December 2008 - Boost-users - lists.preview.boost.org

Re: [Boost-users] [shared_ptr] Is there an interface to set the ptr a shared_ptr holds
by Max 29 Dec '08

29 Dec '08

Hello Scott， Thanks. It seems to be one of the simplest way to get to my needs to declare another companion auto_free_helper. It's better than none, if not the best. I'm using this technique for stream: std::ofstream outstream(strPath); boost::shared_ptr<std::ofstream> os_auto_closer(&outstream, boost::bind(&std::ofstream::close, _1)); B/Rgds Max ----- Original Message ----- From: Scott McMurray <me22.ca+boost(a)gmail.com> To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] [shared_ptr] Is there an interface to set the ptr a shared_ptr holds Date: 2008-12-29 11:35:53 On Sun, Dec 28, 2008 at 22:29, Max <loadcom(a)sina.com> wrote: > > I have the following code scenario: > > A *pa = NULL; > > // pa will be allocated here (by 3rd-party function) > // by setA(), > // whose behavior could not be changed > setA(pa); // prototype: void setA(A*& pa) > > //... > > pa->close(); > I think the following is fine: A *pa_ = 0; setA(pa_); boost::shared_ptr<A> pa(pa_, boost::bind(&A::close, _1)); Since the shared_ptr constructor will call the deleter if its internal allocation fails. ~ Scott _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users ------------------------------------------------------------------- 新浪空间——与朋友开心分享网络新生活！(http://space.sina.com.cn/ )

3 2

Re: [Boost-users] [shared_ptr] Is there an interface to set the ptr a
by Max 29 Dec '08

29 Dec '08

Further, I've written the following macros for such use case: #define DECLARE_VAR_PTR_WITH_AUTO_MEM_CLEANER(CLASS, VAR_PTR, MEM_FN_CLEANER)\ CLASS *VAR_PTR;\ boost::shared_ptr<CLASS> auto__cleaner__for__ ## VAR_PTR\ (VAR_PTR, boost::bind(&CLASS::MEM_FN_CLEANER, _1)); #define DECLARE_VAR_WITH_AUTO_MEM_CLEANER(CLASS, VAR, MEM_FN_CLEANER)\ CLASS VAR;\ boost::shared_ptr<CLASS> auto__cleaner__for__ ## VAR\ (&VAR, boost::bind(&CLASS::MEM_FN_CLEANER, _1)); Usage: DECLARE_VAR_PTR_WITH_AUTO_MEM_CLEANER(AcDbBlockTable, pBlockTable, close); Max Hello Scott， Thanks. It seems to be one of the simplest way to get to my needs to declare another companion auto_free_helper. It's better than none, if not the best. I'm using this technique for stream: std::ofstream outstream(strPath); boost::shared_ptr os_auto_closer(&outstream, boost::bind(&std::ofstream::close, _1)); B/Rgds Max ----- Original Message ----- From: Scott McMurray To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] [shared_ptr] Is there an interface to set the ptr a shared_ptr holds Date: 2008-12-29 11:35:53 On Sun, Dec 28, 2008 at 22:29, Max wrote: > > I have the following code scenario: > > A *pa = NULL; > > // pa will be allocated here (by 3rd-party function) > // by setA(), > // whose behavior could not be changed > setA(pa); // prototype: void setA(A*& pa) > > //... > > pa->close(); > I think the following is fine: A *pa_ = 0; setA(pa_); boost::shared_ptr pa(pa_, boost::bind(&A::close, _1)); Since the shared_ptr constructor will call the deleter if its internal allocation fails. ~ Scott _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users ------------------------------------------------------------------- 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

[shared_ptr] Is there an interface to set the ptr a shared_ptr holds
by Max 29 Dec '08

29 Dec '08

Hello, I have the following code scenario: A *pa = NULL; // pa will be allocated here (by 3rd-party function) // by setA(), // whose behavior could not be changed setA(pa); // prototype: void setA(A*& pa) //... pa->close(); I want to apply shared_ptr to make it easier to maintain, or, more specifically here, to save a close() call. new version: boost::shared_ptr<A> pa(NULL, boost::bind(&A::close, _1)); setA(pa.get()); // because get() returns T* instead of T*&, // this line chokes the compiler //pa->close(); not needed any more Is there any way to get my goal accomplished? Thanks for any help. B/Rgds Max ------------------------------------------------------------------- 新浪空间——与朋友开心分享网络新生活！(http://space.sina.com.cn/ )

2 1

Re: [Boost-users] [shared_ptr] Is it possible for a shared_ptr to
by Max 29 Dec '08

29 Dec '08

Thank you all for your information. Of course, new elegant solution or just simple hints are still welcome. B/Rgds Max ----- Original Message ----- From: Andy Wiese <andyw(a)mitosystems.com> 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 04:54:08 On Dec 27, 2008, at 2:17 PM, Alan M. Carroll wrote: > 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). > I find auto_ptr's strict semantics to be useful in APIs because it makes a clear statement about the lifecycle of an object. If some function returns an auto_ptr, then you know right there that the client has full authority over that objects lifecycle. > 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 > > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users ------------------------------------------------------------------- 新浪空间——与朋友开心分享网络新生活！(http://space.sina.com.cn/ )

1 0

[Serialization] How non-portable is the non-portable Binary Archive?
by James Madison 28 Dec '08

28 Dec '08

I am interested in using boost serialization for a project I am currently working on, but two requirements are hurting me. First, the archives have to be portable. Second, it's critical to keep the message size under a certain threshold which the text archives blow through. So I am wondering if I can somehow get away with using a binary archive, even though my choices seem to be a non-portable version, or a unsanctioned tested version. Oh, and I am pretty much stuck using Boost 1.32. So, basically, I am wondering: 1. How non-portable is the non-portable archive? While this is a mixed windows/linux environment, all the computers are running as x86 or x86_64, so the endian is the same and word boundary restrictions are mostly the same. Unfortunately, the mapping of data types might not be the same. Is there a simple way to test this if the answer is "maybe"? 2. How stable is the portable archive? Reading through the release notes for 1.37 it seems like that the code is solid, except for encoding doubles, which would not be a problem. Of course, I am being forced to use 1.32, which might be much more unstable . . . Thanks in advance for any replies.

2 1

[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