May 2002 - Boost-users - lists.preview.boost.org

creating a shared_ptr<T> from a weak_ptr<T>
by Toon Knapen 08 May '02

08 May '02

When creating a shared_ptr<T> from a weak_ptr<T>, I want the shared_ptr<T> to use the same count as the shared_ptr the weak_ptr was constructed from. I could create a new shared_ptr, based on the pointer I can get() from the weak_ptr. But in this case, this shared_ptr will keep its own count. Let me fill you in on my use-case. An object creates some big data-structure on request of its clients, taking into account some options specified by the client while making the request. Now I'd like the server-object to keep a weak_ptr to these big data-structure to provide quickly the same result once another client performs an identical request. The server should however not keep a shared_ptr because otherwise all datastructure will stay in memory. And therefor I need to create a shared_ptr from a weak_ptr. thanks, toon

6 7

Re: [Boost-Users] AllExamples.dsw from BGL Errata Web-site
by Lie-Quan Lee 07 May '02

07 May '02

Hi Jon, Thanks to remind me that attachment doesn't work for yahoogroups. I put the file at http://groups.yahoo.com/group/boost/files/BGLVCprojects.tgz Just in case, it is also available at: http://www.osl.iu.edu/~llee/BGLVCprojects.tgz Jeremy Siek will update the file on BGL book errata page. On Tue, 2002-05-07 at 11:43, Jon Kalb wrote: > >Hi Richard, > > > >It looks like the BGLVCprojects.tgz online is corrupted somehow. I > >attached the right file (about 466.1K) in this message. Please extracts > >the contents of the tarball using winzip 8.0 or up. The tarball includes > >the missing workspace file and project files. > > > >Thanks for reporting that problem! > >-- > >Lie-Quan Lee (AKA: Rich Lee) > >Research Associate http://www.osl.iu.edu/~llee > >Computer Science Department llee(a)cs.indiana.edu > >Indiana University at Bloomington 812-855-3608(Office) > > > >[Non-text portions of this message have been removed] > > Rich, > > As you can see from the line inserted by Yahoo! Groups in your > message, Boost-Users is configured to not allow attachments. If you > wouldn't mind uploading the attachment to the group's shared file > area you can send a link to it. Or if you want to send the file to > me, I'll take care of it. > > Thanks. > > -- > Jon Kalb > Kalb(a)LibertySoft.com -- Lie-Quan Lee (AKA: Rich Lee) Research Associate http://www.osl.iu.edu/~llee Computer Science Department llee(a)cs.indiana.edu Indiana University at Bloomington 812-855-3608(Office)

1 0

AllExamples.dsw from BGL Errata Web-site
by loopylozzysdad 07 May '02

07 May '02

Hi, I am still having no luck with this file. In the Boost Graph Library book (Great book!), Errata web page (http://siek.info/bgl.html) it says... "On the accompanying CD, the file \BGL\VC++ Projects\AllExamples.dsw is missing. You can download the missing file from here BGLVCprojects.tgz." When I download it I get a 454K file. When I open it with WinZip winzip tells me there is a single file BGLVCprojects.tgz in it. when it opens that single file it only finds ONE file in that accum_compile_times.dsw size 4639 bytes. this does not make sense fo r a file that started at 454K but I don't seem able to get any more out of it. I am using WinZip 6.2. Another machine with WinZIP 8.0 reports 'Invalid compressed header after 1 entry'. Is anyone else getting this result. If not could you email me a file that works for you? All the best and thanks - Richard Howells

2 1

Association Library?
by rzako23 07 May '02

07 May '02

Are there any generic components in Boost (or elsewhere) for *directly* representing associations between classes? Generalization, composition, aggregation and association are major ideas in the object model. Generalization can be implemented fairly directly in C++ using inheritance. Composition and aggregation can be represented directly in C++ using data members (when the number of parts in each whole is small and fixed), or using STL containers (when the number of parts in each whole is large or variable). But Standard C++ (including the STL) provides no *direct* support for representing associations, i.e., collections of links between objects in two or more classes. (David Papurt makes this point in "Inside the Object Model: The Sensible Use of C++.") For example, the STL does not offer a way to *directly* represent the 1-to-many "Supervises" association from the Manager class to the Employee class. One would like to have links in both directions: forward links forthe employees a manager supervises and a back link to the manager each employee is supervised by. Of course, one could code this all "by hand" using a container of pointers to employees for each manager and a back pointer for each employee. But the implementation details really have little to do with managers and employees and would be pretty much the same for parents and children or teachers and students. Moreover, while not difficult, managing the forward and back links correctly (and in an exception-safe manner) is tedious. It would be nice to have templates that can be instantiated to implement specific associations. In brief, STL containers directly support composition and aggregation but not association. I am new to the Boost library, but was not able to find this capability in Boost. If an association library is available in Boost (or elsewhere), I would be happy to know so. If not, is an association library something that would be worth looking into? Thanks, Rob Zako

11 24

Problems building Python lib
by Albrecht Fritzsche 03 May '02

03 May '02

Hi, since I've read in the documentation "Boost.Python is known to have been tested in the following configurations: Against Python 2.0 using the following compiler/library combinations: MSVC++6sp4 with the native library. ... Against Python 1.5.2 using the following compiler/library: MSVC++6sp4 MSVC++6sp4/STLport 4.0" and since I am using the MSVC++6sp4/STLport 4.0 combination I deinstalled Python 2.2 (for which the build failed) and installed Python 1.5.2. Running G:\Environment\boost_1_27_0>jam -sBOOST_ROOT=. -sTOOLS="msvc-stlport" I get the following errors: don't know how to make <python-library@>python21.lib ...patience... ...patience... ...found 2247 targets... ...using 2 temp targets... ...updating 132 targets... ...can't find 1 target... ...can't make 42 targets... msvc-C++-action libs\python\build\bin\libboost_python_static\msvc-stlport\debug\ runtime-link-dynamic\stlport-iostream-off\stlport-version-4.0\types.obj types.cpp .\boost/python/detail/wrap_python.hpp(24) : fatal error C1083: Include-Datei kann nicht geoeffnet werden: 'patchlevel.h': No such file or directory cl /Zm400 -nologo -GX -c -DBOOST_PYTHON_STATIC_LIB=1 -DBOOST_PYTHON_DYNAMI C_LIB -D__STL_NO_SGI_IOSTREAMS=1 -D__STL_DEBUG=1 -D__STL_DEBUG_UNINITIALIZED=1 /Zi /Od /Ob0 /GX /GR /MDd -I"libs\python\build" -I"." -I"c:/tools/python\inclu de" -I"c:/tools/python\PC" -I"\stlport" -I"D:\PROGRA~1\MICROS~1.0\VC98\include" -Fo"libs\python\build\bin\libboost_python_static\msvc-stlport\debug\runtime-li nk-dynamic\stlport-iostream-off\stlport-version-4.0\types.obj" -Tp"libs\python\ build\../src\types.cpp" ...failed msvc-C++-action libs\python\build\bin\libboost_python_static\msvc-stlp ort\debug\runtime-link-dynamic\stlport-iostream-off\stlport-version-4.0\types.ob j ... msvc-C++-action libs\python\build\bin\libboost_python_static\msvc-stlport\debug\ runtime-link-dynamic\stlport-iostream-off\stlport-version-4.0\classes.obj classes.cpp .\boost/python/detail/wrap_python.hpp(24) : fatal error C1083: Include-Datei kan n nicht geoeffnet werden: 'patchlevel.h': No such file or directory ............ [The obvious translation of this error message is: Cannot open include file 'patchlevel.h'.] What am I doing wrong? By the way, all of the other libs including tests are compiling with Jam without any problems at all. And, as a second quesion, why is there no MSVC++6sp4/STLport 4.0 support for the newer Python versions - by the way, there is already a Python v2.2. Thanks for any hints and help Ali *

5 7

Re: [boost] boost organization and installation
by William E. Kempf 02 May '02

02 May '02

----- Original Message ----- From: "David Abrahams" <david.abrahams(a)rcn.com> > > ----- Original Message ----- > From: "William E. Kempf" <williamkempf(a)hotmail.com> > > OK, let me make this a bit more of an official request. If anyone is > > interested in volunteering for this effort send me a private e-mail. > Once I > > have, let's say at least 5, committed volunteers I'll start the effort > of > > creating and coordinating the needed community. > > I don't know whether you want to count me as a volunteer or not. I'm > willing to handle things from the Boost.Build end at some workable pace. > I don't think I can invest anything in other systems, except at the > integration boundary (e.g. if autotools builds Jam and calls it to build > boost). I'd hope that every developer with a current Boost library would be willing to volunteer to at least help with input here, and possibly to committing to some (minor) changes to procedures and/or code. I'm also glad you're volunteering to help from the Boost.Build end of things. But I won't count you as a volunteer for the actual effort here, since I know how busy you are. What I think we really need are volunteers from the user community, though I won't turn down any current Boost developers who think they have the time and experience to help out here. In any event, since I hope that we get volunteers from the user community, I'm CC:ing this to the Boost Users list. Anyone interested in helping out in this effort should e-mail me privately. Bill Kempf

2 1

RE: [Boost-Users] MPL ?
by Aleksey Gurtovoy 01 May '02

01 May '02

Toon Knapen wrote: > Inside a templated class, I need to provide two different > implementations for > a function based on the compile-time decision > ::boost::is_convertible< value_type,double >::value. > How can I do this using MPL ? 1) A simple way: template< typename T > class my { public: typedef typename T::value_type value_type; void foo() { typedef boost::is_convertible< value_type,double > c; do_foo(mpl::bool_c<c::value>()); } private: void do_foo(mpl::true_c) { /*...*/ } void do_foo(mpl::false_c) { /*...*/ } }; Andrei has written an article that covers this technique - http://www.cuj.com/experts/1810/alexandr.htm. 2) A sophisticated one (becomes more relevant when you have several "foo"'s to handle): template< typename T, typename Derived > struct foo_impl { void foo() { Derived& self = static_cast<Derived&>(*this); // ... } }; template< typename Derived > struct foo_double_impl { void foo() { Derived& self = static_cast<Derived&>(*this); // ... } }; template< typename T > class my : public mpl::select_if< boost::is_convertible< typename T::value_type,double > , foo_impl<T, my> , foo_double_impl<T> >::type { // ... private: friend class foo_impl<T, my>; friend class foo_double_impl<T>; // ... }; HTH, Aleksey

1 0

MPL ?
by Toon Knapen 01 May '02

01 May '02

Inside a templated class, I need to provide two different implementations for a function based on the compile-time decision ::boost::is_convertible< value_type,double >::value. How can I do this using MPL ? thanks, toon

1 0

RE: [Boost-Users] shared_ptr and incomplete types
by bjorn.karlsson＠readsoft.com 01 May '02

01 May '02

> From: Jim Hyslop [mailto:jim.hyslop@leitch.com] > > I notice that the implementation available at boost.org does not allow > the single-argument ctor to be called on an incomplete type, e.g.: [snip] > The documentation on the single-argument ctor, though, does not state > that T must be a complete type. Is the implementation > stricter than the > documentation, or is the documentation incomplete? I believe that the implementation is correct, and that belief is confirmed by the latest documentation in the CVS - T must indeed be a complete type. Another change is the addition of a default constructor, which does not require T to be complete. > I'm writing an article on the smart pointers for the CUJ, and I'd like Cool! ;-) > to know if I can correctly state "The default behavior for > shared_ptr is > to require complete types. You can, however, manage an incomplete type > if you provide a function or function object, which will be > called when > the owned object is destroyed." Since shared_ptr requires a complete type only for a few operations, such as the constructor that you mentioned, and reset(T * p), I'd rephrase to something more explicit, like: "The instantiation of a shared_ptr typically requires a complete type. [etc]". Bjorn Karlsson [Non-text portions of this message have been removed]

2 1

shared_ptr and incomplete types
by Jim Hyslop 01 May '02

01 May '02

The documentation at http://www.boost.org/libs/smart_ptr/shared_ptr.htm states that "T may be void, but in that case, either an explicit delete function must be passed in, or the pointed-to object must have a trivial destructor." I notice that the implementation available at boost.org does not allow the single-argument ctor to be called on an incomplete type, e.g.: class simple; simple * createSimple(); shared_ptr<simple> simpleFactory() { return shared_ptr<simple>(createSimple()); } The documentation on the single-argument ctor, though, does not state that T must be a complete type. Is the implementation stricter than the documentation, or is the documentation incomplete? I'm writing an article on the smart pointers for the CUJ, and I'd like to know if I can correctly state "The default behavior for shared_ptr is to require complete types. You can, however, manage an incomplete type if you provide a function or function object, which will be called when the owned object is destroyed." -- Jim

3 2