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

socket library
by Phil Nash 10 May '02

10 May '02

Hi all, Does anyone know what the state of the socket library is? I remember it getting started sometime last year, and there is an implementation in the files section on Yahoo, but that is far from complete and is all I can find. Has it gone off the boil or is someone still looking after it? Many thanks, [)o IhIL..

1 1

Re: [Boost-Users] Re: build setup - what have I missed?
by dick.bridges＠tais.com 10 May '02

10 May '02

Ok. I went back and built bjam and jam from .\tools\build\jam_src. Then I tried: /path-to-jam/jam -sBOOST_ROOT=. sTOOLS=msvc and got: <result> Couldn't find Python 2.1 installation in c:/tools/python skipping Boost.Python library build You can configure the location of your python installation, by setting: PYTHON_ROOT - currently "c:/tools/python" PYTHON_VERSION - currently "2.1" The following are automatically configured from PYTHON_ROOT if not otherwise set PYTHON_INCLUDES - path to Python #include directories; currently PYTHON_LIB_PATH - path to Python library; currently "c:/tools/python/libs" "c:/tools/python/PCBuild" PYTHON_STDLIB_PATH - path to Python standard library modules; currently "c:/tools/python/Lib" D:\dev\boost\tools\build\allyourbase.jam:609: in C++ warning: unknown rule C++-action D:\dev\boost\tools\build\allyourbase.jam:1065: in Object D:\dev\boost\tools\build\allyourbase.jam:1122: in Objects <snip> D:\dev\boost\tools\build\boost-base.jam:1751: in lib libs\thread\build\Jamfile:59: in load-jamfiles D:\dev\boost\tools\build\allyourbase.jam:1580: in SubInclude D:\dev\boost\tools\build\allyourbase.jam:1178: in subinclude Jamfile:3: in load-jamfiles D:\dev\boost\tools\build\bootstrap.jam:15: in boost-build D:\dev\boost\boost-build.jam:2: in module scope ...found 347 targets... ...updating 12 targets... File libs\regex\build\bin-stage\libboost_regex_debug.lib The system cannot find the file specified. copy libs\regex\build\bin\libboost_regex.lib\mvcs\debug\libboost_regex.lib libs\regex\build\bin-stage\libboost_regex_debug.lib ...failed File libs\regex\build\bin-stage\libboost_regex_debug.lib... File libs\regex\build\bin-stage\libboost_regex.lib The system cannot find the file specified. <snip> </result> Should I just wait for the 1.28.0 pre-build Jam executables (and associated instructions) to be posted? "David Abrahams" <david.abrahams@ To: boost-users(a)yahoogroups.com rcn.com> cc: Sent by: news Subject: [Boost-Users] Re: build setup - what have I missed? <news(a)main.gmane .org> 05/08/2002 09:29 PM Please respond to Boost-Users <dick.bridges(a)tais.com> wrote in message news:OFC855DCA5.CAFBF8DF-ON88256BB3.007925A3@tais.net... > This worked once. Then the [yeah, it wasn't backed up] disk drive died. > Target system is Win2000, SP5 and MS Visual C++ 6.0, SP5. <snip> > I've clearly screwed something up but I can't figure out what. Help would > be appreciated. This might hold your answer: http://aspn.activestate.com/ASPN/Mail/Message/boost/1193795 -Dave P.S. Toon, did you post this one to Boost-user? 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

Error report!!
by Ulises Herrera 09 May '02

09 May '02

The file... boost/random/uniform_smallint.hpp need... #include <boost/static_assert.hpp> when boost is compiled with g++-3.0 and a program is linked with it. -- _______________________________________________ Sign-up for your own FREE Personalized E-mail at Mail.com http://www.mail.com/?sr=signup

2 1

building Boost using Jam
by jd.schmidt＠juno.com 09 May '02

09 May '02

Hi, I'm new to using Boost. I've been trying to build Boost using Jam on Windows 2000. My compiler is MSVC6 SP5. As the documentation (www.boost.org/tools/build/index.html) says, I went to my Boost root directory (C:\boost_1_27_0 for me) and used the command jam -sBOOST_ROOT=. -sTOOLS="msvc" and I get the message: 'jam' is not recognized as an internal or external command, operable program, or batch file OK, fine, jam.exe is in C:\boost_1_27_0\boost-build; that's just where I downloaded it to. I've tried to run Jam from the boost-build directory, but I get the message: Jamfile: No such file or directory ...found 8 targets... This doesn't seem very straightforward. What do I need to do here so that I can build Boost? ________________________________________________________________ GET INTERNET ACCESS FROM JUNO! Juno offers FREE or PREMIUM Internet access for less! Join Juno today! For your FREE software, visit: http://dl.www.juno.com/get/web/.

2 1

build setup - what have I missed?
by dick.bridges＠tais.com 09 May '02

09 May '02

This worked once. Then the [yeah, it wasn't backed up] disk drive died. Target system is Win2000, SP5 and MS Visual C++ 6.0, SP5. Downloaded boost-build.win.zip and extracted it to c:\bin\boost-build. Added c:\bin\boost-build to the path. Set environment variables: BOOST_BUILD=c:\bin\boost-build JAM_TOOLSET=VISUALC VISUALC="c:\Program Files\Microsoft Visual Studio\VC98\bin" Checked out from CVS to d:\dev cd'd to d:\dev\boost and executed: %BOOST_BUILD%\jam -sBOOST_ROOT=. -sTOOLS=msvc <result> warning: unknown rule boost-build don't know how to make all ...found 1 target... ...can't find 1 target... </result> I've clearly screwed something up but I can't figure out what. Help would be appreciated. TIA (from the eternal newbie)

2 1

Unable to compile Boost Graph (BGL) inVisual Studio 7 (.NET)
by dmarsh26us 08 May '02

08 May '02

Have just bought the book and downloaded the source it all looks cool.... BUT Can't build a thing! Managed to get hold of AllExamples.dsw eventually converted it to a new VC7 solution ok (few errors on some projects). But no builds, cannot build any of my own code with BGL either. HELP. This is the jist of it:- ------ Build started: Project: dfs_example, Configuration: Debug Win32 ------ Compiling... dfs-example.cpp ../boost_1_25_1\boost\type_traits\cv_traits.hpp(188) : fatal error C1001: INTERNAL COMPILER ERROR (compiler file 'msc1.cpp', line 2844) Please choose the Technical Support command on the Visual C++ Help menu, or open the Technical Support help file for more information Build log was saved at "file://c:\BGL\VC++ Projects\Debug\BuildLog.htm" dfs_example - 1 error(s), 0 warning(s) ---------------------- Done ---------------------- Build: 0 succeeded, 1 failed, 0 skipped The cv traits class seems to trigger the compiler to crash horribly. I don't even know what this class does yet. (only got to page 55 of the book!) Anyone know a workaround? Anyone tried VC7 with boost yet? Is there a microsoft patch yet? Can't go back to VC6 as am using other new libraries. Thanks David Marsh

2 2

Help me pay the piper?
by dick.bridges＠tais.com 08 May '02

08 May '02

Now that I've used all this swell stuph, it's time to pay the piper: turn it over to QC and get it into "The Build". Unfortunately the QC people would like some instructions that describe which version of boost-build I'm using; how to install it; which version of the libs I'm using; etc. <some background> It's been a couple of months since I started with the boost libs, and I don't remember all the gyrations I went through to get it working and I'm POSITIVE I did a LOT more than is required. I remember that, after downloading and extracting boost_1_27_0.tar.gz, I had to get the current version from CVS before I could get a clean compile. (I tried turning over copies of my disk drives to them but it turns out that is not considered acceptable documentation. %>[ ) I've been following the jamboost and Boost-Users mailing lists so I know there've been quite a few changes - especially to boost-build. After building a "clean" testbed machine, I started by downloading boost-build.win.zip and following the instructions in the readme.html. Well, that didn't work. I can do a hand job on the path but I'm not sure what environment variables are actually required. Current plan is to pull the library source directly from CVS. Problem is, I don't know what version to tell QC I'm using and, unless all the folks at boost.org would agree not to add anything to CVS for a while, I SURE can't tell them how to re-create what I did. ;) </some background> I can always take what I've got; check it in and call it a day but that's not the path to getting boost integrated into the development environment around here. (Hey, I'm trying to be an evangelist here.) Anyone by any chance working on some updated instructions and/or docs that maybe aren't ready for prime time but would be willing to share? TIA PS: Current target systems are Win2000, SP2 and Visual C++ 6.0, SP5 and vanilla Red Had Linux 7.2.

2 1

voronoi diagram and BGL
by Louis Lavery 08 May '02

08 May '02

Hi, Is it possible to order the edges of a vertex of a boost graph? >From what I can glean from the BGL documentation, I guess the answer will be "No, but may be some time in the future". I so, how soon in the future? Thanks, Louis.

5 5

Solaris jam build
by cs_jritchie 08 May '02

08 May '02

I am receiving the following on a Solaris system when using: jam -sBOOST_ROOT=. -sTOOLS=gcc status/Jamfile:53: in failed-test-file warning: unknown rule FAIL_EXPECTED status/Jamfile:100: in fail-to-build ./tools/build/boost-base.jam:1407: in subvariant-target ./tools/build/boost-base.jam:1448: in main-target ./tools/build/boost-base.jam:1065: in declare-local-target status/Jamfile:35: in boost-test status/Jamfile:119: in compile-fail status/Jamfile:368: in SubInclude ./tools/build/allyourbase.jam:1141: in subinclude Jamfile:5: in module scope {...above repeated many times...} Unable to load Boost.Build, it seems that the build system files were located, but the build system is not defined. Attempted to use, as the bootstrap, this file: ./tools/build/boost-build.jam. This means that even though that file was loaded it either failed to directly load the build system files, or it failed to specify where to find and load the files by calling the 'build-system [path] ;' rule. Please consult the documentation at 'http://www.boost.org'. What am I doing wrong? Any help is appreciated.

2 4

Re: Problems building Python lib
by pendrildurham 08 May '02

08 May '02

Sorry -- Newbie here -- I don't know exactly what the 'latest CVS state' is or how to get it. I have been trying to build boost-python on Windows95 with MSVC++6. I have boost 1.27 and activestate python 2.2.1 build 222. I can get it to compile but the link fails. Here is the linker message for the boost-example1: Linking... Creating library Debug/hello.lib and object Debug/hello.exp example1.obj : error LNK2001: unresolved external symbol "bool __cdecl boost::python::handle_exception_impl(class boost::function0<void,struct boost::empty_function_policy,struct boost::empty_function_mixin,int>)" (?handle_exception_impl@python@boos t@@YA_NV?$function0@XUempty_function_policy@boost@@Uempty_function_mix in@2@H@2@@Z) bpl_static.lib(types.obj) : error LNK2001: unresolved external symbol "bool __cdecl boost::python::handle_exception_impl(class boost::function0<void,struct boost::empty_function_policy,struct boost::empty_function_mixin,int>)" (?handle_exception_imp l@python@boost@@YA_NV?$function0@XUempty_function_policy@boost@@Uempty _function_mixin@2@H@2@@Z) bpl_static.lib(extension_class.obj) : error LNK2001: unresolved external symbol "bool __cdecl boost::python::handle_exception_impl(class boost::function0<void,struct boost::empty_function_policy,struct boost::empty_function_mixin,int>)" (?handle_exc eption_impl@python@boost@@YA_NV?$function0@XUempty_function_policy@boo st@@Uempty_function_mixin@2@H@2@@Z) bpl_static.lib(extension_class.obj) : error LNK2001: unresolved external symbol "struct _object * __cdecl boost::python::expect_non_null(struct _object *)" (?expect_non_null@python@boost@@YAPAU_object@@PAU3@@Z) bpl_static.lib(functions.obj) : error LNK2001: unresolved external symbol "struct _object * __cdecl boost::python::expect_non_null(struct _object *)" (?expect_non_null@python@boost@@YAPAU_object@@PAU3@@Z) bpl_static.lib(objects.obj) : error LNK2001: unresolved external symbol "struct _object * __cdecl boost::python::expect_non_null(struct _object *)" (?expect_non_null@python@boost@@YAPAU_object@@PAU3@@Z) example1.obj : error LNK2001: unresolved external symbol "struct _object * __cdecl boost::python::expect_non_null(struct _object *)" (?expect_non_null@python@boost@@YAPAU_object@@PAU3@@Z) bpl_static.lib(module_builder.obj) : error LNK2001: unresolved external symbol "struct _object * __cdecl boost::python::expect_non_null(struct _object *)" (?expect_non_null@python@boost@@YAPAU_object@@PAU3@@Z) bpl_static.lib(init_function.obj) : error LNK2001: unresolved external symbol "struct _object * __cdecl boost::python::expect_non_null(struct _object *)" (?expect_non_null@python@boost@@YAPAU_object@@PAU3@@Z) bpl_static.lib(classes.obj) : error LNK2001: unresolved external symbol "struct _object * __cdecl boost::python::expect_non_null(struct _object *)" (?expect_non_null@python@boost@@YAPAU_object@@PAU3@@Z) Debug/hello.dll : fatal error LNK1120: 2 unresolved externals Error executing link.exe. hello.dll - 11 error(s), 0 warning(s) Perhaps there is something simple I can do to fix this? Perhaps there is something I need to include in the project to get the link to work? Can anyone help? Thanks --- In Boost-Users@y..., Albrecht Fritzsche <albrecht.fritzsche@t...> wrote: > David Abrahams wrote: > > >1. Get the latest CVS state. Release 1.27 is broken in several ways. > > > Ooops, ok - I will try the CVS version. > > >The library works with Python 2.2.1 > > > Thanks - that's the info I was hoping for :-) > > Ali

2 1

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