- Boost - lists.boost.org

git clone up
by troy d. straszheim 18 May '09

18 May '09

In case this reaches any other git enthusiasts, I've rigged up a git clone that is automatically tracking boost svn, here: http://sodium.resophonic.com/git/boost/ Click around on it... it's smoking fast, despite the fact that it is running on hardware I found in the trash on the end of a mediocre uplink. There is a little howto here: http://sodium.resophonic.com/git/boost/about/ The repository (minus checkout) is about 150M, it contains branches and tags going all the way back. It will take a while to do the initial clone but fetches after that are very fast. -t

4 8

release notes for 1.39.0?
by Neal Becker 18 May '09

18 May '09

Where can I find some more detailed release notes? I looked here: http://comments.gmane.org/gmane.comp.lib.boost.announce/233

2 1

[serialization] Enum loading may lead to uninitialized value [Ask for change]
by Andrey Upadyshev 18 May '09

18 May '09

Hi! I found that template boost::archive::detail::load_enum_type (boost\archive\detail\iserializer.hpp) may overvrite loading enum's value with uninitialized ones. See comments: template<class Archive, class T> struct load_enum_type { static void invoke(Archive &ar, T &t){ int i; // Initialize 'i' with trash ar >> boost::serialization::make_nvp(NULL, i); // Imagine: some code in archive skips loading t = static_cast<T>(i); // Ooops, get the trash into 't' } }; I think it is more secure to initialize 'i' from 't'. So, if archive's code skips loading, we just get 't' unchanged instead of corrupted. Probably, someone can fix this in svn? Andrey Upadyshev P.S. Sorry for my English.

2 2

[regex] patch for Sun CC
by Steven Watanabe 18 May '09

18 May '09

AMDG To get regex to work with Sun 5.9, I had to add std:: in one place. Most of the tests pass, but a few still fail for mysterious reasons. Okay to commit? In Christ, Steven Watanabe

2 1

[boost1.38][serialization][Patch Request] Request a patch for fixing crashes on SunOS 64bit platform
by Joanne chiang 18 May '09

18 May '09

Hi, Please consider : 1. fixing the regression issues (iostream error/exception, segmentation faults , illegal instruction etc.) happened when running the testsuite bundled with boost.serilation (v1.38, though I believe it applies to other revisions) on SunOS 5.8 sparc v9 (5.9, 5.10 64-bit also get the same crashes). 2. adding SunOS 64bit platform to the boost regression test trunk : http://www.boost.org/development/tests/trunk/developer/serialization.html ,if possible. Thanks very much, Mave

2 4

Re: [boost] [filesystem] copy_file() not atomic proposed fix
by Beman Dawes 17 May '09

17 May '09

On Fri, May 15, 2009 at 5:10 AM, Andre Merzky <andre@...> wrote: > Quoting [Beman Dawes] (May 12 2009): >> >> Ticket #2925 complains that "boost::filesystem cannot atomically copy >> file if target exists" ... >> >> ... >> >> The proposed solution is to add an option to copy_file(): >> >> enum class copy_option { fail_if_exists, overwrite_if_exists }; >> >> void copy_file( const Path & from_path, const Path & to_path, >> copy_option option=copy_option::fail_if_exists ); >> >> The implementation would be atomic on both POSIX and Windows. >> >> Comments? > > Makes sense. OK, fix committed to trunk. It will go in branches/release once sufficient regression tests have cycled. > I would suggest to check the complete lib for similar > problems. I think most operational and convenience > functions are prone to similar problems. Hum... "most" is too strong; many of the operational functions are implemented as a single operating system API call. I really don't want to spend time analyzing and fixing the V2 codebase. Instead, I've added an item to the V3 do-list to do an atomicity review. I'm planning to spend a lot of time on Filesystem.V3 this summer, and release a V3 beta as the code firms up. Thanks, --Beman

1 0

PropertyTree's XML parsers survey
by Sebastian Redl 17 May '09

17 May '09

Hi, PropertyTree currently has 5 (five!) different xml readers using 4 different parsers. I don't want to maintain them all. Looking at the open tickets for PT shows that most bugs occur in the various xml readers, and there's no point in having more than one. I just wonder which one to keep. To those who use PropertyTree's XML format, which parser do you use? Do you have any experience with the pros and cons of the various parsers? All feedback is highly appreciated. Sebastian

3 2

[test] Boost.Test with Sun CC
by Steven Watanabe 17 May '09

17 May '09

AMDG The attached patch allows the library to build with Sun 5.9. In Christ, Steven Watanabe

2 2

aligned storage problems with borland
by Robert Ramey 17 May '09

17 May '09

Serialization tests indicated that a recent change to aligned storage has broken serialization tests for borland. Could someone please look into this? Error E2031 ..\boost/aligned_storage.hpp 126: Cannot cast from 'aligned_storage<4,8> * const' to 'detail::aligned_storage::aligned_storage_imp<4,8> *' in function aligned_storage<4,8>::address() *** 1 errors in Compile *** Robert Ramey

2 1

[test] Using the Boost Unit Test Framework with shared libs
by Peter 17 May '09

17 May '09

Hi I am using Boost.Test from the current boost 1.38 , on cygwin ( gcc ) and I want to use it with multiple test files and with shared libs ( in order to speed up the compilation time which is too long for me when using the header include variant ). What is the recommended way ? ( I read the doc, several times and I read the _long_ threads on the mailing list concerning the static/shared library issue, but I am not sure that I understood all the details). I did the following: *in mainTest.cpp:* #define BOOST_TEST_DYN_LINK #define BOOST_TEST_MAIN #include "../inc/globalFixture.hpp" #include "../inc/testfile1.hpp" #include "../inc/testfile2.hpp" *testfile1.hpp:* #include "globalFixture.hpp" BOOST_FIXTURE_TEST_SUITE(ATlib1, Fixture) BOOST_AUTO_TEST_CASE(Resource_Manager) { ... ....BOOST_CHECK .... } BOOST_AUTO_TEST_SUITE_END() *testfile2.hpp:* #include "globalFixture.hpp" BOOST_FIXTURE_TEST_SUITE(ATlib2, Fixture) BOOST_AUTO_TEST_CASE(FileManager { ... BOOST_CHECK .... } BOOST_AUTO_TEST_SUITE_END() *globalfixture.hpp: *#ifndef _GLOBALFIXTURE_HPP #define _GLOBALFIXTURE_HPP #define BOOST_TEST_MODULE ATlib #include <iostream> #include <boost/test/unit_test.hpp> #include <boost/test/detail/unit_test_parameters.hpp> #include <boost/bind.hpp> #include <boost/test/output_test_stream.hpp> using namespace boost::unit_test; struct GlobalConfig { GlobalConfig() { std::cout << "GlobalConfig setup\n"; using namespace boost::unit_test; if (runtime_config::log_level() < log_warnings) unit_test_log.set_threshold_level(log_successful_tests); }; ~GlobalConfig() { std::cout << "GlobalConfig teardown\n"; } }; BOOST_GLOBAL_FIXTURE(GlobalConfig); struct Fixture { Fixture() { BOOST_TEST_MESSAGE("setup Fixture"); using namespace boost::unit_test; unit_test_log.set_threshold_level(log_successful_tests); } ~Fixture() { BOOST_TEST_MESSAGE("teardown Fixture"); } }; #endif /* _GLOBALFIXTURE_HPP */* *I then link to boost_unit_test_framework-gcc34-mt-1_38.so The link passes , but I get a warning from gcc++ 3.4.4 that auto-linking has been enabled ( I will provide the exact details if useful ). I can execute the resulting mainTest.exe without problems, but the compiler warnings tells that something is not quite right. Is the above example the right way to use multiple test files ? Are the flags for shared libs correctly set ? Thanks for your help. PS: I think that Boost.Test is not easy to grasp ( at least for me ). I think this is partially due to a lack of complete examples ( involving for examples a test suite with more than one test file ). Also the static/shared library topic and the related macros ( BOOST_TEST_DYN_LINK etc ) could do with a concrete example. Maybe the mentioned case above could serve as an example after clarification. Peter

3 3