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

VC++ 6.0 ,SGI STL and Boost
by Joshua 23 Oct '02

23 Oct '02

I'm just trying to include the files for the shared_ptr class and keep having problems. I was able to fix a problem with a duplicate definition of 'max' and 'min' (in suffix.hpp, it was conflicting with a file from the SGI STL implementation) I mainly just commented it out, if someone has a better plan on how to fix it I would love to know how to fix it. I have tried Boost 1.28 and 1.29 and still got the same set of errors. In addition to the 'max' and 'min' conflict I am getting these errors : e:\programming projects\c++ projects\boost_1_28_0\boost\checked_delete.hpp(24) : error C2061: syntax error : identifier 'T' e:\programming projects\c++ projects\boost_1_28_0\boost\checked_delete.hpp(24) : error C2061: syntax error : identifier 'T' e:\programming projects\c++ projects\boost_1_28_0\boost\checked_delete.hpp(30) : error C2061: syntax error : identifier 'T' e:\programming projects\c++ projects\boost_1_28_0\boost\checked_delete.hpp(30) : error C2061: syntax error : identifier 'T' e:\programming projects\c++ projects\boost_1_28_0\boost\detail\shared_count.hpp(184) : error C2504: 'counted_base' : base class undefined e:\programming projects\c++ projects\boost_1_28_0\boost\detail\shared_count.hpp(206) : see reference to class template instantiation 'detail::counted_base_impl<P,D>' being compiled e:\programming projects\c++ projects\boost_1_28_0\boost\detail\shared_count.hpp(253) : error C2061: syntax error : identifier 'Y' e:\programming projects\c++ projects\boost_1_28_0\boost\detail\shared_count.hpp(254) : error C2065: 'Y' : undeclared identifier e:\programming projects\c++ projects\boost_1_28_0\boost\detail\shared_count.hpp(390) : fatal error C1506: unrecoverable block scoping error So I'm not sure if I'm doing something wrong (like not setting up the include order correctly) or if Microsoft is once again making my life hell. I am being bad and using the __int64 data type but I may need to rethink it and convert it to standard C++ (well sorta standard, I think gcc supports a longlong to get a 64bit int.. if not I may change it around to a bit_vector or something). But if anyone can shed some light on why I'm getting these errors listed above I'd really appreciate it. thanks. Joshua.

2 3

VC++ 6.0 with SGI STL and Boost
by Joshua 23 Oct '02

23 Oct '02

I'm using VC++ 6.0 with the SGI implementation of the STL. I'm trying to use the shared_ptr but when I try to include the <boost/shared_ptr.hpp> file I get about 40 errors. Most are related to errors redefining template functions. Its conflicting with the SGI STL header files but I'm not sure how to fix it. I keep thinking that I need to adjust the include order (Currently the Boost files are first, the the STL, then Microsofts includes. It doesn't matter where the Boost include directory is though I still get the same errors : this is a small sample --------------------Configuration: DryDock - Win32 Debug-------------------- Compiling... AbstractComponent.cpp e:\programming projects\c++ projects\boost_1_29_0\boost\config\suffix.hpp(256) : error C2995: 'min' : template function has already been defined e:\programming projects\c++ projects\drydock\drydock\stl\stl_algobase.h(106) : see declaration of 'min' e:\programming projects\c++ projects\boost_1_29_0\boost\config\suffix.hpp(260) : error C2995: 'max' : template function has already been defined e:\programming projects\c++ projects\drydock\drydock\stl\stl_algobase.h(112) : see declaration of 'max' e:\programming projects\c++ projects\boost_1_29_0\boost\checked_delete.hpp(24) : error C2061: syntax error : identifier 'T' e:\programming projects\c++ projects\boost_1_29_0\boost\checked_delete.hpp(24) : error C2061: syntax error : identifier 'T' e:\programming projects\c++ projects\boost_1_29_0\boost\checked_delete.hpp(30) : error C2061: syntax error : identifier 'T' e:\programming projects\c++ projects\boost_1_29_0\boost\checked_delete.hpp(30) : error C2061: syntax error : identifier 'T' e:\programming projects\c++ projects\boost_1_29_0\boost\detail\shared_count.hpp(213) : error C2504: 'counted_base' : base class undefined e:\programming projects\c++ projects\boost_1_29_0\boost\detail\shared_count.hpp(235) : see reference to class template instantiation 'detail::counted_base_impl<P,D>' being compiled And it goes on like that to a grand total of 40 errors. I know its probably just something like the order of the includes or I need to add something to specify the namespace correctly. Has any tried using this combination of libraries (VC++, SGI STL, Boost) and gotten it to work? Thanks.

2 1

Newbie trying to use multi_array ...
by Jon Pendergrass 23 Oct '02

23 Oct '02

I'm new to Boost, so please excuse my lack of knowledge. I'm trying to use the multi_array class and have hit a small snag. I'd like to have a class with a 4-D array. Initially it won't contain any data. However, Microsoft Visual C++ v6.0 complains about not finding a constructor. I want to do something like this: #include "boost/multi_array.hpp" typedef boost::multi_array<std::complex<double>, 4> t_Array4D; class foobar { t_Array4D ar; }; void foobar::SetValue() { // set extents code // set value this->ar[0][0][0][0] = 2.0; }

2 1

Error in date_time example -- exp-time_periods.html
by David Clark 22 Oct '02

22 Oct '02

date_time/doc/exp-time_periods.html class day_period : public { should be class day_period : public time_period {

1 0

Smart Pointers and Autocomplete
by eldritchshadows 22 Oct '02

22 Oct '02

Has anybody ever found a program which worked under Visual C++ that would list the member functions of the pointers wrapped by smart pointers. An example of what I'm talking about would be Visual Assist (except it doesn't seem to work). It would make life easier. Thanks in advance. Nathan Janken

1 0

Error in date_time example -- exp-print_month.html
by David Clark 21 Oct '02

21 Oct '02

There are two errors in the date_time/doc/exp-print_month.html example. std::cout<<"Enter Year: "; int should be std::cout << "Enter Year: "; int year,month; and //Use the calendar to get the last day of the month int eom_day = gregorian_calendar::end_of_month_day( should be int eom_day = gregorian_calendar::end_of_month_day(year,month);

3 2

[BGL] Why does depth_first_search not accept a source vertex parameter?
by kweeheong 21 Oct '02

21 Oct '02

Hi, I'm just learning about the BGL, and I've been sucessful in compiling and learning with examples on my VC6 Windows machine. I have a newbie question that someone with more knowledge of BGL might be able to answer : Q : Why does depth_first_search not accept a source vertex parameter, while breadth_first_search do? Is this just to be consistent with the historical definition of DFS? However, depth_first_visit and breadth_first_visit both accept a source vertex parameter. It would seem that DFS is more useful ( output is more deterministic ) if the source vertex can be specified, just like BFS. Reference : ( from the online BGL documentation ) template <class VertexListGraph, class P, class T, class R> void breadth_first_search(VertexListGraph& G, typename graph_traits<VertexListGraph>::vertex_descriptor s, const bgl_named_params<P, T, R>& params); template <class IncidenceGraph, class P, class T, class R> void breadth_first_visit(IncidenceGraph& G, typename graph_traits<IncidenceGraph>::vertex_descriptor s, const bgl_named_params<P, T, R>& params); template <class VertexListGraph, class class P, class T, class R> void depth_first_search(VertexListGraph& G, const bgl_named_params<P, T, R>& params); template <class IncidenceGraph, class DFSVisitor, class ColorMap> void depth_first_visit(IncidenceGraph& g, typename graph_traits<IncidenceGraph>::vertex_descriptor s, DFSVisitor& vis, ColorMap color);

1 0

Error in date_time example -- exp-print_holidays.html
by David Clark 20 Oct '02

20 Oct '02

Two errors in date_time/doc/exp-print_holidays.html In function print_date std::cout << d << " [" << d. should be std::cout << d << " [" << d.day_of_week().as_short_string() << "]\n"; and year is not defined in main() std::cout << "Enter Year: "; int should be std::cout << "Enter Year: "; int year;

2 2

A New Boost Mailing List
by Jon Kalb 18 Oct '02

18 Oct '02

As you know there are currently three Boost mailing lists: Boost - the discussion list for Boost library developers Boost-Users - the discussion list for Boost library users Boost-Announce - for library review and release announcements But we feel the need for an additional list. This list, Boost-Interest, will be an announcement list for items of interest to members of the Boost community. We see this as a low volume, moderated list on which commercial announcements are acceptable if relevant. We have written a proposed charter (below), but we invite comments and volunteers to moderate this list. Boost-Interest: Proposed Charter Boost-Interest is a moderated mailing list of announcements of interest to the Boost community. On topic messages will include announcements of books, magazine articles, papers, talks, seminars, products, tools, events, or conferences on advanced uses of C++, generic/generative/meta-programming, and, of course, the Boost libraries. Off topic will be discussions of any kind and job postings. If you are have comments on this list or its charter please send them to either or both Doug and Jon (addresses below). Also contact us if you are interested in moderating this list. Please do not ask for subscription information yet. We will announce details when we know them. -- Doug Gregor <gregod(a)cs.rpi.edu> Jon Kalb <Kalb(a)LibertySoft.com>

2 1

^M
by Felix E. Klee 18 Oct '02

18 Oct '02

Hi, I found that at least part of the Boost code contains DOS style carriage returns. Why is that? This causes parsing errors on UNIX. Felix

1 1

RE: [Boost-Users] Re: Error in Date_Time example - exp-days_alive.html
by Jeff Garland 18 Oct '02

18 Oct '02

David wrote: > I just checked the example code in > libs/date_time/example/gregorian/date_alive.cpp > > the correction should be > > else if (days_alive < date_duration(0)) { Thx -- I'll clean that up. Jeff

1 0

Template class static boost::mutex initialization !?!?!
by Luc Bergeron 17 Oct '02

17 Oct '02

Hello everyone, i'm currently trying to implement a thread safe singleton using template. In the class I need to use a boost::mutex in order to protect access to the singleton pointer. The problem i'm currently encountering is that I need to initialize this static boost::mutex, but how !?!?!?! I tried using a pointer to a mutex and initialize it to null but that didn't work. I also tried to new it in the initialization but that failed too. :( Can someone help me on this one ??? Thanks in advance for any help. Luc. P.S. I'm using boost 1.29 :) P.S.S Here is my current class definition : (A link to where I got this implementation : http://www.drizzle.com/~scottb/publish/gpgems1_singleton.htm) template <typename T> class ThreadSafeSingleton { static T* d_ThreadSafeSingleton; static boost::mutex d_mutex; public: ThreadSafeSingleton() { assert(!d_ThreadSafeSingleton); int offset = (int)(T*)1 - (int)(ThreadSafeSingleton <T>*)(T*)1; d_ThreadSafeSingleton = (T*)((int)this + offset); } ~ThreadSafeSingleton() { assert(d_ThreadSafeSingleton); d_ThreadSafeSingleton = 0; } static T& GetSingleton(void) { boost::mutex::scoped_lock scoped_lock(d_mutex); assert(d_ThreadSafeSingleton); return *d_ThreadSafeSingleton; } static T* GetSingletonPtr(void) { boost::mutex::scoped_lock scoped_lock(d_mutex); return d_ThreadSafeSingleton; } }; template <typename T> T* ThreadSafeSingleton <T>::d_ThreadSafeSingleton = 0;

1 1

RE: [Boost-Users] min & max #undef'd by config.hpp?
by richard_cox＠dell.com 17 Oct '02

17 Oct '02

On 17 October 2002 09:14, Chris Russell [mailto:cdr@encapsule.com] wrote: > > I'm confused. Using MS Visual Studio 6.0 w/the Intel 6.0 > compiler (or the > native VC compiler for that matter) the min and max macros > are defined in > the MS header WINDEF.H. It seems that config.hpp #undef's > these (although I > haven't figured out where or why yet). My environment seems > to be perfectly > happy using std::min/std::max. But what of all these > MS-defined macros that > use the undecorated min/max? I really don't want to go > chasing these down in > the MS + third-party headers if I can avoid it. Will someone > please explain > to me what's going on with config.hpp? There is nothing in windows.h that uses min & max. In fact if you define NOMINMAX *before* you include windows.h they will not be defined. > --- simple example --- > > #include <windows.h> > #include <boost/config.hpp> > #if defined(min) > #pragma message "min is defined" > #else > #pragma message "min not defined" > #endif > #if defined(max) > #pragma message "max is defined" > #else > #pragma message "max not defined" > #endif > int main(int argc, char* argv[]) > { > int a = 10, b = 20; int c = min(a,b); int d = max(a,b); > return 0; > } You either need: using std::min; using std::max; or fully qualify those names... > > Deleting intermediate files and output files for project 'test - Win32 > Debug'. > --------------------Configuration: test - Win32 > Debug-------------------- > Compiling... > StdAfx.cpp > test.cpp > . . . Boost configuring for Intel C++ compiler > min not defined > max not defined > L:\boost\test\test.cpp(24): error: identifier "min" is undefined > int c = min(a,b); ... hence this error, there is no ::min. > L:\boost\test\test.cpp(25): error: identifier "max" is undefined > int d = max(a,b); > ^ Richard Cox Senior Software Developer Dell Technology Online All opinions and statements mine and do not in any way (unless expressly stated) imply anything at all on behalf of my employer

2 1

Trivial doc error in http://boost.org/libs/smart_ptr/shared_ptr.htm
by Smith, Eric V. 17 Oct '02

17 Oct '02

Not sure if this is the right place to send this, but here goes. About half way down the page, "rile" should be "rule".

2 1

how to build boost with special gcc
by wugui 17 Oct '02

17 Oct '02

i install gcc-3.2 in a new directory. pl. /usr/local/gcc-3.2/ how build boost with it. Thanks.

2 1

Can't build library under vc7
by Alexander Semenov 17 Oct '02

17 Oct '02

Hi, I use > cl Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 13.00.9254 for 80x86 > cd /wrk/sft/boost/libs/python/build > bjam -sTOOLS=vc7 ...found 961 targets... ...updating 21 targets... vc-C++ ..\..\..\libs\python\build\bin\boost_python.dll\vc7\debug\runtime-link-dynam ic\function.obj function.cpp C:\wrk\sft\boost\boost\python\args_fwd.hpp(40) : error C2118: negative subscript or subscript is too large C:\wrk\sft\boost\boost\python\args_fwd.hpp(41) : see reference to class template instantiation 'boost::python::detail::error::more_keywords_than_function_arguments<,>' being compiled etc... Compiling fails on: C:\wrk\sft\boost\boost\python\args_fwd.hpp(40) : error C2118: negative subscript or subscript is too large C:\wrk\sft\boost\boost\python\args_fwd.hpp(41) : see reference to class template instantiation 'boost::python::detail::error::more_keywords_than_function_arguments<,>' being compiled C:\wrk\sft\boost\boost\pending\detail\property.hpp(75) : error C2039: 'result<`template-parameter258'>' : is not a member of 'boost::detail::property_value_end<PropertyList>' C:\wrk\sft\boost\boost\pending\detail\property.hpp(75) : fatal error C1507: previous user errors and subsequent error recovery halt further compilation Alexander Semenov.

2 1

Error in Date_Time example - exp-days_alive.html
by David Clark 17 Oct '02

17 Oct '02

There is an error in the following date_time example: libs/date_time/doc/exp-days_alive.html else if (days_alive < should be else if (days_alive < one_day) {

1 1

64-bit support for integer_traits on MSVC
by John Harris (TT) 17 Oct '02

17 Oct '02

It seems that there should be a way to get 64-bit integer support into integer_traits.hpp on MSVC. It looks like the only real problem is the missing feature of class constants, and the fact that integer_traits is using enums (which can't hold const_max and const_min values). But if there were specializations on the detail::integer_traits_base for MS 64-bit types, the constants could be declared in-class as __int64 but defined out-of-class (still in the header), safely. It's not as pretty as BOOST_STATIC_CONSTANT, but wouldn't it work? john harris trading technologies

1 0

min & max #undef'd by config.hpp?
by Chris Russell 17 Oct '02

17 Oct '02

I'm confused. Using MS Visual Studio 6.0 w/the Intel 6.0 compiler (or the native VC compiler for that matter) the min and max macros are defined in the MS header WINDEF.H. It seems that config.hpp #undef's these (although I haven't figured out where or why yet). My environment seems to be perfectly happy using std::min/std::max. But what of all these MS-defined macros that use the undecorated min/max? I really don't want to go chasing these down in the MS + third-party headers if I can avoid it. Will someone please explain to me what's going on with config.hpp? - Thanks & regards Chris Russell --- simple example --- #include <windows.h> #include <boost/config.hpp> #if defined(min) #pragma message "min is defined" #else #pragma message "min not defined" #endif #if defined(max) #pragma message "max is defined" #else #pragma message "max not defined" #endif int main(int argc, char* argv[]) { int a = 10, b = 20; int c = min(a,b); int d = max(a,b); return 0; } Deleting intermediate files and output files for project 'test - Win32 Debug'. --------------------Configuration: test - Win32 Debug-------------------- Compiling... StdAfx.cpp test.cpp . . . Boost configuring for Intel C++ compiler min not defined max not defined L:\boost\test\test.cpp(24): error: identifier "min" is undefined int c = min(a,b); ^ L:\boost\test\test.cpp(25): error: identifier "max" is undefined int d = max(a,b); ^ compilation aborted for L:\boost\test\test.cpp (code 2) Error executing xicl6.exe. Creating browse info file... BSCMAKE: error BK1506 : cannot open file '.\Debug\test.sbr': No such file or directory Error executing bscmake.exe. test.exe - 3 error(s), 0 warning(s)

1 0

integrating shared_ptr and building with gcc 2.7.2
by Nik Kolev 16 Oct '02

16 Oct '02

Hi everybody: I would like to make some of my classes reference counted, and I like what shared_ptr offers. The problem is (or is it really?) that I am building (have to) with gcc 2.7.2. The gcc.h header file which is part of the shared_ptr library does not like versions of gcc prior to 2.9.0. I would greatly appreciate it if somebody could point out a way to use the boost's shared_ptr technique and build with gcc prior to 2.9 Thanks, Nik

2 2

newbie here. can't get anywhere with the regex search example
by suzanne 16 Oct '02

16 Oct '02

hi all. i'm brand new with this and just trying to figure everything out. i build the regex search example using VC++ 5, but i can't seem to make it 'go'. in debug mode, my argv (the name of a .cpp file) is blank! argc is incremented properly, so it's seeing something. but no value at all for the argv i'm feeding it - just a good old "". i've got to take this thing apart & learn it so i can implement it in a tool i'm building - can anyone help me get started? i want to step through the various examples to learn it. thanks, suzanne

1 0

RE: boost vs. The Lawyers (non-technical)
by Brian Neal 16 Oct '02

16 Oct '02

Thanks for the input so far. I too would like to call for some blanket license that all boost developers agree to (if possible). It looks like I'll be contacting developers individually and trying to get them to sign something for us. Not fun for either side I guess. Thanks, Brian Neal

3 2

[random] Known problems compiling random lib with Dinkumware STL?
by Chris Russell 16 Oct '02

16 Oct '02

I'm using the Intel 6.0 compiler under MS Visual Studio 6.0 SP6 and just attempted to plug the Dinkumware STL distribution into the mix. I ran into a slew of compile-time errors in the random library and am wondering if anyone has seen this before I ride into battle. Thanks - Chris

3 4

RE: [Boost-Users] Re: Type traits direct compiler support
by Beman Dawes 15 Oct '02

15 Oct '02

At 11:11 AM 10/15/2002, scleary(a)jerviswebb.com wrote: >> It is the vagueness of the term "compiler support", which you did not >> alleviate in your answer, which I am trying to clarify. > >"Compiler support" means "anything the compiler can give us so that we can >implement that class correctly". > >It could be a specially-recognized preprocessor define, a quasi-class >(e.g., >SGI's type traits), a pragma, a keyword, or anything else that the compiler >provides. Yes. "Compiler support" can be seen as a courtesy by library experts, so they aren't trying to tell core language experts exactly how to do something. --Beman

2 1

RE: [Boost-Users] Re: Type traits direct compiler support
by Mark Storer 15 Oct '02

15 Oct '02

What exactly can't you do with existing techniques? I was Most Impressed by Modern C++ Design, and I don't recall any limits stated on what was possible... the whole concept is relatively new to me --Mark Storer Software Engineer Cardiff Software #include <disclaimer> typedef std::disclaimer<Cardiff> Discard;

1 0

RE: [Boost-Users] Re: Running the Boost_1_29_0 regression tests
by Beman Dawes 15 Oct '02

15 Oct '02

At 12:02 PM 10/15/2002, Arjun Bijanki wrote: >"David Abrahams" writes: > >> "Arjun Bijanki" <arjunb(a)microsoft.com> writes: > >> > Hi, I'm having trouble running the Boost_1_29_0 regression tests. A >lot >> > >> >> Arjun, >> >> What are you doing to compile/link the tests? >> The supported method uses bjam. Have you tried that? >> If you use the -d+2 option you can look at the command-lines used and >> replicate them in whatever environment you like. >> >> -Dave > >Thanks. Previously, I was using /status/regression.cpp to generate the >status table; I'll try using bjam instead. We probably should have removed regression.cpp from the 1.29.0 release. The thought occurred to me, but that was before we decided to include all the new regression test components in the current CVS main trunk. Now that the main trunk has everything needed for the new tests, there is no reason to keep regression.cpp around. I'll remove it. --Beman

1 0

RE: [Boost-Users] Re: Running the Boost_1_29_0 regression tests
by Arjun Bijanki 15 Oct '02

15 Oct '02

"David Abrahams" writes: > "Arjun Bijanki" <arjunb(a)microsoft.com> writes: > > Hi, I'm having trouble running the Boost_1_29_0 regression tests. A lot > > > > Arjun, > > What are you doing to compile/link the tests? > The supported method uses bjam. Have you tried that? > If you use the -d+2 option you can look at the command-lines used and > replicate them in whatever environment you like. > > -Dave Thanks. Previously, I was using /status/regression.cpp to generate the status table; I'll try using bjam instead. Arjun

1 0

RE: [Boost-Users] Re: Type traits direct compiler support
by scleary＠jerviswebb.com 15 Oct '02

15 Oct '02

> -----Original Message----- > From: Edward Diener [mailto:eddielee@tropicsoft.com] > > > > From: Edward Diener [mailto:eddielee@tropicsoft.com] > > > > > > What does it mean in the type_traits documentation that "direct compiler > > > support for that traits class is required" ? I do not see that terminology > > > explained anywhere and am lost as far as to what it refers. > > > > It means that there is no known way to implement that class (in a > > Standard-compliant, portable manner). > > > > The type traits library is in a rather uncomfortable position, since it is > > essentially trying to add reflection to the C++ language after the fact. > > > > As currently implemented, the type traits library bends over backwards to > > try to provide as much as possible to as many different compilers as > > possible. But the simple fact is that it should not really be a library, > > but a part of the supplied language. > > > > Some type traits classes technically require compiler support. However, > > those classes will all try to take their "best guess" at the correct answer > > -- sometimes defaulting to a "safe" default if they can't figure out the > > correct answer. > > What sort of compiler support is being talked of here, since you mention > that it can't be done in a standard-compliant manner ? Does this mean > looking at preprocessor defines to determine the correct results, or what ? > It is the vagueness of the term "compiler support", which you did not > alleviate in your answer, which I am trying to clarify. "Compiler support" means "anything the compiler can give us so that we can implement that class correctly". It could be a specially-recognized preprocessor define, a quasi-class (e.g., SGI's type traits), a pragma, a keyword, or anything else that the compiler provides. -Steve

1 0

Returning a tokenizer by value
by Brian_X7 15 Oct '02

15 Oct '02

Is there a problem with returning a tokenizer from a function? Here's what I've got: typedef boost::tokenizer< boost::char_separator<char> > tokenizer; tokenizer getTokens( istream& is ) { boost::char_separator<char> sep(",","", boost::keep_empty_tokens); std::string str; getline(is,str,'\n'); tokenizer toks( str, sep ); return toks; } .. tokenizer tokens = getTokens(input); tokenizer::iterator it = tokens.begin(); (*it) now contains garbage, interesting enough, iterating through the rest of the container works. when I move the processing of getTokens into the main code body the first token (*it) is then correct. Am I missing something fundamental here?

1 0

RE: [Boost-Users] Type traits direct compiler support
by scleary＠jerviswebb.com 15 Oct '02

15 Oct '02

> -----Original Message----- > From: Edward Diener [mailto:eddielee@tropicsoft.com] > > What does it mean in the type_traits documentation that "direct compiler > support for that traits class is required" ? I do not see that terminology > explained anywhere and am lost as far as to what it refers. It means that there is no known way to implement that class (in a Standard-compliant, portable manner). The type traits library is in a rather uncomfortable position, since it is essentially trying to add reflection to the C++ language after the fact. As currently implemented, the type traits library bends over backwards to try to provide as much as possible to as many different compilers as possible. But the simple fact is that it should not really be a library, but a part of the supplied language. Some type traits classes technically require compiler support. However, those classes will all try to take their "best guess" at the correct answer -- sometimes defaulting to a "safe" default if they can't figure out the correct answer. -Steve

2 2

[ANN] Quaternions, Octonions and Special Functions: please download from CVS.
by Hubert HOLIN 15 Oct '02

15 Oct '02

Somewhere in the E.U., le 15/10/2002 Bonjour Please note that due to some unfortunate circumstances beyond our control, the quaternion, octonion and special functions libraries which shipped with Boost 1.29.0 were not in the state they were meant to be in. You might want to download new versions of these libraries from the CVS in a couple of days. The version which should be present there adds compatibility with M$VCPP 7, and fixes things which CWPro 8 has broken in some circumstances. In a few days there will be another version with a user-contributed fix for gcc2.95.x and stlport on linux, among other things. Merci Hubert Holin Hubert.Holin(a)Bigfoot.com

1 0

Running the Boost_1_29_0 regression tests
by Arjun Bijanki 14 Oct '02

14 Oct '02

Hi, I'm having trouble running the Boost_1_29_0 regression tests. A lot of tests don't appear to have main() defined. As an example, one of the tests fails with the following: *** libs/utility/ref_test.cpp *** ** Microsoft Visual C++ cl /Feboosttmp.exe /Zc:wchar_t /nologo /Zm400 /MDd /W3 /GR /GX /Zi /Od /GZ /I".." /D "WIN32" /D "_DEBUG" /D "_MBCS" /_CONSOLE" ../libs/utility/ref_test.cpp /link /incremental:no user32.lib ref_test.cpp LINK : fatal error LNK1561: entry point must be defined 0.80 s Return code: 2 Fail libs/utility/ref_test.cpp has the following lines: #define BOOST_INCLUDE_MAIN #include <boost/test/test_tools.hpp> But this doesn't seem to define a main() function like it did in Boost_1_28_0. Do I need to do some different kind of installation for Boost_1_29_0, or did I pick up wrong headers, or is this function missing? I downloaded Boost_1_29_0 from: http://boost.sourceforge.net/release/boost_1_29_0.zip Thanks! Arjun Bijanki Microsoft Visual C++

2 1

Formal Review: The Spirit Parser Framework.
by John Maddock 14 Oct '02

14 Oct '02

The formal review for the Spirit Parser Framework begins today, and runs until the end of Sunday 20th October. Discussion should take place on the main boost developer list (boost(a)lists.boost.org) please state in your review whether you think the framework should be accepted into boost or not. The framework can be downloaded from: http://spirit.sourceforge.net/index.php?doc=download/v1_5.html Synopsis: ~~~~~~ Spirit is an object oriented recursive descent parser generator framework implemented using template meta-programming techniques. Expression templates allow us to approximate the syntax of Extended Backus Normal Form (EBNF) completely in C++. Parser objects are composed through operator overloading and the result is a backtracking LL(inf) parser that is capable of parsing rather ambiguous grammars. The Spirit framework enables a target grammar to be written exclusively in C++. Inline EBNF grammar specifications can mix freely with other C++ code and, thanks to the generative power of C++ templates, are immediately executable. In retrospect, conventional compiler-compilers or parser-generators have to perform an additional translation step from the source EBNF code to C or C++ code. Regards, John Maddock (Spirit review manager). http://ourworld.compuserve.com/homepages/john_maddock/index.htm

2 1

Type traits direct compiler support
by Edward Diener 14 Oct '02

14 Oct '02

What does it mean in the type_traits documentation that "direct compiler support for that traits class is required" ? I do not see that terminology explained anywhere and am lost as far as to what it refers.

1 0

RedHat 7.3
by S. Scott Collis 14 Oct '02

14 Oct '02

I appologize if this has been addressed before. I did a quick search of the of the new group and didn't find anything. I just tried to port a code that uses Boost to a new computer loaded with RedHat 7.3. It seems that in 7.3 the /usr/include/asm/atomic.h header is different and generates a warning (i.e. #warning Using kernel header in userland program. BAD! instead of supplying routines like atomic_add and atomin_sub that are apparently required by Boost for the shared_ptr implementation. I ended up replacing this 7.3 header with one from 7.1 and this fixed the problem, but if anyone has a better solution, I would appreciate it. Thanks, Scott -- ===================================================================== Dr. Scott Collis, Asst. Prof. | Tel: (713) 348-3617 Dept. of Mechanical Engineering | Fax: (713) 348-5423 and Materials Science - MS 321 | Rice University | Email: collis(a)rice.edu Houston, TX 77005-1892 | http://mems.rice.edu/~collis =====================================================================

2 1

d:\boost\boost_1_29_0\libs\thread\src\thread.cpp(107) : error C2065: '_beginthre
by satya_vajja 13 Oct '02

13 Oct '02

Hello, I am getting the following error while i compile boost thread library under VC++ 6.0. Could any of you please help me out. d:\boost\boost_1_29_0\libs\thread\src\thread.cpp(107) : error C2065: '_beginthreadex' : undeclared identifier When i define _MT compile flag the compiling error goes off but i get lot of linker errors. satya_vajja(a)yahoo.com Thanks Satya

3 5

regex questions in 1_29_0 with g++ 3.1.1
by Anibal Jodorcovsky 13 Oct '02

13 Oct '02

Hi all, I've been using regex 3.20 with g++ 2.95.3 with no problems. We're in the process of moving towards g++ 3.1.1 (we have a library we need to use and it core dumps with 2.95.3 due to a bug in the stdc++ that goes with it in ofstream). So, I have g++ 3.1.1 working fine in redhat 6.2. I couldn't find the regex standalone library so I ended up downloading the whole boost suite (1_29_0) and I built only the regex++ library. I have one comment regarding this version (which I don't know what it is BTW. What's the version of regex++ in boost 1_29_0?, I guess it's 3.31). It seems that boost:cmatch doesn't work anymore when doing a regex_search, I ended up modifying boost:cmatch with boost::match_results<std::string::const_iterator>, I hope that's ok. I'm also having some trouble compiling some code but this is a problem with the new STL implementation that ships with g++ 3.1.1 (the v3 STL). I can't seem to find the solution, that's why I'm asking here, if anybody has any pointers I'll be very grateful. Look at what I get: commit.cpp: In function `void findSopInstances(IDB_HANDLE**, const std::vector<CommitSopInfo, std::allocator<CommitSopInfo> >&, std::vector<CommitSopInfo, std::allocator<CommitSopInfo> >&, std::vector<CommitSopInfo, std::allocator<CommitSopInfo> >&)': commit.cpp:737: conversion from `__gnu_cxx::__normal_iterator<const CommitSopInfo*, std::vector<CommitSopInfo, std::allocator<CommitSopInfo> > > ' to non-scalar type `__gnu_cxx::__normal_iterator<const CommitSopInfo*, std::vector<const CommitSopInfo, std::allocator<const CommitSopInfo> > >' requested commit.cpp:738: no match for `__gnu_cxx::__normal_iterator<const CommitSopInfo*, std::vector<const CommitSopInfo, std::allocator<const CommitSopInfo> > >& < __gnu_cxx::__normal_iterator<const CommitSopInfo*, std::vector<CommitSopInfo, std::allocator<CommitSopInfo> > >' operator commit.cpp:742: cannot convert `__gnu_cxx::__normal_iterator<const CommitSopInfo*, std::vector<const CommitSopInfo, std::allocator<const CommitSopInfo> > >' to `const CommitSopInfo*' for argument `2' to `bool haveSopInstance(IDB_HANDLE**, const CommitSopInfo*)' commit.cpp: In function `int buildSopSeq(const std::vector<CommitSopInfo, std::allocator<CommitSopInfo> >&, SQ_TYPE, void***)': commit.cpp:913: conversion from `__gnu_cxx::__normal_iterator<const CommitSopInfo*, std::vector<CommitSopInfo, std::allocator<CommitSopInfo> > > ' to non-scalar type `__gnu_cxx::__normal_iterator<const CommitSopInfo*, std::vector<const CommitSopInfo, std::allocator<const CommitSopInfo> > >' requested commit.cpp:914: no match for `__gnu_cxx::__normal_iterator<const CommitSopInfo*, std::vector<const CommitSopInfo, std::allocator<const CommitSopInfo> > >& < __gnu_cxx::__normal_iterator<const CommitSopInfo*, std::vector<CommitSopInfo, std::allocator<CommitSopInfo> > >' operator make: *** [commit.o] Error 1 And the code it complains about is this: static void findSopInstances( IDB_HANDLE** idbHandle, const std::vector<CommitSopInfo>& rSopList, std::vector<CommitSopInfo>& rSuccesses, std::vector<CommitSopInfo>& rFailures ) { static const char * func = "(findSopInstances)"; // Iterate over list of SOP instances for ( std::vector<const CommitSopInfo>::iterator pos = rSopList.begin(); pos < rSopList.end(); pos++ ) { // Test whether we have this instance if ( haveSopInstance( idbHandle, pos ) ) { ...... Where haveSopInstance is defined as: static bool haveSopInstance( IDB_HANDLE** idbHandle, const CommitSopInfo* pSopInfo ) { now, this used to compile fine under g++2.95.3 with its stdc++. I don't see why it's having difficuly converting the interator to the CommitSopInfo*. Any ideas? thanks a lot, -Anibal

2 1

controlling edge sequence during traversal
by Jakub Plichta 12 Oct '02

12 Oct '02

Hello I am using an adjacency list graph. During a dfs traversal I would like to be control the order of edge traversal. Specifically I have an edge property with an index value and I would like to traverse the higher index edges first. I have tried getting the out_edges iterator and resorting the edges base on their index but that does not seem to make any difference. Any suggestions would be very greatly appreciated. Thanks Jakub Plichta [Non-text portions of this message have been removed]

2 1

RE: [Boost-Users] Re: boost vs. The Lawyers (non-technical)
by Elliot Murphy 11 Oct '02

11 Oct '02

That would be helpful. Some of my colleagues have had terribly frustrating experiences trying to get our lawyers to approve the use of various boost libraries. The lawyers don't seem to like the idea of going into every source file to read a license agreement. -elliot -----Original Message----- From: James Curran/MVP [mailto:jamescurran@mvps.org] Sent: Thursday, October 10, 2002 9:58 AM To: boost-users(a)yahoogroups.com Subject: [Boost-Users] Re: boost vs. The Lawyers (non-technical) "Douglas Gregor" <gregod(a)cs.rpi.edu> wrote in message news:200210090000.23639.gregod@cs.rpi.edu... > Each Boost library has its own license that _must_ meet those requirements. Of > course, that won't satisfy the lawyers, so you'll probably have to send them > the individual library licenses for approval. > > We really could do much better in this area: at the very least, we could > provide a comprehensive list of the licenses for each library. Perhaps we should add to our "standard" a requirement that a printable license page by included with each libraries' documentation (with one or more "fill-in-the-blank" templates/samples available on www.boost.org) -- Truth, James Curran www.NovelTheory.com (Personal) www.NJTheater.com (Professional) www.aurora-inc.com (Day job) Yahoo! Groups Sponsor ADVERTISEMENT 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 the Yahoo! Terms of Service.

4 5

RE: [Boost-Users] Problem with boost::any.
by Jim.Hyslop 11 Oct '02

11 Oct '02

Luc Bergeron [mailto:bluc@videotron.ca] wrote: > Greets, > > I'm trying to use the boost::any component but i'm having > problem with > it !!! I'm using it because I need a way to store parameters of > different type. That real fine but I can't retrieve the value > from the > boost::any. This is what i've tried : > > std::string GetStringFromAny(const boost::any& val) const > { > std::string strVal; > try > { > strVal = boost::any_cast<std::string>(val); > } > catch(boost::bad_any_cast &err) > { > std::cerr << err.what() << std::endl; > return ""; > } > return strVal; > } > > But I always have a bad_any_cast exception. From the > documentation, at > leats what I understand of it, "If passed a value or reference, it > returns a copy of the value content if successful". Can > someone point me > in the right direction as to how can I extract the value within a > boost::any type, I think i'm lost !!! :) The following program works for me, without any error (using MSVC6): #include <boost/any.hpp> #include <string> #include <iostream> std::string GetStringFromAny(const boost::any& val) { std::string strVal; try { strVal = boost::any_cast<std::string>(val); } catch(boost::bad_any_cast &err) { std::cerr << err.what() << std::endl; return ""; } return strVal; } int main() { boost::any anAny( std::string("Hello, world") ); std::string converted = GetStringFromAny( anAny ); std::cout << "Converted: [" << converted << "]" << std::endl; } The output, as I expected, was: Converted: [Hello, world] Note that if I add more to main: int main() { boost::any anAny( std::string("Hello, world") ); std::string converted = GetStringFromAny( anAny ); std::cout << "Converted: [" << converted << "]" << std::endl; anAny = 23; converted = GetStringFromAny( anAny ); std::cout << "Converted: [" << converted << "]" << std::endl; } the second conversion fails. The conversion will only succeed if the type that boost::any currently holds can be implicitly converted into the type you want to retrieve. If you want to convert from (for example) an int to a string, you'll need boost::lexical_cast. -- Jim -- Jim

1 0

Problem with boost::any.
by Luc Bergeron 11 Oct '02

11 Oct '02

Greets, I'm trying to use the boost::any component but i'm having problem with it !!! I'm using it because I need a way to store parameters of different type. That real fine but I can't retrieve the value from the boost::any. This is what i've tried : std::string GetStringFromAny(const boost::any& val) const { std::string strVal; try { strVal = boost::any_cast<std::string>(val); } catch(boost::bad_any_cast &err) { std::cerr << err.what() << std::endl; return ""; } return strVal; } But I always have a bad_any_cast exception. From the documentation, at leats what I understand of it, "If passed a value or reference, it returns a copy of the value content if successful". Can someone point me in the right direction as to how can I extract the value within a boost::any type, I think i'm lost !!! :) Thanks in advance for any help. Luc.

2 1

Possible bug in boost_1_28
by Christopher Henrich 11 Oct '02

11 Oct '02

This bug comes up when I try to include <boost/array.hpp> in certain CodeWarrior projects. (CodeWarrior 8.0, running under Mac OS 10.1.5 on G3 "beige tower") If I initialize the project with stationery "C++ Console Mach-O", and I change HelloWorld.cp to include <boost/array.hpp>, then I get this error message: Error: undefined identifier 'ptrdiff_t' suffix.hpp line 204 included from config.hpp: 57 array.hpp:27 HelloWorld.cp: 10 If I initialize with PowerPlant stationery, I do not have this problem. Chris Henrich

6 6

BOOST regex problem using Libarys with Borland Builder 4
by Alastair 11 Oct '02

11 Oct '02

Hi I have just found out about the boost library's so I am new to them. I am trying to use the regex within a VCL Borland Builder 4. I have followed the installation. and tried a given example. I found that the example works well for a console program but for a windows program it seems to crash with the message. EAccessViolation with message ŽAccess violation at address 0040987B. Read address 00000032. I have compiled both within the Borland IDE and I believe that I don't have to do anything and that the IDE and Library will automatically configure itself. What am I doing wrong? Thanks in advance Alastair

2 1

Thread Lib Build errors in xtime.cpp
by Glenn Thorne 10 Oct '02

10 Oct '02

I'm getting the following errors when trying to build Boost using the Intel C++ v6.0 compiler. The build completes as expected with msvc. Is there a problem with boost::uint64_t in this configuration or am I missing a setting somewhere? Any help will be appreciated. Thanks! Glenn ******** C:\boost_1_28_0>set INTELC="C:\Program Files\Intel\Compiler60\IA32" C:\boost_1_28_0>bjam "-sTOOLS=intel-win32" ...found 330 targets... ...updating 12 targets... vc-C++ libs\thread\build\bin\libboost_thread.lib\intel-win32\debug\runtime-link- static\threading-multi\xtime.obj xtime.cpp libs\thread\build\../src\xtime.cpp(80): error: namespace "boost" has no member " uint64_t" const boost::uint64_t TIMESPEC_TO_FILETIME_OFFSET = ((boost::uint64_t) 27111902UL << 32) + (boost::uint64_t)3577643008UL; ^ libs\thread\build\../src\xtime.cpp(80): error: namespace "boost" has no member " uint64_t" const boost::uint64_t TIMESPEC_TO_FILETIME_OFFSET = ((boost::uint64_t) 27111902UL << 32) + (boost::uint64_t)3577643008UL; ^ libs\thread\build\../src\xtime.cpp(80): error: expected a ")" const boost::uint64_t TIMESPEC_TO_FILETIME_OFFSET = ((boost::uint64_t) 27111902UL << 32) + (boost::uint64_t)3577643008UL; ^ libs\thread\build\../src\xtime.cpp(80): error: namespace "boost" has no member " uint64_t" const boost::uint64_t TIMESPEC_TO_FILETIME_OFFSET = ((boost::uint64_t) 27111902UL << 32) + (boost::uint64_t)3577643008UL; ^ libs\thread\build\../src\xtime.cpp(80): error: expected a ";" const boost::uint64_t TIMESPEC_TO_FILETIME_OFFSET = ((boost::uint64_t) 27111902UL << 32) + (boost::uint64_t)3577643008UL; ^ compilation aborted for libs\thread\build\../src\xtime.cpp (code 2) "C:\Program Files\Intel\Compiler60\IA32/bin/icl" /Zm400 -nologo -GX -c /Z 7 /Od /Ob0 /GX /GR MTd -I"libs\thread\build" -I"C:\boost_1_28_0" -I"C:\PROGR A~1\MICROS~3\VC98\include" -I"../include" -Fo"libs\thread\build\bin\libboo st_t hread.lib\intel-win32\debug\runtime-link-static\threading-multi\xtime.obj" -Tp" libs\thread\build\../src\xtime.cpp" ...failed vc-C++ libs\thread\build\bin\libboost_thread.lib\intel-win32\debug\run time-link-static\threading-multi\xtime.obj... ...skipped <libs!thread!build\libboost_thread.lib\intel-win32\debug\runtime-link -static\threading-multi>libboost_thread.CMD for lack of <libs!thread!build\libbo ost_thread.lib\intel-win32\debug\runtime-link-static\threading-multi>xtime.o bj.. . ...skipped <libs!thread!build\libboost_thread.lib\intel-win32\debug\runtime-link -static\threading-multi>libboost_thread.lib for lack of <libs!thread!build\libbo ost_thread.lib\intel-win32\debug\runtime-link-static\threading-multi>xtime.o bj.. . ********

1 0

CVS fails to build
by Björn Lindberg 10 Oct '02

10 Oct '02

Boost CVS isn't working for me right now. When I try to build programs against it I get /home/compbio/d95-bli/include/boost/preprocessor/cat.hpp:18:49: boost/preprocessor/config/config.hpp: Filen eller katalogen finns inte /home/compbio/d95-bli/include/boost/preprocessor/cat.hpp:22:28: missing binary operator before '(' /home/compbio/d95-bli/include/boost/preprocessor/cat.hpp:29:28: missing binary operator before '(' ... Is it (a) a bug, or (b) me doing something wrong, or (c) something else? I am running g++-3.2 on GNU/Linux. Björn

1 0

boost vs. The Lawyers (non-technical)
by Brian Neal 10 Oct '02

10 Oct '02

I hope this is appropriate for this mailing list... I am a software engineer and our development team loves boost. However, I can already see a coming battle with our company lawyers trying to convince them to let us use boost in our software. We informally talked with one of our software quality engineers. He said their major concern would be assuring that we had permission to use the boost software. In other words, how would we know that the boost developers had permission to release their libraries so that anyone could use it? He also thought we would need something in writing to this effect. Since their isn't any official boost organization, I am getting scared already. I am not a lawyer, nor do I play one on TV, but I would like to hear from any boost developer or user who has experiences in dealing with this nonsense. 1) How did you get your company lawyers off your backs and let you use boost? 2) Should there be (is there?) a central boost license that all boost libraries have that we can point our lawyers at, instead of going to each and every boost developer to get permission to use his/her library? 3) What has been your experience in this situation? Its ironic that something as wonderful as boost, which gives us so much productivity (and that savings is passed along to our customers) could be considered so dangerous by our lawyers. Again, if this is not appropriate or has been rehashed to death I apologize. If that is the case, please email me off list. I'll collect the responses and give them to anyone who is interested. Thanks, Brian Neal

4 3

tuple vs SGI/MIPSPro
by Kevin Wheatley 10 Oct '02

10 Oct '02

Hi, I'm starting to get to grips with boost and in particular on the SGI platform. I was trying to use the boost::tuples and found that with the SGI compiler v7.3.1.3m I was getting the following error CC -LANG:std -I/images/src/boost/from_cvs/boost -I/images/src/boost/from_cvs/boost/boost/compatibility/cpp_c_headers/ -o tupple tupple.cpp cc-1239 CC: ERROR File = /images/src/boost/from_cvs/boost/boost/tuple/tuple_io.hpp, Line = 113 "close" is ambiguous. case close : c = i.widen(')'); break; ^ A template was detected during header processing. instantiation of "char boost::tuples::detail::format_info::get_manipulator(std:: basic_ios<char, std::char_traits<char>> &, boost::tuples::detail::format_info::manipulator_type)" at line 437 instantiation of "std::basic_istream<char, std::char_traits<char>> &boost::tuples::detail::extract_and_check_delimiter(std:: basic_istream<char, std::char_traits<char>> &, boost::tuples::detail::format_info::manipulator_type)" at line 511 instantiation of "std::basic_istream<char, std::char_traits<char>> &boost::tuples::operator>>(std::basic_istream<char, std::char_traits<char>> &, boost::tuples::cons<double, boost::tuples::detail::map_tuple_to_cons<double, double, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type, boost::tuples::null_type>::type> &)" at line 630 of "/usr/include/CC/stl_iterator.h" instantiation of "void std::istream_iterator<density, char, std::char_traits<char>, ptrdiff_t>::_M_read()" at line 604 of "/usr/include/CC/stl_iterator.h" instantiation of "std::istream_iterator<density, char, std::char_traits<char>, ptrdiff_t>::istream_iterator(std::istream_iterator<densit y, char, std::char_traits<char>, ptrdiff_t>::istream_type &)" at line 19 of "tupple.cpp" I'm working with the code from CVS because it compiled better on the SGI, I looked at the code, and basically I think it's getting confused due to my test program including a "using namespace std;" so I modified the tuple_io.hpp to specifically mention detail::format_info::close rather than the bare close and then my example works (I also made explicit the other cases of the switch statement). Is there a more recomended way of doing this? Almost everywhere else in that file the full specification of the enum is given. My code is something like: #include <vector> #include <iostream> #include <algorithm> #include <numeric> #include <iterator> #include "boost/tuple/tuple.hpp" #include "boost/tuple/tuple_comparison.hpp" #include "boost/tuple/tuple_io.hpp" using namespace std; using namespace boost::tuples; typedef tuple<double, double, double> density; int main() { vector<density> v; istream_iterator<density> in(cin); istream_iterator<density> eos; ostream_iterator<density> out(cout); density total; copy(in, eos, back_inserter(v)); copy(v.begin(), v.end(), out); cout << endl; return 0; } Thanks Kevin -- | Kevin Wheatley | These are the opinions of | | Senior Do-er of Technical Things | nobody and are not shared | | Cinesite (Europe) Ltd | by my employers |

1 0

Building regex on AIX w VisualAge
by Tim Daplyn 10 Oct '02

10 Oct '02

Can anyone assist me with this? I've tried every compiler option I can think of combined with disabling BOOST support for a whole bunch of stuff and I still can't get this library to build on AIX. There are just too many compile errors, and I was really hoping I wouldn't have to go modifying the source.

2 1

How to Create Calculated Internal Properties?
by Phillip Ratzloff 10 Oct '02

10 Oct '02

I wish to create an internal calculated property that is derived from existing properties in a vertex or edge. Id also like to be able to access it like any other internal property. For instance, I want a vertex property called weight_sum that would sum the weights for all the out edges and return the answer. I want to be able to use a statement "get(vertex_weight_sum,G,v)" to get the value. Later on, I may decide that I want to cache the value on the vertex (like a normal property) and existing code should work without change. I have been unable to figure out any template class specializations to do this. Additionally, if I try to define the property using the property<> template in the graph declaration it reserves storage for the property, which is undesired. Am I missing something? Is there a way to define a calculated internal property that doesnt reserve storage?

1 0

using weak_ptr and shared_ptr for singleton
by Stephan Born 09 Oct '02

09 Oct '02

Hello! I do some experiments on shared_ptr and weak_ptr. So I tried to use them for a singleton-class, which does not need an explicit singleton::release_instance method. The singleton-instance will be destroyed everytime the last external shared_ptr holding the instance goes out of scope. Is this a valid usage for weak_ptr? Are there any traps I do not see using these smart_ptrs in this context? The drawback of using shared_ptr with singleton is the need for a public destructor. But this is dangerous....it is possible to write singleton::singleton_ptr first_ptr = singleton::instance(); delete first_ptr.get(); Declaring the destructor as private is only possible when defining boost::checked_delete as friend of the singleton-class. ( there is a thread on comp.lang.c++.moderated with the topic 'specialization of function template as friend with MSVC6SP5 not possible' which deals with this problem of friendship... ) But then it is still possible to write the following code singleton::singleton_ptr first_ptr = singleton::instance(); boost:checked_delete( first_ptr.get() ); OK, it's very unlikely to do it by chance, but it is possible. Is there a way to allow destruction of the instance only to shared_ptr? <code> #include <boost/shared_ptr.hpp> #include <boost/weak_ptr.hpp> #include <boost/utility.hpp> class singleton : boost::noncopyable { public: typedef boost::shared_ptr< singleton > singleton_ptr; static singleton_ptr instance() { singleton_ptr tmp; if( s_instance.expired() ) { tmp.reset( new singleton ); s_instance = tmp; } else { tmp = boost::make_shared( s_instance ); } return tmp; } void foo() { // do something i++; } // like to make it private in order to prohibit delete on it, // but boost::shared_ptr<> can't handle it....so sad :( ~singleton(){}; private: typedef boost::weak_ptr< singleton > internal_singleton_ptr; singleton(){}; int i; static internal_singleton_ptr s_instance; }; singleton::internal_singleton_ptr singleton::s_instance; int main() { { singleton::singleton_ptr first_ptr = singleton::instance(); { singleton::singleton_ptr second_ptr = singleton::instance(); second_ptr->foo(); } first_ptr->foo(); } singleton::singleton_ptr another_ptr = singleton::instance(); another_ptr->foo(); return 0; } </code> Regards, Stephan -- --------------------------------------------------------------- Dipl.-Inf. (FH) Stephan Born | beusen Solutions GmbH fon: +49 30 549932-0 | Landsberger Allee 366 fax: +49 30 549932-21 | 12681 Berlin mailto:stephan.born@beusen.de | Germany --------------------------------------------------------------- PGP-Key verfügbar | PGP-Key available ---------------------------------------------------------------

3 6