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

Possible interest in select_by_predicate?
by Jonathan Brownell 11 Feb '02

11 Feb '02

I work in doing programming involving quite a bit of data analysis, and lately I've gotten sick and tired of doing routine searches on vectors and other linear containers to pull out maximum/minimum values. I ended up sitting down and writing a unary function object that retains a value based on any predicate. An example is worth a thousand words: int array[10] = { 5, 4, 6, 3, 9, 0, 7, 2, 8, 1 }; int maximum = std::for_each(&array[0], &array[10], select_by_predicate<int, std::greater>()); int minimum = std::for_each(&array[0], &array[10], select_by_predicate<int, std::less>()); assert(maximum == 9 && minimum == 0); Another example, using select_by_predicate independently: select_by_predicate<int, std::greater> max_error(0); for(loop through data samples) { double error; ... analyze data sample error ... max_error.compare(error); // same as max_error(error) } std::cout << "Maximum error: " << max_error; Is select_by_predicate something that boost users would be interested in? It's been helpful to me in all kinds of contexts. -Jonathan

3 4

Any use of a Singleton<> template
by sixfurdk 08 Feb '02

08 Feb '02

I currently working on a Singleton<> template, which currently has these features. 1) Supports the same interface/concepts as auto_ptr<> 2) One Line option to create a working singleton 3) Possible to use different Singletons of same type using names (simultaneously) 4) Possible to store different Singleton Objects under the same name (not simultaneously) 5) Possible to Release By Reference (Optional, Not Default) 6) Automatic Cleanup at program termination (Optional, Default) It can be used in the following manner: class SingletonObject { void someFunction(); ... }; Singleton<SingletonObject> singleton; singleton->someFunction(); Is there any interrest in such a class? Peter Sixhøj

2 2

tokenizer compilation problem with gcc 3.0.3 and stlport
by fintaeph 07 Feb '02

07 Feb '02

Hi all, I have been having trouble getting the tokenizer examples to compile using gcc 3.0.3 and stlport. I am using the example code from http://www.boost.org/libs/tokenizer/escaped_list_separator.htm, with the following compiler output: test.cpp: In instantiation of `::boost::::boost::iterator_ada ptor<::boost::detail::::boost::detail::token_iterator_base<co nst char *>,::boost::detail::::boost::detail::tokenizer_polic y<::_STL::::_STL::basic_string<char,::_STL::::_STL::char_trai ts<char>,::_STL::::_STL::allocator<char> >,::boost::::boost:: escaped_list_separator<char,::_STL::::_STL::char_traits<char> > >,::_STL::::_STL::basic_string<char,::_STL::::_STL::char_t raits<char>,::_STL::::_STL::allocator<char> >,const ::_STL::: :_STL::basic_string<char,::_STL::::_STL::char_traits<char>,:: _STL::::_STL::allocator<char> > &,const ::_STL::::_STL::basic _string<char,::_STL::::_STL::char_traits<char>,::_STL::::_STL ::allocator<char> > *,::_STL::forward_iterator_tag,int>': test.cpp:11: instantiated from here test.cpp:11: `sizeof' applied to incomplete type `::boost:::: boost::STATIC_ASSERTION_FAILURE<false>' If anyone has encountered and solved this problem before, I would greatly appreciate knowing what I'm doing wrong. Thanks, Chris. P.S. I get similar results with all the tokenizer examples.

3 3

modifed boost::stringtok
by hicks 06 Feb '02

06 Feb '02

Hello, I have created a modifed version of boost::stringtok() (the template function in boost\pending) with the aim of proving a more convient programmer interface. (It is called stringtok_2() below, but the interface is identical to stringtok()). The present version expects a container of objects which can be assigned a std::string. The modifed version is designed to accept a container of objects which can be converted to from std::string, at present "int" and "float" in addition to "std::string". E.g. char ac[] = "1 2 3 4 5"; std::vector<int> vi; boost::stringtok_2(vi,ac); std::vector<float> vf; boost::stringtok_2(vf,ac); std::vector<std::string> vs; boost::stringtok_2(vs,ac); As you can see in the cases for "int" and "float", the programmer can avoid having to create a vector of std::string and then convert them to int (or float) after calling stringtok. The implementation is achieved through template specialization. Behind the interface is an overloaded stringtok_2, template<typename Container, class Cvt> stringtok_2(Container l, Cvt cvt, const std::string s, const std::string ws) which takes a functor argument cvt which does the string to type conversion. The hithertoe used interface template<typename Container> stringtok_2(Container l, const std::string s, const std::string ws) calls the overloaded stringtok_2 with a converting functor string_cvt_ftr<Container::value_type>. string_cvt_ftr has a default declaration and two specializations, one for "int" and one for "float". template <class T> struct string_cvt_ftr { T operator()(const std::string& s) { return s; } }; template <> struct string_cvt_ftr<int> { int operator()(const std::string& s) { return std::atoi(s.c_str()); } }; template <> struct string_cvt_ftr<float> { float operator()(const std::string& s) { return std::atof(s.c_str()); } }; That is the partial specialization. Note that a vector of std::string can still be passed as the container argument. Since the interface semantics of are a superset of boost::stringtok(), I wonder if this could be substituted as a version-up, and use the same name. However, judging from the mail I see on this list, it is possible that some compilers might not handle it, so maybe that wouldn't be a good idea. I am sending this to boost-users since I do not currently recieve plain boost mail. I include the full code (including test code) below. Craig Hicks Engineer, KGK Tokyo //////////////////////////////////////////////////////////////////// // stringtok_2.h //--------------------------------------------------------------------------- #ifndef stringtok_2H #define stringtok_2H //--------------------------------------------------------------------------- #include <string> #include <cstring> // for strchr /***************************************************************** // modified with partial template specialization for int, float containers 2002.2.6 // Craig Hicks /***************************************************************** * This is the only part of the implementation that I don't like. * It can probably be improved upon by the reader... */ namespace stringtok_2_ns { inline bool isws (char c, char const * const wstr) { using namespace std; return (strchr(wstr,c) != NULL); } } namespace boost { /***************************************************************** * Simplistic and quite Standard, but a bit slow. This should be * templatized on basic_string instead, or on a more generic StringT * that just happens to support ::size_type, .substr(), and so on. * I had hoped that "whitespace" would be a trait, but it isn't, so * the user must supply it. Enh, this lets them break up strings on * different things easier than traits would anyhow. */ template <typename Container, typename Cvt> void stringtok_2 ( Container &l, Cvt cvt, std::string const &s, char const * const ws = " \t\n" ) { using namespace stringtok_2_ns; typedef std::string::size_type size_type; const size_type S = s.size(); size_type i = 0; while (i < S) { // eat leading whitespace while ((i < S) && (isws(s[i],ws))) ++i; if (i == S) return; // nothing left but WS // find end of word size_type j = i+1; while ((j < S) && (!isws(s[j],ws))) ++j; // add word l.push_back(cvt(s.substr(i,j-i))); // set up for next loop i = j+1; } } template <class T> struct string_cvt_ftr { T operator()(const std::string& s) { return s; } }; template <> struct string_cvt_ftr<int> { int operator()(const std::string& s) { return std::atoi(s.c_str()); } }; template <> struct string_cvt_ftr<float> { float operator()(const std::string& s) { return std::atof(s.c_str()); } }; template <typename Container> void stringtok_2 ( Container &l, std::string const &s, char const * const ws = " \t\n" ) { string_cvt_ftr<Container::value_type> cvt; stringtok_2(l, cvt, s, ws); } } // namespace boost #endif /////////////////////////////////////////////////////////////////// // test code void test_stringtok_2(std::ostream& os) /* Expected output:: int 1, 2, 3, 4, 5, float 1, 2, 3, 4, 5, string 1, 2, 3, 4, 5, */ { char ac[] = "1 2 3 4 5"; os << "int" << std::endl; std::vector<int> vi; boost::stringtok_2(vi,ac); std::copy(vi.begin(), vi.end(), std::ostream_iterator<int,char>(os,", ")); os << std::endl; os << "float" << std::endl; std::vector<float> vf; boost::stringtok_2(vf,ac); std::copy(vf.begin(), vf.end(), std::ostream_iterator<float,char>(os,", ")); os << std::endl; os << "string" << std::endl; std::vector<std::string> vs; boost::stringtok_2(vs,ac); std::copy(vs.begin(), vs.end(), std::ostream_iterator<std::string,char>(os,", ")); os << std::endl; }

2 2

Tokenizer question
by Edward Diener 06 Feb '02

06 Feb '02

Two tokenizer questions 1) I do not understand the constructor parameters to the char_delimits_separator class which is the default TokenizerFunction model. I assume the constructor would enable one to specify the delimiters for each token. I assume that the "returnable" parameter is a string literal of delimiters which are also returned as tokens and that "nonreturnable" is a string literal of delimiters which are not returned as tokens. I assume that the intersection of these string literals consists of all the delimiters. Given these assumptions, what is the purpose of the bool "return_delims" first parameter ? If "returnable" specifies a non-empty string literal, then why shouldn't these be the delimiters returned as tokens irregardless of the setting of "return_delims". If "returnable" specifies an empty string literal, then no delimiters will be returned as tokens once again irregardless of the setting of "return_delims". Would someone please explain why "return_delims" exists as a parameter in conjunction with the meaning of the "returnable" parameter ? It seems utterly redundant and unnecessary. 2) How does one create one own's TokenizerFunction to be plugged into the tokenizer class template as the first template type ? The documentation on TokenizerFunction concept leaves me utterly confused on a practical level. I assume I am creating a template class here despite the name which suggests it is a function, but the documentation never explains what it is. Maybe someone from Boost or John Bandela should consider updating it.

2 2

Another Tokenizer question
by s1imshady86 06 Feb '02

06 Feb '02

How do you specify that a white space is a returnable delimiter? The obvious, char_delimiters_separator<char> func (true, " "); Doesnt work. I know it's rather newbish question, but please help. A little more documentaion on token_iterator and TokenizerFunction concept (and perhaps the three specialized ones) would help greatly! Thanks, Ray

1 0

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

05 Feb '02

-------- Original Message -------- Subject: Re: [Boost-Users] Four Questions on boost::shared_ptr Date: Tue, 05 Feb 2002 10:08:25 +0900 From: hicks <hicks(a)kgk.co.jp> To: Boost-Users(a)yahoogroups.com, Duane Murphy <duanemurphy(a)mac.com> References: <B8838630.133F2%darin(a)bentspoon.com> <20020204170343.12718(a)mail.murphyslogic.com> Hi >>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. > >I consider user vector<> often. However there is a cost. For a fixed size >allocation of a buffer there is only the data allocated. If you use a >vector<> the vector also includes (for most implementations) the current >capacity, the current size, and the allocator. There is another sometimes very big overhead incurred by std::vector, and not incurred by by new T[]/ delete[]. Most implementions of std::vector will enter a loop and call a destructor on each element of the array, even if there is no machine code associated with that destructor. At least Borlands implemention of delete[] automatically detects this condition and does not insert a loop to call the destructors. This could be avoided if std::vector used a traits type to determine whether to enter a loop calling destructors. Craig Hicks KGK Tokyo Duane Murphy wrote: >--- At Mon, 4 Feb 2002 00:28:32 -0800, Darin Adler wrote: > >>On 2/3/02 11:54 PM, "Jon Wang" <webmaster(a)boost.100ok.com> <mailto:webmaster@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. >> > >I consider user vector<> often. However there is a cost. For a fixed size >allocation of a buffer there is only the data allocated. If you use a >vector<> the vector also includes (for most implementations) the current >capacity, the current size, and the allocator. > >I would be interested in finding a class that is somewhere between >allocating my own buffers and vector<>. Low overhead for simple >allocations; no resizing or growing. Even in this case, in order to >support iterator semantics the size is going to be needed. > >Sigh. > >I suppose a case like this might be when scoped_array would be used? > > ...Duane > > > > >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/ > > > > >

1 0

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

Cannot find regular expression symbols in Solaris debug
by phlasnicek 31 Jan '02

31 Jan '02

Hi, I am using Solaris Workshop 5.0 and I am using boost libraries for regular expressions. Everything builds fine and runs ok except when I try to attach to the process and run the debugger. I get the following error: warning: cannot demangle '__1c9uX__unnamed_6ofsRKVdR8zYNQre_get_error_str6FI_pkc_' Anyone have any suggestions, I would appreciate it very much. Thanks Paul

1 0

RE: [Boost-Users] Digest Number 81
by Anthony Kolarik 31 Jan '02

31 Jan '02

> Date: Tue, 29 Jan 2002 20:22:32 -0500 (EST) > From: Jeremy Siek <jsiek(a)cs.indiana.edu> > Subject: RE: RE: Dependencies for tokenizer? > > On Tue, 29 Jan 2002, Jeff Garland wrote: > jeff> > jeff> I'm sure there is. It is a matter of getting the "25 lines of > jeff> actual code" (my guess) in those 47 headers that tokenizer > jeff> really needs into a small header. I'm sure with a small > jeff> effort you could do it. Perhaps you could even find a way to > jeff> improve the overall design in the process. > > Well, it could get a little messy. The iterator adaptors use > some stuff > from the type_traits headers, which have lots of workarounds > for different > compilers, etc., etc. > > For a super simple C++ stringtok see the header > > boost/pending/stringtok.hpp > > It's not an official boost library, but you might find it fits > your needs. > > Cheers, > Jeremy Thank you for pointing out the pitfalls of reworking tokenizer and for pointing out boost/pending/stringtok.hpp, it serves fine for my purposes. -- Tony K.

1 0

1_26_0: regex/thread on W32 build via jam?
by John Maddock 31 Jan '02

31 Jan '02

>that's what I'm doing now. >I just want to fully understand the new jam way of building/using boost, >and eg to see what is neccessary to use the 'new way'. >Will the 'old way' be deprecated someday? Probably not any time soon. >Does/will the new way support static libs? It does for regex, can't comment on anything else though. - John Maddock http://ourworld.compuserve.com/homepages/john_maddock/

1 0

RE: [Boost-Users] Re: boost::pool + shared memory
by Tom Matelich 30 Jan '02

30 Jan '02

I've been using the ACE_Malloc framework from ACE (http://cs.wustl.edu/~schmidt/ACE.html) to solve a similar problem (sharing statics between DLL's). It takes a little while to figure out because the documentation is a little slim, but once you do it pays off pretty quickly. Doxygen documentation for ACE_Malloc is at: http://doc.ece.uci.edu/Doxygen/Current/html/ace/classACE__Malloc__T.html and the shared_mem memory pool is at: http://doc.ece.uci.edu/Doxygen/Current/html/ace/classACE__Shared__Memory__Po ol.html I ended up not having to write an STL allocator using ACE_Malloc, but I don't think it would be too difficult. If you get it right, a lot of ACE users would be your friend :) Tom Matelich > -----Original Message----- > From: xpav_2000 [mailto:alex@pavloff.net] > Sent: Tuesday, January 29, 2002 6:01 PM > To: Boost-Users(a)yahoogroups.com > Subject: [Boost-Users] Re: boost::pool + shared memory > > > --- In Boost-Users@y..., Tom Matelich <tmatelich@z...> wrote: > > Looks like pool may not be what I need, darn. I like > > to weasel as many boost classes into my project as possible. > > I've got a similar question -- I've got a whole pile of strings that > I want to share between processes, and I was thinking a reference- > counted string implementation with the actual data in shared memory > would be a good way to do that. > > Do you think a memory pool placed in shared memory, with the string > class modified to allocate/deallocate there would work? > > > > ------------------------ Yahoo! Groups Sponsor > ---------------------~--> > Tiny Wireless Camera under $80! > Order Now! FREE VCR Commander! > Click Here - Only 1 Day Left! > http://us.click.yahoo.com/nuyOHD/7.PDAA/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/

1 0

1_26_0: regex/thread on W32 build via jam?
by Bruno.Voigt＠ic3s.de 30 Jan '02

30 Jan '02

> Date: Tue, 29 Jan 2002 07:55:46 -0500 > From: John Maddock <john_maddock(a)compuserve.com> > Subject: 1_26_0: regex/thread on W32 build via jam? > > >But how do I have to change my project settings for a static executable? > >I just changed the the lib search path to this new monster path and it > >does not work > >The linker is missing the vc7-re300dm.lib then. > > why not carry on using the libs you built from the makefiles? Hi John, that's what I'm doing now. I just want to fully understand the new jam way of building/using boost, and eg to see what is neccessary to use the 'new way'. Will the 'old way' be deprecated someday? Does/will the new way support static libs? > >And how to link the boost thread stuff? > >Is it supported for statically linked apps at all? > Build threadmon.cpp into a dll, and the rest of the source into a static > lib with the options you want. Okay, thanks, I will try this combination. bye, Bruno

1 0

RE: [Boost-Users] boost::pool + shared memory
by Tom Matelich 30 Jan '02

30 Jan '02

Looks like pool may not be what I need, darn. I like to weasel as many boost classes into my project as possible. > -----Original Message----- > From: Tom Matelich [mailto:tmatelich@zetec.com] > Sent: Monday, January 07, 2002 2:49 PM > To: 'boost-users(a)yahoogroups.com' > Cc: 'shammah(a)voyager.net' > Subject: [Boost-Users] boost::pool + shared memory > > > I need to build a system for having miscellaneous objects in > a shared memory > segment. Essentially, we need a few inter-process singletons. These > classes have STL containers, so I was looking at using > pool_allocator and > writing a UserAllocator to use ACE_Mem_Map ( > http://doc.ece.uci.edu/Doxygen/Current/html/ace/classACE__Mem_ > _Map.html#_det > ails ) or something. I just wanted to see if anyone had done anything > resembling this and could give me any pointers or tips. > > Thanks, > > Tom Matelich > > PS Why can't I see who is a member of this group? I wanted > to check to see > if Steve Cleary was a member. Sorry if you are Steve. > > ------------------------ Yahoo! Groups Sponsor > ---------------------~--> > Tiny Wireless Camera under $80! > Order Now! FREE VCR Commander! > Click Here - Only 1 Day Left! > http://us.click.yahoo.com/WoOlbB/7.PDAA/ySSFAA/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

RE: Dependencies for tokenizer?
by Anthony Kolarik 30 Jan '02

30 Jan '02

> > > Any ideas about how to get this functionality w/o all the headers? Thanks, > > > -- Tony K. > > > In order to use the tokenizer classes I end up requiring 47 files, which > > What's wrong with 47 files? How much longer does it take to compile? It actually does slow things down noticeably, since it's a small project. > He might be sub setting boost and only wants to include the needed files. Exactly. I have 47 more files checked into source control that everybody else has to download (and have the tool check every time they refresh) so that I can parse a command line. I suppose I could check in the whole boost library, (~2000 files in the download I got,) but I'd rather just take what I need for now. > Anyway, this was brought up during the review, but at the time only 20 files > were needed (the very first tokenizer only need 2 boost headers). It was the > switch to iterator_adaptors that caused the include explosion at the time... Evidently reviewers thought 20 was reasonable, and I assume that some major problems are solved by the 47 headers that weren't by 20. I just wondered if there was a lighter weight way in boost to do my little tokenizing task. Is there? -- Tony K.

4 3

1_26_0: regex/thread on W32 build via jam?
by John Maddock 29 Jan '02

29 Jan '02

>But how do I have to change my project settings for a static executable? >I just changed the the lib search path to this new monster path and it >does not work >The linker is missing the vc7-re300dm.lib then. why not carry on using the libs you built from the makefiles? >And how to link the boost thread stuff? >Is it supported for statically linked apps at all? Build threadmon.cpp into a dll, and the rest of the source into a static lib with the options you want. - John Maddock http://ourworld.compuserve.com/homepages/john_maddock/

1 0