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

[boost-users][thread] second lock locks the owning thread?
by Igor R 28 Dec '08

28 Dec '08

Hello, I encountered strange situation (1.37, VC9SP1, WinXPSP3): boost::mutex m; boost::mutex::scoped_lock lock(m); boost::mutex::scoped_lock lock(m); // this one locks the thread Is this correct behavior or a bug? Thanks.

1 1

[intrusive,thread] Problems with VS in 64-bit mode
by Zeljko Vrba 28 Dec '08

28 Dec '08

I'm trying to use thread and intrusive libraries with VS 2008 (9.0.21022.8) In 32-bit mode (Win32 platform) everything works just fine. However, when I switch the configuration to x64, two problems appear: Intrusive problem: ================== With default settings, just including <boost/intrusive/list.hpp> produces the follwing error: C:\Users\zvrba\Work\boost_1_37_0\boost/intrusive/detail/mpl.hpp(191) : error C2953: 'boost::intrusive::detail::is_unary_or_binary_function_impl <R(__cdecl *)(T0)>' : class template has already been defined I have traced that back to the BOOST_INTRUSIVE_TT_TEST_MSC_FUNC_SIGS define, which is active when compiling with MS extensions turned on. Thread problem: =============== I turned off MS extensions, and this fixes the problem with the intrusive library (the header is included before the thread header). However, then I get the following error: C:\Users\zvrba\Work\boost_1_37_0\boost/config/requires_threads.hpp(29) : fatal error C1189: #error : "Threading support unavaliable: it has been explicitly disabled with BOOST_DISABLE_THREADS" This error is dependent on the language extensions being enabled or not. Fix: ==== I have temporarily solved the problem by hacking the list.hpp header: #if defined(_MSC_EXTENSIONS) && !defined(__BORLAND__) //#define BOOST_INTRUSIVE_TT_TEST_MSC_FUNC_SIGS #endif (the commented line is the fix). The cause is probably the following comment in the documentation: "On Itanium Processor Family (IPF) and x64 processors, __cdecl is accepted and ignored by the compiler" (same for __stdcall) This makes the decorated and undecorated signatures equivalent, so I tend to believe that this fix would be correct with the additional [&& !defined(_WIN64)] clause.

2 2

[Regex] Why is my RE not working?... :-)
by Adem 28 Dec '08

28 Dec '08

Given these definitions: RE: " (ip=|\\(|\\[)(\\d{1,3}\\.\\d{1,3}\\.\\d{1,3}\\.\\d{1,3})(\\]|\\)|,|;| |\n)"; Text: "BLAH ([1.2.3.4]) [5.6.7.8] ip=100.100.200.200 BLAH2 (77.48.32.42)\n"; the following code prints: IP=5.6.7.8 100.100.200.200 77.48.32.42 I wonder why it skips the first IP, ie. 1.2.3.4 ? Is there something wrong in my RE definition above, the code below, or in boost::regex? I'm sure it's just a silly error of mine but I don't see it... :-( ... bool GetIP(const char* ApszText, char* ARet_szIP) { // find all IPs in ApszText and return them via ARet_szIP (seperate 'em by a blank if multiple found) const char* szRe = " (ip=|\\(|\\[)(\\d{1,3}\\.\\d{1,3}\\.\\d{1,3}\\.\\d{1,3})(\\]|\\)|,|;| |\n)"; boost::regex re(szRe); string s = ApszText; string::const_iterator start = s.begin(), end = s.end(); boost::match_results<string::const_iterator> what; boost::match_flag_type flags = boost::match_default; int cFound = 0; while (boost::regex_search(start, end, what, re, flags)) { string sIP(what[2].first, what[2].second); if (!cFound++) ARet_szIP[0] = 0; // clear only if found any else strcat(ARet_szIP, " "); strcat(ARet_szIP, sIP.c_str()); start = what[0].second; // set the next start pos } return cFound > 0; } int main() { const char* szLine = "BLAH ([1.2.3.4]) [5.6.7.8] ip=100.100.200.200 BLAH2 (77.48.32.42)\n"; char szIP[256] = ""; bool f = GetIP(szLine, szIP); printf("IP=%s\n", szIP); return 0; }

4 4

Re: [Boost-users] [shared_ptr] Is it possible for a shared_ptr to
by Max 28 Dec '08

28 Dec '08

Thanks for your time. the reason why I'm not letting B accept a shared_ptr is that the B class in my code just mimics the 3rd party code, which is not customizable. B/Rgds Max ----- Original Message ----- From: Alan M. Carroll To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] [shared_ptr] Is it possible for a shared_ptr to hand over it ownership Date: 2008-12-28 04:17:38 I don't see why you would ever want to do that. Why not hand over ownership to another shared_ptr? Why can B in your example not be of type shared_ptr<>? Then you can hand over ownership by b = p1; p1 = 0; Or, if B must be a class, why not have a member that's a shared_ptr and have getOwnership(shared_ptr& src) do _member = src; src = 0; The object can be explicitly destroyed by assigning 0 to _member. I would recommend against using auto_ptr, as it is just so _odd_. I used it once when Boost.scoped_ptr wasn't quite what I wanted, but that was a mistake (although I did find a nifty bug in the Microsoft compiler implemementation of auto_ptr). At 06:57 AM 12/25/2008, Max wrote: >Suppose I have a shared_ptr which is the only owner of the pointee. >I hope there is a way to hand over its ownership to another object >without rendering the pointee being deleted. > >Or, if the answer is no, is there an alternative class that has >this feature? > >The use case is like this: > >B b; >{ >shared_ptr p1(A()), p2(A()); > >b.GetOwnership(p1.HandOver()); >// now p1 is empty >} >// p2 is deleted here > >// but the A obj p1 used to be pointing to >// has been handed over to b and is still >// alive here > >b.DestroyA(); >// the A obj holded by b is destroyed by b >// explicitly ------------------------------------------------------------------- 新浪空间——与朋友开心分享网络新生活！(http://space.sina.com.cn/ ) ------------------------------------------------------------------- 新浪空间——与朋友开心分享网络新生活！(http://space.sina.com.cn/ )

1 0

[fusion] Constructing a vector with non-default/non-copy constructor
by Tilo Nitzsche 27 Dec '08

27 Dec '08

class A { public: A(int i) : i(i) {}; private: A(); A(const A&); int i; }; typedef boost::fusion::vector<A, A> VecA; boost::fusion::vector<int, int> int_vec(1, 2); VecA v(int_vec); // doesn't compile, needs copy constructor I'm trying to construct a vector of classes that don't have either a default constructor or a copy constructor. Is that possible at all? Thanks, Tilo

2 1

Re: [Boost-users] [shared_ptr] Is it possible for a shared_ptr to hand over it ownership
by Alan M. Carroll 27 Dec '08

27 Dec '08

I don't see why you would ever want to do that. Why not hand over ownership to another shared_ptr? Why can B in your example not be of type shared_ptr<>? Then you can hand over ownership by b = p1; p1 = 0; Or, if B must be a class, why not have a member that's a shared_ptr and have getOwnership(shared_ptr& src) do _member = src; src = 0; The object can be explicitly destroyed by assigning 0 to _member. I would recommend against using auto_ptr, as it is just so _odd_. I used it once when Boost.scoped_ptr wasn't quite what I wanted, but that was a mistake (although I did find a nifty bug in the Microsoft compiler implemementation of auto_ptr). At 06:57 AM 12/25/2008, Max wrote: >Suppose I have a shared_ptr which is the only owner of the pointee. >I hope there is a way to hand over its ownership to another object >without rendering the pointee being deleted. > >Or, if the answer is no, is there an alternative class that has >this feature? > >The use case is like this: > >B b; >{ >shared_ptr p1(A()), p2(A()); > >b.GetOwnership(p1.HandOver()); >// now p1 is empty >} >// p2 is deleted here > >// but the A obj p1 used to be pointing to >// has been handed over to b and is still >// alive here > >b.DestroyA(); >// the A obj holded by b is destroyed by b >// explicitly

2 1

Re: [Boost-users] [ASIO] design question, multiple listenning sockets vs. one
by Jose Martinez 27 Dec '08

27 Dec '08

Overmind, Etienne, Thank you very much for your advice. I switched over to one socket design, one thread doing all the recving. I setup some worker threads to deal with the parsing and processing. I am putting the final touches on the new code now (integrating the new and altered classes into main() ). So hopefully if I have the stamina I should be testing it tonight. Also, the UDP buffers have been moved up to 16megs to deal with the onslaught. thanks again. jose

1 0

Is there a portable way to determine whether a type is functor ?
by ancode tran 26 Dec '08

26 Dec '08

Hi all, The following code only works under g++ 4.x: #include <boost/type_traits.hpp> template <class T> struct is_functor_helper { static boost::type_traits::no_type test (...); template <class U> static boost::type_traits::yes_type test (U*, typeof(&U::operator()) dummy = NULL); BOOST_STATIC_CONSTANT (bool, value = sizeof(test((T*)NULL)) == sizeof(boost::type_traits::yes_type)); }; template <class T> struct is_functor: public boost::integral_constant<bool, is_functor_helper<T>::value> { }; Can anyone give me a hint to make it work under other toolset, please ? Thanks in advance. Regards, Trung An Tran --------------------------------- Hãy cùng chat với bạn bè trên Yahoo! Pingbox mới toanh! Tạo một không gian riêng để chat trên blog là chuyện nhỏ.

2 1

what is boost "jam"?
by Robert P. J. Day 26 Dec '08

26 Dec '08

i admit that this is inevitably a dumb question but what is boost "jam" as opposed to the generic boost library? thanks. rday p.s. i want to install boost on a fedora 9 (soon to be fedora 10) system, and i just want to know what's essential and what isn't. ======================================================================== Robert P. J. Day Linux Consulting, Training and Annoying Kernel Pedantry: Have classroom, will lecture. http://crashcourse.ca Waterloo, Ontario, CANADA ========================================================================

3 2

Why does not ublas overload operators?
by Allen Chen 26 Dec '08

26 Dec '08

Six years ago, I design my own matrix library like matlab. I overload operators, it make it conveninent to do matrix manupulations. For example matrix A, B, C, D; A = B*C +D; It is easy to use and read. Why does not ublas overload such operators as +, -, *, /, ^, [], (),... Allen

2 1