- Boost-users - lists.boost.org

Re: [Boost-Users] Four Questions on boost::shared_ptr
by Darin Adler 05 Feb '02

05 Feb '02

On 2/3/02 11:54 PM, "Jon Wang" <webmaster(a)boost.100ok.com> wrote: > 1. Mr Stroustrup said that there wasn't a need for auto_array, for "a better > solution is to use a vector". So, should we use vector or > scoped_array/shared_array? You should normally use std::vector unless you have some special reason to use boost:: scoped_array and boost::shared_array. Those classes are provided both for completeness, and because occasionally there are such special reasons. But not often. > 2. In the document, four implementations are said to have been discussed: > Direct detached, Indirect detached, Embedded detached and Placement detached. > I've tried to describle them (Are I right? See the attached file), but I don't > understand the Placement detached. Could someone please tell me what it's > like? The concept here is that the object is allocated in a special way, which require use of placement, so the count is in a header before the object. This is a form of the "embedded attached" scheme that does not require changes to the original object's class, but it does require a special form of new, so you couldn't just call the normal "new" and then give the object to such a smart pointer. [Note: There was no attached file.] > 3. In the file smart_ptr.hpp, we can see the private function > shared_ptr::share. > > I'm sorry that I don't understand that reason why ++*rpn should be done before > dispose(). Would someone please give me an example? Or is it like this? This function is gone from the latest implementation of shared_ptr slated for the next boost release, so it's probably not worthwhile to discuss this particular aspect of the old implementation in detail. The code you enclosed looks something like the kind of code that caused trouble -- the key is that deleting the old object pointed to can end up deleting the new shared pointer in some cases, and this must not happen before incrementing the reference count. The new implementation avoids this issue by using the well-known "copy and swap" idiom for implementing assignment, and is easier to understand. > 4. Why not use policies? I'm afraid that the word "parameterization will > discourage users" might not be always true, for we can use default parameters. We'll probably be using policies in future version, if we can work out all the issues. Andrei Alexandrescu is considering submitting his smart pointer class template from his Loki library to Boost for formal review, which is the most likely candidate for a Boost policy-based smart pointer. If it turns out well enough, it may replace all the existing smart pointer class templates. -- Darin

3 3

Re: [Boost-Users] Four Questions on boost::shared_ptr
by Jon Wang 04 Feb '02

04 Feb '02

>[Note: There was no attached file.] Oh, sorry! Here is the attached code. Am I right? Thanks very much. But still I haven't understand placement attached:-) #ifndef COUNTED_PTR_HPP #define COUNTED_PTR_HPP #include <cstring> #include <boost/config.hpp> // counted_ptr --------------------------------------------------------------// template <typename T> class RefLinking { protected: RefLinking(T* p) { init(p); } ~RefLinking() { dispose(); } template<typename Y> RefLinking(const RefLinking<Y>& r) { insert(&r); } // never throws RefLinking(const RefLinking& r) { insert(&r); } // never throws public: template<typename Y> void operator=(const RefLinking<Y>& r) { share(&r); } void operator=(const RefLinking& r) { share(&r); } void reset(T* p=0) { if ( px == p ) return; // fix: self-assignment safe dispose(); init(p); } // reset T* get() const { return px; } // never throws long use_count() const { long i = 0; const RefLinking* p = prev; while(p){ p = p->prev; ++i; } p = this; while(p){ p = p->next; ++i; } return i; } // never throws #if ( defined(BOOST_NO_MEMBER_TEMPLATES) && !defined(BOOST_MSVC6_MEMBER_TEMPLATES) ) || !defined( BOOST_NO_MEMBER_TEMPLATE_FRIENDS ) private: #endif T* px; RefLinking *prev, *next; // Don't split this line into two; that causes problems for some GCC 2.95.2 builds #if !defined( BOOST_NO_MEMBER_TEMPLATES ) && !defined( BOOST_NO_MEMBER_TEMPLATE_FRIENDS ) template<typename Y> friend class RefLinking; #endif void init(T* p) { px = p; prev = 0; next = 0; } void dispose() { if(prev) prev->next = next; else if(next) next->prev = prev; else delete px; } template <class Y> void insert(Y* rr) { RefLinking* r = (RefLinking*)rr; px = r->px; prev = (r); next = r->next; r->next = this; if(next) next->prev = this; } template <class Y> void share(Y* r) { if ((void*)this != (void*)r) { dispose(); insert(r); } } // share }; //*/ template <typename T> class EmbeddedAttached { protected: EmbeddedAttached(T* p) { init(p); } ~EmbeddedAttached() { dispose(); } template<typename Y> EmbeddedAttached(const EmbeddedAttached<Y>& r) : ptr(r.ptr) { // never throws if(ptr) ++(ptr->pn); } EmbeddedAttached(const EmbeddedAttached& r) : ptr(r.ptr) { if(ptr) ++(ptr->pn); } // never throws public: template<typename Y> void operator=(const EmbeddedAttached<Y>& r) { share(r.ptr); } void operator=(const EmbeddedAttached& r) { share(r.ptr); } void reset(T* p=0) { if ( ptr == p ) return; // fix: self-assignment safe dispose(); init(p); } // reset T* get() const { return ptr; } // never throws long use_count() const { return (ptr)? ptr->pn: 0; } // never throws #if ( defined(BOOST_NO_MEMBER_TEMPLATES) && !defined(BOOST_MSVC6_MEMBER_TEMPLATES) ) || !defined( BOOST_NO_MEMBER_TEMPLATE_FRIENDS ) private: #endif T* ptr; // Don't split this line into two; that causes problems for some GCC 2.95.2 builds #if !defined( BOOST_NO_MEMBER_TEMPLATES ) && !defined( BOOST_NO_MEMBER_TEMPLATE_FRIENDS ) template<typename Y> friend class EmbeddedAttached; #endif void init(T* p) { ptr = p; if(ptr) ++(ptr->pn); } void dispose() { if(ptr&&--(ptr->pn)== 0) delete ptr; } void share(T* r) { if (ptr != r && r) { // Q: why not px != rpx? A: fails when both == 0 ++(r->pn); // done before dispose() in case rpn transitively // dependent on *this (bug reported by Ken Johnson) dispose(); ptr = r; } } // share }; template <typename T> class IndirectDetached { protected: IndirectDetached(T* p) { init(p); } ~IndirectDetached() { dispose(); } template<typename Y> IndirectDetached(const IndirectDetached<Y>& r) : ptr(r.ptr) { // never throws ++(ptr->pn); } IndirectDetached(const IndirectDetached& r) : ptr(r.ptr) { ++(ptr->pn); } // never throws public: template<typename Y> void operator=(const IndirectDetached<Y>& r) { share(r.ptr); } void operator=(const IndirectDetached& r) { share(r.ptr); } void reset(T* p=0) { if ( ptr->px == p ) return; // fix: self-assignment safe dispose(); init(p); } // reset T* get() const { return ptr->px; } // never throws long use_count() const { return ptr->pn; } // never throws #if ( defined(BOOST_NO_MEMBER_TEMPLATES) && !defined(BOOST_MSVC6_MEMBER_TEMPLATES) ) || !defined( BOOST_NO_MEMBER_TEMPLATE_FRIENDS ) private: #endif struct IDHelper{ T* px; long pn; IDHelper(T* p): px(p), pn(1) {} ~IDHelper() { delete px; } }; IDHelper* ptr; // Don't split this line into two; that causes problems for some GCC 2.95.2 builds #if !defined( BOOST_NO_MEMBER_TEMPLATES ) && !defined( BOOST_NO_MEMBER_TEMPLATE_FRIENDS ) template<typename Y> friend class IndirectDetached; #endif void init(T* p) { try { ptr = new IDHelper(p); } catch (...) { delete p; throw; } } void dispose() { if(--(ptr->pn) == 0) delete ptr; } template <class Y> void share(IDHelper* r) { if ((void*)ptr != (void*)r) { // Q: why not px != rpx? A: fails when both == 0 ++(r->pn); // done before dispose() in case rpn transitively // dependent on *this (bug reported by Ken Johnson) dispose(); ptr = r; } } // share }; //*/ template <typename T> class DirectDetached { protected: DirectDetached(T* p) : px(p) { try { pn = new long(1); } catch (...) { delete p; throw; } } ~DirectDetached() { dispose(); } template<typename Y> DirectDetached(const DirectDetached<Y>& r) : px(r.px) { // never throws ++*(pn = r.pn); } DirectDetached(const DirectDetached& r) : px(r.px) { ++*(pn = r.pn); } // never throws public: template<typename Y> void operator=(const DirectDetached<Y>& r) { share(r.px, r.pn); } void operator=(const DirectDetached& r) { share(r.px, r.pn); } void reset(T* p=0) { if ( px == p ) return; // fix: self-assignment safe if (--*pn == 0) { delete px; } else { // allocate new reference counter try { pn = new long; } // fix: prevent leak if new throws catch (...) { ++*pn; // undo effect of --*pn above to meet effects guarantee delete p; throw; } // catch } // allocate new reference counter *pn = 1; px = p; } // reset T* get() const { return px; } // never throws long use_count() const { return *pn; } // never throws // Tasteless as this may seem, making all members public allows member templates // to work in the absence of member template friends. (Matthew Langston) // Don't split this line into two; that causes problems for some GCC 2.95.2 builds #if ( defined(BOOST_NO_MEMBER_TEMPLATES) && !defined(BOOST_MSVC6_MEMBER_TEMPLATES) ) || !defined( BOOST_NO_MEMBER_TEMPLATE_FRIENDS ) private: #endif T* px; // contained pointer long* pn; // ptr to reference counter // Don't split this line into two; that causes problems for some GCC 2.95.2 builds #if !defined( BOOST_NO_MEMBER_TEMPLATES ) && !defined( BOOST_NO_MEMBER_TEMPLATE_FRIENDS ) template<typename Y> friend class DirectDetached; #endif void dispose() { if (--*pn == 0) { delete px; delete pn; } } void share(T* rpx, long* rpn) { if (pn != rpn) { // Q: why not px != rpx? A: fails when both == 0 ++*rpn; // done before dispose() in case rpn transitively // dependent on *this (bug reported by Ken Johnson) dispose(); px = rpx; pn = rpn; } } // share }; template<typename T, class Storage = DirectDetached<T> > class counted_ptr: public Storage { public: typedef T element_type; explicit counted_ptr(T* p = 0) : Storage(p) {} template<typename Y, typename S> counted_ptr(const counted_ptr<Y, S>& r) : Storage(r) {} // never throws template<typename Y, typename S> counted_ptr& operator=(const counted_ptr<Y, S>& r) { *static_cast<Storage*>(this) = static_cast<const S&>(r); return *this; } counted_ptr(const counted_ptr& r) : Storage(r) {} // never throws counted_ptr& operator=(const counted_ptr& r) { *static_cast<Storage*>(this) = static_cast<const Storage&>(r); return *this; } T& operator*() const { return *get(); } // never throws T* operator->() const { return get(); } // never throws bool unique() const { return use_count() == 1; } // never throws void swap(counted_ptr& other) // never throws { using std::memcpy; const int size = sizeof(counted_ptr); char temp[size]; memcpy(temp, &other, size); memcpy(&other, this, size); memcpy(this, temp, size); } }; // counted_ptr #endif // COUNTED_PTR_HPP Jon Wang 2002-02-04

1 0

Re: [Boost-Users] Four Questions on boost::shared_ptr
by Darin Adler 04 Feb '02

04 Feb '02

On 2/4/02 12:51 AM, "Jon Wang" <webmaster(a)boost.100ok.com> wrote: > [Non-text portions of this message have been removed] I should have been more clear. This list doesn't allow attached files. -- Darin

1 0

Re: [Boost-Users] Four Questions on boost::shared_ptr
by Jon Wang 04 Feb '02

04 Feb '02

>[Note: There was no attached file.] Oh, sorry! Here is the attached file. Am I right? Jon Wang 2002-02-04 [Non-text portions of this message have been removed]

1 0

Unable to compile thread example on the document.
by 御風只配做笨木頭 04 Feb '02

04 Feb '02

Hi, I cannot compile the thread example listed on the document: http://www.boost.org/libs/thread/doc/thread.html#Example I'm using g++ 2.95.3 on a FreeBSD 4.x box. Here are the exact environment and compile error messages: % uname -a FreeBSD XXX.foo.bar 4.5-PRERELEASE FreeBSD 4.5-PRERELEASE #0: Tue Jan 8 03:14:40 CST 2002 root@XXX.foo.bar:/usr/src/sys/compile/kernel.XXX i386 % g++ -v Using builtin specs. gcc version 2.95.3 20010315 (release) [FreeBSD] % cat test_thread0.cpp #include <boost/thread/xtime.hpp> #include <boost/thread/thread.hpp> #include <iostream> struct thread_alarm { thread_alarm(int secs) : m_secs(secs) { } void operator()() { boost::xtime xt; boost::xtime_get(&xt, ::boost::TIME_UTC); xt.sec += m_secs; boost::thread::sleep(xt); std::cout << "alarm sounded..." << std::endl; } int m_secs; }; void run() { int secs = 5; std::cout << "setting alarm for 5 seconds..." << std::endl; boost::thread thrd(thread_alarm(secs)); std::cout << "main()" << std::endl; thrd.join(); } int main() { run(); } % g++ -I/usr/local/include test_thread0.cpp test_thread0.cpp: In function `void run()': test_thread0.cpp:28: request for member `join' in `thrd', which is of non-aggregate type `boost::thread ()(thread_alarm)' The boost library is installed through FreeBSD's ports collection which the boost version is 1.26.0. I've tried to figure out where the problem is and failed to find the definition of class thread. Do I miss something? Or, is it just a freebsd porting problem? ps. The example seems lack of a xtime inclution. So I add the #include at the first line: #include <boost/thread/xtime.hpp> -- Chien-Chou Hung / Jeff Hung / 洪健洲 (Big5) National Chiao Tung University Institute of Information Management 交通大學資訊管理所碩士生 http://www.jeffhung.idv.tw/ mailto:jeffhung.iim90g@nctu.edu.tw

2 1

Four Questions on boost::shared_ptr
by Jon Wang 04 Feb '02

04 Feb '02

Hi, I've got several questions about the former version of boost::shared_ptr. Would someone please help me? 1. Mr Stroustrup said that there wasn't a need for auto_array, for "a better solution is to use a vector". So, should we use vector or scoped_array/shared_array? 2. In the document, four implementations are said to have been discussed: Direct detached, Indirect detached, Embedded detached and Placement detached. I've tried to describle them (Are I right? See the attached file), but I don't understand the Placement detached. Could someone please tell me what it's like? 3. In the file smart_ptr.hpp, we can see the private function shared_ptr::share. void share(T* rpx, long* rpn) { if (pn != rpn) { ++*rpn; // done before dispose() in case rpn transitively // dependent on *this (bug reported by Ken Johnson) dispose(); px = rpx; pn = rpn; } } // share I'm sorry that I don't understand that reason why ++*rpn should be done before dispose(). Would someone please give me an example? Or is it like this? #include <boost/smart_ptr.hpp> struct A { boost::shared_ptr<A> v; }; int main() { A* pa = new A; boost::shared_ptr<A> ptr(pa); ptr = pa->v; } 4. Why not use policies? I'm afraid that the word "parameterization will discourage users" might not be always true, for we can use default parameters. Jon Wang 2002-01-31

1 0

Fw: Warning to all list owners and moderators
by Ihsan Ali Al Darhi 04 Feb '02

04 Feb '02

I don't know if this is true, but I received it on another group.... > > > > > > I don't know how valid this is but those of you > who belong to yahoo groups > might > want to pass it along to your list owners. > > > > I just got this from another group owner. I have > > heard of this happening, > > > and in one case banned a guy that made a similar > > attempt with one of my > > > groups. > > > Someone has been joining communities with the > nickname YAHOO COMMUNITY > MONITOR, and then posting the following: Yahoo > Community Monitor says: > Yahoo Group Managers have been receiving a few > complaints about your > group.I am required to join your site and monitor > it for about one month.I > will not do anything to damage or ruin your site,I > am only required to > monitor it.The only thing required of you is to > accept my application and > make me a Manager of your group for the time I am > there. If you cannot > meet > these requirements,Yahoo will have to shut your > group down. > DO NOT!! make this person a manager! SHE/HE is a fake, and has been deleting groups once made a manager. BAN the member immediately. Please let other groups know about this problem. I just wanted to let ya'll know. Wouldn't want anything to happen to your groups. Mohammed

1 0

Re: [Boost-Users] Suggestion -- Boost Building
by John Maddock 02 Feb '02

02 Feb '02

>> I used to be able to build this, but since the boost.build stuff came along, > I've been completely unable to build anything at all. The documentation is > worse than useless because it's so self-contradictory.... Don't forget that the regex makefiles are still there if you want them, there is nothing that says that you have to use Jam if you don't want to... - John Maddock http://ourworld.compuserve.com/homepages/john_maddock/

1 0

boost python and smart pointers
by Martin Strubel 01 Feb '02

01 Feb '02

Hello, I've lately experimented with Boost.Python and found it pretty useful. Though I've run into one thing which I couldn't figure out by the documentation, I hope someone can give me a hint there: I have a function returning a pointer to an 'owned' structure, i.e. its container takes care of creation and deletion. In Python, I want to wrap this pointer into a direct reference to the struct, i.e. no copying. To guard the access and make sure the object can't be accessed after the deletion, I want to wrap the smart pointer AS a python object. for example, a C++ function returning an 'owned' face from a mesh: FaceSmartPtr Mesh::addFace() { .. } should be wrapped for access in Python like: f = mesh.addFace() with subsequent calls for face manipulation like: f.addVertex((0.0, 0.0, 0.0)) where this should again call something like if (f_sptr.isValid()) f_sptr->addVertex(...); else throw_pyException_ptrInvalid(); Does Boost support this automatically ? Or do I have to wrap all these functions myself ? Greetings, - Strubi -- Martin Strubel, Software Hooligan strubi(a)blender.nl www.blender.nl Not a Number - 3D Enabling Technology tel. +31 40 2501231 Eindhoven, The Netherlands

1 0

RE: [Boost-Users] Suggestion -- Boost Building
by Wismar, John 01 Feb '02

01 Feb '02

In my case (the most recent how-can-I-build-this question) I'm trying (unsuccessfully) to use the regex library, which requires compiling.... I used to be able to build this, but since the boost.build stuff came along, I've been completely unable to build anything at all. The documentation is worse than useless because it's so self-contradictory.... -------------------------- John Wismar mailto:jwismar@alldata.com > -----Original Message----- > From: Duane Murphy [mailto:duanemurphy@mac.com] > Sent: Saturday, January 26, 2002 12:15 AM > To: Boost Users > Subject: [Boost-Users] Suggestion -- Boost Building > > > It seems a lot of questions come up on the list about > "building" the boost libraries. > > I wonder how many of these people realize that most of the > functionality of the Boost "libraries" is completely > encapsulated in the header files? There are significant > portions like regex and the graph library (?) that do require > compiling, but much of it does not. > > Is there anything that can be done to make it more clear that > compiling is often not necessary? > > I'm just trying to relieve some stress from people who may > very well not require compiling. And I have compassion for > those that do and are having trouble. > > I'm not sure exactly what can be done, but it might be worth > some thought. > > ...Duane > > > > ------------------------ Yahoo! Groups Sponsor > ---------------------~--> Sponsored by VeriSign - The Value > of Trust Pinpoint the right security solution for your > company - FREE Guide from industry leader VeriSign gives you > all the facts. > http://us.click.yahoo.com/lWSNbC/WdiDAA/yigFAA> /EbFolB/TM > > > -------------------------------------------------------------- > -------~-> > > Info: <http://www.boost.org> > Wiki: <http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl> > Unsubscribe: <mailto:boost-users-unsubscribe@yahoogroups.com> > > > Your use of Yahoo! Groups is subject to > http://docs.yahoo.com/info/terms/ > >

2 1