- Boost - lists.boost.org

serialisation - prevent restoring archive of class X into class Y
by Oliver.Kowalke＠infineon.com 10 Feb '06

10 Feb '06

Hello, is it possible to prevent that the archive of class X can be restored in class Y (both have the same layout)? Reagards, Oliver

3 2

[thread] Patch: suppress report of tss_hook "leak"
by Kevin Bluck 10 Feb '06

10 Feb '06

Recently I took up with the thread library. Using MSVC, I quickly ran afoul of a memory leak report from the CRT debugger. Naturally assuming that I am more likely to screw up than a well-reviewed library, I spent a not insignificant amount of effort trying to track down my own stupid mistake. Eventually realizing the cause was in boost.thread, I spent more time researching and finally discovering this thread: http://lists.boost.org/boost-users/2005/08/13538.php Apparently, this issue is well understood and the consensus is that it is benign and the CRT is just being overly pedantic. Although I understand that the issue is generally not regarded as a bug or even a "memory leak" as such, I can't help but imagine that other library users are in the same boat as me. I decided it would be nice if MSVC didn't unnecessarily alarm ignorant newcomers like myself and continually annoy others who understand the issue but have a personal conviction against allowing lint in their builds. So, if there is no plan to change the underlying issue, I figured I could at least shut up MSVC. Attached is a patch for tss_hook.cpp, which uses the function _CrtSetDbgFlag to temporarily turn off heap allocation tracking for the particular allocation in question. It should have no effect except on MSVC debug builds. On my configuration (msvc-8_0) it successfully eliminates the leak warnings. I figure it's the moral equivalent of suppressing a warning that can't reasonably be eliminated otherwise. I hope it seems useful. --- Kevin

1 0

Help with documentation
by John Phillips 10 Feb '06

10 Feb '06

If you have a great idea for a library, but you don't feel the documentation is keeping to the same high standards, there is now a place to look. If you are frustrated by the documentation of a library and would love a chance to help make it better, there is now a place to offer your services. On the Wiki, you will now find a link to Documentation Helpers. This is a list of people who are volunteering to work with developers to improve the documentation of libraries and proposed libraries. If you want to help, go add your name to the list. If you want help, go look at the list and contact someone. Since this is the result of a conversation buried deep in a thread that some people may not be following, I'm also putting it in its own thread. John Phillips

5 7

[typeof][spirit][msvc gurus] is it possible to work around this bug??
by Tobias Schwinger 10 Feb '06

10 Feb '06

Hi List, The code below fails to compile with MSVC8 when the macro MAKE_THINGS_FAIL is defined. I hope the code reads well enough to communicate the problem and why it would be quite hot to get it fixed ;-). Thanks a lot for your help, Tobias #include <boost/spirit/core.hpp> #include <boost/typeof/typeof.hpp> #include BOOST_TYPEOF_INCREMENT_REGISTRATION_GROUP() BOOST_TYPEOF_REGISTER_TYPE(boost::spirit::anychar_parser) BOOST_TYPEOF_REGISTER_TEMPLATE(boost::spirit::action, 2) BOOST_TYPEOF_REGISTER_TEMPLATE(boost::spirit::positive, 1) #define an_expression +anychar_p[ a_placeholder ] typedef void (* fptr)(char const * from, char const * to); void echo(char const * from, char const * to) { // [...] } #ifndef MAKE_THINGS_FAIL using boost::spirit::anychar_p; class a_rule_parser : boost::spirit::parser<a_rule_parser> { fptr a_placeholder; struct __rule { // hide the placeholder member variable of the outer class with a static // variable of the same type static fptr & a_placeholder; typedef BOOST_TYPEOF( an_expression ) type; }; __rule::type __expr; public: a_rule_parser(fptr placeholder) : a_placeholder( placeholder ) , __expr( an_expression ) { } // [...] (it's stripped down code to show a bug) }; int main() { a_rule_parser x(& echo); return 0; } // so far, so good... #else // ...but the same code in templated form makes msvc8 do strange things // using boost::spirit::anychar_p; using namespace boost::spirit; // ADL doesn't seem to work, either but I could live with this problem ;-) template<typename T> class a_rule_parser_template : boost::spirit::parser< a_rule_parser_template<T> > { T a_placeholder; struct __rule { // hide the placeholder member variable of the outer class with a static // variable of the same type static T & a_placeholder; typedef BOOST_TYPEOF_TPL( an_expression ) type; }; typename __rule::type __expr; public: a_rule_parser_template(T const & placeholder) : a_placeholder( placeholder ) , __expr( an_expression ) { } // [...] (it's stripped down code to show a bug) }; int main() { a_rule_parser_template<fptr> x(& echo); return 0; } #endif

3 9

[foreach] Borland porters needed
by Eric Niebler 09 Feb '06

09 Feb '06

BOOST_FOREACH is no longer working for any version of Borland. Any Borland users out there care to submit a patch and get their names in the acknowledgements? :-) -- Eric Niebler Boost Consulting www.boost-consulting.com

4 10

How is BOOST_SLIST_HEADER meant to be used?
by Robert Ramey 09 Feb '06

09 Feb '06

I've modified my code to read: #ifdef BOOST_HAS_SLIST #include BOOST_SLIST_HEADER Now it is failing with the following compilers: mingw borland What am I missing here? Robert Ramey

3 2

[asio] Tutorial 7 using strdup, new
by Peter Dimov 09 Feb '06

09 Feb '06

Why is http://asio.sourceforge.net/boost-asio-proposal-0.3.6/libs/asio/doc/tutoria… using raw pointers, strdup, new, free and delete? This is soo 1996. Tutorials shoud teach. -- Peter Dimov

7 43

Re: [boost] serialisation - prevent restoring archive of class Xintoclass Y
by Oliver.Kowalke＠infineon.com 09 Feb '06

09 Feb '06

>Oliver.Kowalke(a)infineon.com wrote: >> Hello, >> is it possible to prevent that the archive of class X can be restored >> in class Y (both have the same layout)? >I don't understand what this means. >Robert Ramey I don't want Y created from a archive containing a X. Please look at the code below: #include <iostream> #include <sstream> #include <stdexcept> #include <boost/archive/text_oarchive.hpp> #include <boost/archive/text_iarchive.hpp> #include <boost/archive/xml_iarchive.hpp> #include <boost/archive/xml_oarchive.hpp> #include <boost/serialization/nvp.hpp> #include <boost/serialization/level.hpp> #include <boost/serialization/export.hpp> class X { public: friend class boost::serialization::access; template< typename Archive > void serialize( Archive & ar, unsigned int version) { ar & BOOST_SERIALIZATION_NVP( degrees); ar & BOOST_SERIALIZATION_NVP( city); } int degrees; std::string city; X(){}; X( int d, std::string const& c) : degrees( d), city( c) {} }; // BOOST_CLASS_EXPORT(X) // BOOST_CLASS_IMPLEMENTATION(X, boost::serialization::object_class_info) class Y { public: friend class boost::serialization::access; template< typename Archive > void serialize( Archive & ar, unsigned int version) { ar & BOOST_SERIALIZATION_NVP( degrees); ar & BOOST_SERIALIZATION_NVP( city); } int degrees; std::string city; Y(){}; Y( int d, std::string const& c) : degrees( d), city( c) {} }; // BOOST_CLASS_EXPORT(Y) // BOOST_CLASS_IMPLEMENTATION(Y, boost::serialization::object_class_info) int main() { try { std::stringstream ss; int degrees = -1342; std::string city; city += "abc"; { boost::archive::text_oarchive oa( ss); const X g( degrees, city); oa << g; } std::cout << ss.str() << std::endl; { boost::archive::text_iarchive ia( ss); Y g; ia >> g; // I would expect an exception; at least not Y rtestored if ( g.degrees == degrees && g.city == city) std::cout << "equal" << std::endl; else std::cout << "not equal" << std::endl; } return 0; } catch ( std::exception const& e) { std::cerr << e.what() << std::endl; } catch (...) { std::cerr << "unhandled" << std::endl; } return -1; }

2 1

[hash] hash_friend_test problem
by Vladimir Prus 09 Feb '06

09 Feb '06

Hi, the hash library had a new test, 'hash_friend_test' added to Jamfiles less than an day ago. However, I don't see any 'hash_friend_test.cpp' added, and bjam reports this: don't know how to make <@boost!libs!functional!hash!test>hash_friend_test.cpp Seems like "cvs add" + "cvs ci" is in order. - Volodya

1 0

[Hash] Changes for 1.34
by Daniel James 09 Feb '06

09 Feb '06

Hello everyone, This is just a quick summary of the changes I'm trying to get ready for the next version of boost. I've added a forward header for the library, which is intended for adding hash support to other boost libraries without pulling in the whole library. This is currently located at <boost/functional/hash_fwd.hpp>, but it might better placed in a detail directory. I'll document this soon. Although, I'm having some problems with some compilers, so I might revert this for the next release. I've changed this library to use forward declarations for the standard library, so that it doesn't need to include every header it supports. This is of course non-standard, but when it was discussed before there were no objections. To make my life easier I'm just including all the headers for compilers/libraries where it's tricky to get right. So far that's more recent versions of Borland and the gcc library in debug mode. This is to make a move to using a single header for the hash library, deprecating the <boost/functional/hash/*.?pp> headers (of course, it's not required, but makes a single header more acceptable). After that I will add the BOOST_HASH_NO_EXTENSIONS macro I discussed before to allow the user to turn of the extensions. I'm going to try get this all in before Friday (I've already implemented most of this but don't want to dump the whole thing into the regression tests in one big go). Also, I'm changing the portable example to reflect some portability problems that I've discovered - but please, don't review it just yet, I'm still experimenting. And if I'm lucky, I might even throw in support for std::complex. Sorry that this is all very last minute, I haven't had internet access at home recently, and only limited access right now. I'll revert anything which is problematic as none of this is particularly important. Oh, and it looks like I can't access CVS tonight so it isn't up to date. So it goes. Daniel

3 6