August 2006 - Boost-users - lists.boost.org

enable_if problem
by Imre 02 Aug '06

02 Aug '06

Hi I have some non-compiling code, and I'd like to know if it's a compiler bug, or something I'm doing wrong. In the following code (it's a simplified version of the real code where I met the problem) the call to F<int> instantiates S<int>. I think the compiler should choose the specialized version of S<>, but it seems to use the primary template instead, leading to an invalid non-lvalue int -> int& conversion. Strangely, if I explicitly instantiate S<int> (uncomment the commented line), then it compiles fine. I also tried implementing enable_if and is_fundamental myself, and came to the exact same results. Also tested that version (with the custom enable_if and is_fundamental) with the online Comeau compiler, and it compiled fine. So, I'd like to know whether this stuff should work according to the standard, and if not, why; and if anyone could suggest a workaround that works with VC++ 7.1. Thanks, Imre Here's the code: #include <boost/type_traits.hpp> #include <boost/type_traits/is_fundamental.hpp> #include <boost/utility/enable_if.hpp> template <typename T, typename Enable = void> struct S { typedef T& type; }; template <typename T> struct S<T, typename boost::enable_if_c<boost::is_fundamental<T>::value>::type> { typedef T type; }; //template struct S<int>; // Compiles fine with VC if this line is uncommented template <typename T> typename S<T>::type F() { return 0; } int main(int argc, char* argv[]) { F<int>(); return 0; }

1 0

regex_constants::perl
by Winson Yung 02 Aug '06

02 Aug '06

Hello, I saw the following thread talking about how to set regex to work in perl extended mode: http://groups.google.com/group/boost-list/browse_thread/thread/4d5588fb7021… I am using 1.28.0 and has coded something like this already: boost::RegEx expr(is10k ? keywords_10k[i].m_Key : keywords_10q[i].m_Key, true); // Pass callback, strings to RegEx library unsigned int result = expr.Grep(cb, pBuf); question is how do I set this boost::regex_constants::perl like the follow example: boost::regex rxBlah("([a-z](?:_?[a-z0-9])*)_dvs", boost::regex_constants::perl | boost::regex_constants::icase); /Winson

2 3

replace_all_regex on c-string help
by llwaeva＠21cn.com 02 Aug '06

02 Aug '06

Hi there, I am using in-place replacement with replace_all_regex replace_all_regex( source, boost::regex("(%f)"), "(?1 xxx)", format_all); where source is of type std::string. But for some reason, I need to perform replacement on a C-string (char *), is replace_all_regex safe to do so? Will replace_all_regex handle memory issue for us? Thanks.

3 2

Re: [Boost-users] replace_all_regex on c-string help
by llwaeva＠21cn.com 02 Aug '06

02 Aug '06

>> But for some reason, I need to perform replacement on a >> C-string (char *), is replace_all_regex safe to do so? Will replace_all_regex handle memory >> issue for us? >Of course this would be impossible. Boost.Regex can't know how the >memory was allocated (e.g. malloc or new or some custom allocator), or >even if it *was* allocated. >Why do you need to operate on raw char*'s? Why can't you operate on >an std::string? I got the string (char*s), which is too big (>10M), from an external library, it's too slow to convert the raw string to std::string before it is handled by replace_all_regex

2 1

[shmem] problem retrieving objects from map
by Pataki Szilard 02 Aug '06

02 Aug '06

Hello, I have a problem retrieving objects from a map placed in shared memory. I have two processes: proc#1: creates the shared memory, puts a map into it and populates it with objects proc#2: connects to the shared memory and tries to read out the objects from the map Whenever I try to read out the objects from the map, I get zeroes for the internal members (a, b). The class definition for the objects: -------------------------------------------------- class ShmNode { private: int a; int b; public: ShmNode(); ShmNode(int newa, int newb); ~ShmNode(); void set(int newa, int newb); int getA(); int getB(); }; -------------------------------------------------- I'm posting the source codes for the two programs. I removed the synchronization stuff with semaphores for the sake of clarity: Proc#1: -------------------------------------------------- #include <boost/shmem/named_shared_object.hpp> #include <boost/shmem/allocators/allocator.hpp> #include <boost/shmem/containers/vector.hpp> #include <boost/shmem/containers/map.hpp> #include <boost/shmem/sync/named_semaphore.hpp> #include <iostream> #include "../shmem/ShmNode.hpp" typedef std::pair< const int, ShmNode*> MyPair; typedef boost::shmem::allocator < MyPair, boost::shmem::named_shared_object::segment_manager> MyAlloc; typedef boost::shmem::map< const int, ShmNode*, std::less<const int>, MyAlloc> MyMap; int main() { boost::shmem::named_shared_object segment; segment.create("/myshmem", 20000); MyAlloc nodesAlloc(segment.get_segment_manager()); std::less<const int> comparator; MyMap *table = segment.construct<MyMap>("table")(comparator, nodesAlloc); MyPair myPair1(1, new ShmNode(1,2)); MyPair myPair2(2, new ShmNode(3,4)); MyPair myPair3(3, new ShmNode(5,6)); table->insert(myPair3); table->insert(myPair1); table->insert(myPair2); return 0; } -------------------------------------------------- Proc#2 -------------------------------------------------- /* same includes */ /* same typedefs */ int main() { boost::shmem::named_shared_object segment; segment.open("/myshmem"); MyMap *table = segment.find<MyMap>("table").first; std::cout << "table size: " << table->size() << std::endl; MyMap::iterator iter; for (iter = table->begin(); ; iter++) { if (iter == table->end()) break; std::cout << iter->second->getA() << " " << iter->second->getB() << std::endl; } return 0; } -------------------------------------------------- The problem once again: in proc#2, iter->second->getA() and iter->second->getB() return 0. If I put the same loop into proc#1, the correct values are returned. As a further hint, the size of the map object in proc#2 is returned correctly (=3). What am I doing wrong? Regards, Szilard

3 3

[format] bind_arg with empty string
by Edson Tadeu 01 Aug '06

01 Aug '06

Boost.format bind_arg with an empty string is not working for Boost 1.33.1(it used to work in 1.32). The offending code is this one: format f("%s%-.32s:%|40t|%|.6|%|54t|%25s\n"); f.bind_arg(1,""); It crashes because of a null pointer in a buffer (buf_).

2 1

boost unit test compilation problem
by Minkoo Seo 01 Aug '06

01 Aug '06

Hi list. I'm trying to compile and run simple unit test, but only to fail: #include <boost/test/unit_test.hpp> using namespace std; using namespace boost::unit_test; using namespace boost; void force_division_by_zero() { BOOST_CHECKPOINT("About to force division by zero!"); int i = 1, j = 0; i = i / j; } test_suite* init_unit_test_suite(int, char*) { test_suite* test = BOOST_TEST_SUITE("Unit test example2"); test->add(BOOST_TEST_CASE(&force_division_by_zero), 1); return test; } This does not compile though this is the program shown in the document. The errors are: /usr/lib/gcc/x86_64-redhat-linux/4.0.2/../../../../lib64/crt1.o (.text+0x20): In function `_start': : undefined reference to `main' /tmp/ccytTZBu.o(.text+0x8f): In function `force_division_by_zero()': backoff_ngram_lm_test.cpp: undefined reference to `boost::test_tools::tt_detail::check_impl(boost::test_tools:: predicate_result const&, boost::basic_wrap_stringstream<char>&, boost::unit_test::basic_cstring<char const>, unsigned long, boost::test_tools::tt_detail::tool_level, boost::test_tools::tt_detail::check_type, unsigned long, ...)' /tmp/ccytTZBu.o(.text+0x173): In function `init_unit_test_suite(int, char*)': backoff_ngram_lm_test.cpp: undefined reference to `boost::unit_test::test_suite::test_suite(boost::unit_test:: basic_cstring<char const>)' /tmp/ccytTZBu.o(.text+0x1e8):backoff_ngram_lm_test.cpp: undefined reference to `boost::unit_test::test_suite::add(boost::unit_test::test_unit*, unsigned long, unsigned int)' /tmp/ccytTZBu.o(.gnu.linkonce.t._ZN5boost9unit_test14make_test_case ERKNS0_9callback0INS0_9ut_detail6unusedEEENS0_13basic_cstringIKcEE [boost::unit_test::make_test_case(boost::unit_test::callback0 <boost::unit_test::ut_detail::unused> const&, boost::unit_test::basic_cstring<char const>)]+0x35): In function `boost::unit_test::make_test_case(boost::unit_test::callback0 <boost::unit_test::ut_detail::unused> const&, boost::unit_test::basic_cstring<char const>)': backoff_ngram_lm_test.cpp: undefined reference to `boost::unit_test::ut_detail::normalize_test_case_name(boost:: unit_test::basic_cstring<char const>)' /tmp/ccytTZBu.o(.gnu.linkonce.t._ZN5boost9unit_test14make_test_ caseERKNS0_9callback0INS0_9ut_detail6unusedEEENS0_13basic_cstring IKcEE[boost::unit_test::make_test_case(boost::unit_test::callback0 <boost::unit_test::ut_detail::unused> const&, boost::unit_test::basic_cstring<char const>)]+0x65):backoff_ngram_lm_test.cpp: undefined reference to `boost::unit_test::test_case::test_case(boost::unit_test::basic_ cstring<char const>, boost::unit_test::callback0<boost::unit_test::ut_detail:: unused> const&)' collect2: ld returned 1 exit status I'm compiling the sorce like g++ filename.cpp. My environments: $ g++ -v Using built-in specs. Target: x86_64-redhat-linux Configured with: ../configure --prefix=/usr --mandir=/usr/share/man --infodir=/usr/share/info --enable-shared --enable-threads=posix --enable-checking=release --with-system-zlib --enable-__cxa_atexit --disable-libunwind-exceptions --enable-libgcj-multifile --enable-languages=c,c++,objc,java,f95,ada --enable-java-awt=gtk --with-java-home=/usr/lib/jvm/java-1.4.2-gcj-1.4.2.0/jre --host=x86_64- redhat-linux Thread model: posix gcc version 4.0.2 20051125 (Red Hat 4.0.2-8) Any idea? Sincerely, Minkoo Seo

4 6

[regex] Thread safety patch.
by John Maddock 01 Aug '06

01 Aug '06

With apologies to all concerned, and thanks to Aleksey Sanin for tracking down the problem, there is a const-correctness bug in Boost.Regex-1.33.x that strikes only in multithreaded programs on Unix platforms, if: * Multiple threads construct regexes concurrently, or * Multiple threads perform search and replace operations concurrently. There are two possible fixes: * Apply the attached patch (this has just gone into cvs along with an updated test case that detects the issue). Or, * define BOOST_REGEX_USE_C_LOCALE in boost/user.hpp and rebuild everything. Regards, John Maddock.

2 2

boost::threads
by Noel Yap 01 Aug '06

01 Aug '06

I was just wondering what the status of boost::threads is. It looks like it hasn't been released since last year. Is there an ETA for the next release? Thanks, Noel

1 0

Re: [Boost-users] (de)serializing map<const A*>
by Cheok Yan Cheng 01 Aug '06

01 Aug '06

Hmmm - I think the map size is already saved to disk. If you just copy the current templated implementation for map loading and instantiate a specialized version for your case, I would think that would handle the job. Robert Ramey ------------------------------------------------------- I am sorry. However, I do not quite understand. Let me more specific to my case. I have a data structure: map<const FiducialPair *, vector<RectPair> > fiducialPairSearchRegionMap; Both FiducialPair and RectPair can be serialized with boost no problem. I try to serialize the whole map by void save(Archive & ar, const unsigned int version) const { ar & fiducialPairSearchRegionMap; } void load(Archive & ar, const unsigned int version) { ar & fiducialPairSearchRegionMap; } I would get the following error: C:\Boost\include\boost-1_33_1\boost\archive\detail\iserializer.hpp(487): error C2440: 'reinterpret_cast' : cannot convert from 'typef *__w64 ' to 'void ** ' C:\Boost\include\boost-1_33_1\boost\archive\detail\iserializer.hpp(493): error C2664: 'T *boost::archive::detail::load_pointer_type<Archive,Tptr>::pointer_tweak<const FiducialPair>(const boost::serialization::extended_type_info &,void *,T &)' : cannot convert parameter 2 from 'typef ' to 'void *' with [ T=const FiducialPair, Archive=boost::archive::text_iarchive, Tptr=typef ] Conversion loses qualifiers May I know how I can apply ur suggested solution to my case? Thank you. __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

2 1