std::map< std::string, std::vector< toyAndColoPair > > stringToToyColorPairMap; // ... methods that use the defined types... } This class currently requires template parameters of the following form to compile: ToyPicker< boost::variant< boost::shared_ptr<ToyModel> >, boost::variant< boost::shared_ptr<BlueToy>, boost::shared_ptr<RedToy>, boost::shared_ptr<GreenToy> > > toyPicker; How do I use an mpl list so that I can allow the following much simpler definition for users, then convert it into the one above inside my class? ToyPicker< boost::mpl::list< ToyModel >, boost::mpl::list< BlueToy, RedToy, GreenToy > > toyPicker; I really appreciate all of the user help you all give, and hope that as I become more experienced I may be able to contribute myself. Cheers! Andrew Hundt

2 1

boost::date_time: date & time in RFC 3339 format
by Stuart 17 Dec '10

17 Dec '10

Hello I'm trying to use boost::date_time to format the date and time (with fractional seconds) in RFC3339 format http://tools.ietf.org/html/rfc3339. I'd like my date to look like "1996-12-19T16:39:57.76-08:00". date +'%FT%T.%N%:z' gives me the format that I want 2010-12-17T17:09:41.079937078+01:00 The boost/C++ code at the end of this message gives me 2010-12-17T16:10:03.141889Z this is the correct time but I want the time in my time zone with an indication of the time zone (+0100) not in UTC with a "Z". Can anyone give me a hint? Another question, is there anyway to control the number of decimals after the point for the seconds? thanks Stuart #include <boost/date_time.hpp> std::cout.imbue(std::locale(std::cout.getloc(), new boost::local_time::local_time_facet("%Y-%m-%dT%H:%M:%S%F%Q"))); std::cout << boost::local_time::local_microsec_clock::local_time(boost::local_time::time_zone_ptr()) << std::endl;

1 0

Simple boost::python::dict question (I hope)
by Tim Aerts 17 Dec '10

17 Dec '10

Hello, I have a pretty urgent matter. Is it possible to add an element to a Python's dict using boost::python? I tried: extrafields["validated"] = "False"; With 'extrafields' in this case being a boost::python::dict But that crashes. I hope this will be an easy one... :) thx Tim

2 1

[Regex] Can't create dylib that links regex statically
by Marcos Pearson 17 Dec '10

17 Dec '10

Hi there, I'm hoping someone can give me some hints on what I'm doing wrong. I'm trying to create a library that uses regex and has no external dependencies. My understanding is that I need to compile boost to be static and link static at runtime. My project is multi-platform, but I'm currently working on OSX 10.6.5. To build boost, I'm using the following command: sudo ./bjam link=static threading=multi runtime-link=static architecture=combined address-model=32_64 macosx-version=10.5 macosx-version-min=10.5 install In Xcode, I've included the produced library, libboost_regex.a, in my Xcode project. I compile my dylib without errors or warning. However, I'm unable to load it. Checking the dependencies, my library seems to want the libboost_regex.dylib. Why? $ otool -L libglobalkvp.dylib libglobalkvp.dylib: /usr/local/lib/libglobalkvp.dylib (compatibility version 1.0.0, current version 1.0.0) /Users/marcospearson/Library/Frameworks/iqvectorsdk.framework/Versions/A/iqvectorsdk (compatibility version 1.0.0, current version 1.0.0) libboost_regex.dylib (compatibility version 0.0.0, current version 0.0.0) /usr/lib/libstdc++.6.dylib (compatibility version 7.0.0, current version 7.4.0) /usr/lib/libgcc_s.1.dylib (compatibility version 1.0.0, current version 1.0.0) /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 111.1.4) I've built, and re-build the boost libraries more times that I can remember and have read post after post, but haven't been able to figure out a solution. Any help would be appreciated. Thanks, Marcos

3 2

serialization/fusion: non intrusive serialize() function templated on type being serialized
by Hicham Mouline 17 Dec '10

17 Dec '10

Hello, I have a base class and a large number of derived classes that are all adapted as fusion sequences. The intention is to serialize through the base pointer. struct Base; struct D_1 : public Base;struct D_2 : public Base;... All the D_i are adapted as fusion sequences. Besides, there is an mpl vector of all the D_i called DVECTOR. In a normal setting, I would have written this: template <typename Archive>void serialize(Archive& ar, Base& b, unsigned int ){ ar & b.member1;} template <typename Archive>void serialize(Archive& ar, D1& d, unsigned int ){ ar & boost::serialization::base_object<Base>(d); ar & d.memberd1;}BOOST_CLASS_EXPORT_KEY(D1) and the same for all D_i. Then I would need BOOST_CLASS_EXPORT_IMPLEMENT for all the D_i as well in the relevant translation units. The base_object part of serialize() is the same for all D_i The D_i specific part of serialize(), I have the ability to generate with fusion meta functions. This leads to a template function with 2 template arguments: template <typename Archive, template Derived>void serialize(Archive& ar, Derived& d, unsigned int ){ ar & boost::serialization::base_object<Base>(d); // fusion code to work on the Derived fusion sequence to generate a " ar & ... " line for each member of Derived} Robert, My question is about the BOOST_CLASS_EXPORT_KEY() and BOOST_CLASS_EXPORT_IMPLEMENT() macros. Am I able to generate all of them from DVECTOR? regards,

1 0