February 2008 - Boost-users - lists.preview.boost.org

Building boost 1.34.1 with MSVC 8 (64-bit)
by Eric Woodruff 25 Feb '08

25 Feb '08

Hello, I hope John Maddock is reading this as he had some knowledge of the Intel compiler problem. Similar to the thread quoted below, I am getting the "Windows cannot open this file" dialog when trying to build boost using VC8 as 64-bit. (Doesn't have /this/ problem when I don't set address-model=64 or use the VC7.1 x86 compiler for 32-bit builds.) This is with boost-jam-3.1.16-1-ntx86. I've tried adding the full path to cl.exe in user-config.jam without success. For historical reasons, I am using an nmake Makefile (under a Cygwin environment) to fit the boost build into an existing build system. It essentially operates as follows: tar jxf boost_1_34_1.tar.bz2 echo using msvc : 8.0 : "C:/WINDDK/3790.1830/bin/win64/x86/amd64/cl.exe" ; > boost_1_34_1\tools\build\v2\user-config.jam echo using stlport : 5.1.5 : c:\cygwin\home\ewoodruff\build\trunk\winnt\amd64\x64-WINNT\TOOLS\usr\include\stlport c:\cygwin\home\ewoodruff\build\trunk\winnt\amd64\x64-WINNT\TOOLS\usr\lib ; >> boost_1_34_1\tools\build\v2\user-config.jam sed -ie 's,\\\\,/,g' boost_1_34_1\tools\build\v2\user-config.jam cd boost_1_34_1 && "c:\cygwin\home\ewoodruff\build\trunk\winnt\amd64\boost\boost-jam-3.1.16-1-ntx86\bjam.exe" "-q" "--toolset=msvc-8.0" "--prefix=c:\cygwin\home\ewoodruff\build\trunk\winnt\amd64\x64-WINNT\TOOLS" "runtime-link=shared" "link=static" "threading=multi" "stdlib=stlport-5.1.5" "address-model=64" "--without-python" "--without-wave" stage install Once the first directory leaf is created is when the .rcp file tries to get run. Here's the content of that file: $ cat bin.v2/libs/serialization/build/msvc-8.0/debug/address-model-64/link-static/stdlib-stlport-5.1.5/stdlib-stlport-5.1.5/threading-multi/basic_archive.obj.rsp "libs\serialization\src\basic_archive.cpp" -Fo"bin.v2\libs\serialization\build\msvc-8.0\debug\address-model-64\link-static\stdlib-stlport-5.1.5\stdlib-stlport-5.1.5\threading-multi\basic_archive.obj" -TP /Z7 /Od /Ob0 /W3 /GR /MDd /Zc:forScope /Zc:wchar_t /favor:blend /wd4675 /EHs /Gy -c -DBOOST_ALL_NO_LIB=1 -D_STLP_DEBUG=1 -D_STLP_DEBUG_UNINITIALIZED=1 -D_STLP_USE_DYNAMIC_LIB=1 "-I." "-Ic:\cygwin\home\ewoodruff\build\trunk\winnt\amd64\x64-WINNT\TOOLS\usr\include\stlport" I'm also curious, is stdlib-stlport-5.1.5 supposed to be in the path twice? Thanks in advance, Eric > > > *Subject:* [Boost-users] Building boost 1.34.1 > > > *From:* James Emerton > > > *Date:* 2007-08-31 18:08:46 > > ------------------------------------------------------------------------ > I'm having difficulties building boost 1.34.1 using the intel > toolchain. Once bjam tries to invoke the compiler, I get the "Windows > cannot open this file" dialog, where the file is an .obj.rsp file. > > It would appear that perhaps the compiler itself is not making it into > the command line, thus causing the system to try executing the .rsp > file directly. I am not having this problem if I use the msvc > toolchain. > > > Command to invoke the build: > bjam --toolset=intel-9.1 > Compiler: > Intel(R) C++ Compiler for 32-bit applications, Version 9.1 > Build 20060519Z > bjam: > Boost.Jam Version 3.1.14. OS=NT. > > > Some environment variables: > > ICPP_COMPILER91=C:\Program Files\Intel\Compiler\C++\9.1 > INTEL_BASE_MSVC_TOOLSET=vc-7_1 > INTEL_LICENSE_FILE=C:\Program Files\Common Files\Intel\Licenses > INTEL_PATH=C:\Program Files\Intel\Compiler\C++\9.1\Ia32 > INTEL_SHARED=C:\Program Files\Common Files\Intel\Shared Files > > I've run the iclvars batch script, icl.exe is on the path, and I've > used this environment to build boost 1.33 successfully. I've tried > various debug levels of bjam, and I haven't seen anything that makes a > problem obvious to me. > > Thanks > James > _______________________________________________ > Boost-users mailing list > Boost-<email removed> > http://lists.boost.org/mailman/listinfo.cgi/boost-users > > ------------------------------------------------------------------------ > > > *From:* John Maddock > > > *Date:* 2007-08-31 19:17:38 > > ------------------------------------------------------------------------ > James Emerton wrote: > /> I'm having difficulties building boost 1.34.1 using the intel/ > /> toolchain. Once bjam tries to invoke the compiler, I get the "Windows/ > /> cannot open this file" dialog, where the file is an .obj.rsp file./ > />/ > /> It would appear that perhaps the compiler itself is not making it into/ > /> the command line, thus causing the system to try executing the .rsp/ > /> file directly. I am not having this problem if I use the msvc/ > /> toolchain./ > > Yep, it's bug report already: http://zigzag.cs.msu.su/boost.build/ticket/137 > > It does work if you add an entry for intel to a user-config.jam in your HOME > directory, mine looks like: > > using intel-win : 9.1 : "c:/program > files/intel/compiler/c++/9.1/ia32/bin/icl.exe" ; > using intel-win : 10.0 : "c:/program > files/intel/compiler/c++/10.0.025/ia32/bin/icl.exe" ; > using intel-win : 8.1 : "c:/program > files/intel/cpp/compiler80/ia32/bin/icl.exe" ; > #using intel-win : 7 : "c:/program > files/intel/cpp/compiler70/ia32/bin/icl.exe" ; > #using intel-win : 6 : "c:/program > files/intel/cpp/compiler60/ia32/bin/icl.exe" ; > > See the Boost.Build docs for more info on setting this up. And of course in > most cases you can always build the source from your IDE should you prefer > :-) > > HTH, John. > > _______________________________________________ > Boost-users mailing list > Boost-<email removed> > http://lists.boost.org/mailman/listinfo.cgi/boost-users >

3 4

boost::crc32 - visual studio crt - bug?!
by gfsdgadas＠Safe-mail.net 24 Feb '08

24 Feb '08

I want to compute the Crc32 checksum of a bunch of bytes. To do so, I simply use crc_32_type and process_bytes. Unfortunately this does not work in debug mode. Visual Studio (2008) breaks at line 574 of crc.hpp. // Compare a byte to the remainder's highest byte static unsigned char index( value_type rem, unsigned char x ) { return x ^ rem; } Error message: Run-Time Check Failure #1 - A cast to a smaller data type has caused a loss of data. If this was intentional, you should mask the source of the cast with the appropriate bitmask. For example: char c = (i & 0xFF); Changing the code in this way will not affect the quality of the resulting optimized code. I do not like changing the boost source code, so what can I do to workaround this problem?

2 1

GPGPU
by Carlo Alberto Ferraris 24 Feb '08

24 Feb '08

Just a casual question: has anyone considered the possibility to write and include a GPGPU library in boost? CAF

3 3

[call_traits] call_traits<enum>::param_type is const reference
by Sergei Politov 23 Feb '08

23 Feb '08

As I can see, call_traits for enum types defines param_type as const reference. I cannot find a reason for it. It is possible to use is_enum to determine whether type is enum or not, and define param_type for enums similar to value_type. Could anybody describe the situation?

1 0

Re: [Boost-users] All orderings of boost::graph::topological_sort()
by Matan Nassau 23 Feb '08

23 Feb '08

Excuse me for the double-post: my previous post had a bad subject line... Thank you very much for your prompt reply, Stephen. On Fri, Feb 22, 2008 at 12:51, Stephen Nuchia wrote: > For any significant-sized input the number of orderings will be too > large to do anything useful with. At the moment my input consists of about 32 vertices, although I'd like not to count on it to remain this way. I can probably assume the number of nodes will remain in the order of magnitude of hundreds or less. I'm talking about a schedule generating program for an undergraduate student. It generates a schedule for a semester for a student, given some constraints (time of day, courses done already and so on) and the course dependency (this course must not be taken before that one and so on). I would very much like to support other complexities such as some courses may be taken while a certain dependent is being taken, while others must be taken before any of their dependents may be taken. But I leave such complexities as worries for future versions. For the time being, all I want is to be able to generate all the different schedules (i.e., all various topo sorts). Indeed, something like std::next_permutation() would make a lot of sense. It would be fast and flexible. My next task would then be to find an algorithm to quickly estimate the number of different orderings. :) > For an input small enough that the > set of feasible orderings will fit in memory I would simply generate all > permutations and screen for feasibility rather than attempting to create > an enumeration algorithm. I don't follow. Can I change the ordering of vertices without ruining the graph's integrity? How? Would this affect the toposort output? > Create a comparability matrix and use it to > prune the permutation tree, perhaps? N^2 bits doesn't sound > prohibitive. I don't follow... > This has probably been studied but I'm at work so I'll speculate. For > larger inputs with many constraints it may be worthwhile to create an > enumeration algorithm. > > For purposes of empirical testing of build-ordering systems it might be > sufficient to go into the algorithm and randomize the selection > processes whenever it makes arbitrary choices. This may or may not be > easily achieved by randomly pre-sorting the input, depending on the > internals of the algorithm. If the sort is stable that will work. That is the reason for my query on my OP in the first place. I don't quite understand how manipulate the input vertices of the depth_first_search(). For what I understand, depth_first_search() works on a set of start vertices (which have no incoming edges), and for each of these the algorith dives in to the depth of the vertex's tree. Permuting the order of these vertices would give me advance me a lot, but I don't know how to do that. > If there is enough interest to justify adding an efficient deterministic > enumerator for poset orders I'll write one. Graph-theoretical > algorithms are a fun way to spend the weekend :-) I personally think the topological_sort() algorithm begs for an algorithm that would allow iterating over all unique orderings. As I stated above, I think something of the sort of std::next_permutation() would make a lot of sense, especially because graph algorithms, and finding all different toposorts in particular, tend to be expensive operations. Simply asking for one ordering at a time gives a lot of flexibility to the user with the added benefit of efficiency and integration with the STL's way of doing things. -- Matan

2 1

Using boost::barrier dynamically
by Gonzalo Garramuño 23 Feb '08

23 Feb '08

I'm trying to have several threads synced with a boost::barrier. But I'm finding its design very limiting (or, more likely, my understanding of its use is too poor). I want to have a barrier created dynamically (based on some parameters that determine how many threads run). And I also want to be able to delete the barrier at any point in time (from a thread that does not wait on the barrier). I expect this to force all threads that are waiting on the barrier to continue (ie. send a notify_all). However, boost::barrier's destructor just does nothing, so I get a hang. What's the right way to achieve this? PS. On a separate note... is there any reason why all of boost primitives have their variables private? I really would find it pretty useful to sometimes be able to query, for example, the lock count of some objects. -- Gonzalo Garramuño ggarra(a)advancedsl.com.ar AMD4400 - ASUS48N-E GeForce7300GT Xubuntu Gutsy

1 0

Re: [Boost-users] All orderings of boost::graph::topological_sort()"
by Matan Nassau 23 Feb '08

23 Feb '08

Thank you very much for your prompt reply, Stephen. On Fri, Feb 22, 2008 at 12:51, Stephen Nuchia wrote: > For any significant-sized input the number of orderings will be too > large to do anything useful with. At the moment my input consists of about 32 vertices, although I'd like not to count on it to remain this way. I can probably assume the number of nodes will remain in the order of magnitude of hundreds or less. I'm talking about a schedule generating program for an undergraduate student. It generates a schedule for a semester for a student, given some constraints (time of day, courses done already and so on) and the course dependency (this course must not be taken before that one and so on). I would very much like to support other complexities such as some courses may be taken while a certain dependent is being taken, while others must be taken before any of their dependents may be taken. But I leave such complexities as worries for future versions. For the time being, all I want is to be able to generate all the different schedules (i.e., all various topo sorts). Indeed, something like std::next_permutation() would make a lot of sense. It would be fast and flexible. My next task would then be to find an algorithm to quickly estimate the number of different orderings. :) > For an input small enough that the > set of feasible orderings will fit in memory I would simply generate all > permutations and screen for feasibility rather than attempting to create > an enumeration algorithm. I don't follow. Can I change the ordering of vertices without ruining the graph's integrity? How? Would this affect the toposort output? > Create a comparability matrix and use it to > prune the permutation tree, perhaps? N^2 bits doesn't sound > prohibitive. I don't follow... > This has probably been studied but I'm at work so I'll speculate. For > larger inputs with many constraints it may be worthwhile to create an > enumeration algorithm. > > For purposes of empirical testing of build-ordering systems it might be > sufficient to go into the algorithm and randomize the selection > processes whenever it makes arbitrary choices. This may or may not be > easily achieved by randomly pre-sorting the input, depending on the > internals of the algorithm. If the sort is stable that will work. That is the reason for my query on my OP in the first place. I don't quite understand how manipulate the input vertices of the depth_first_search(). For what I understand, depth_first_search() works on a set of start vertices (which have no incoming edges), and for each of these the algorith dives in to the depth of the vertex's tree. Permuting the order of these vertices would give me advance me a lot, but I don't know how to do that. > If there is enough interest to justify adding an efficient deterministic > enumerator for poset orders I'll write one. Graph-theoretical > algorithms are a fun way to spend the weekend :-) I personally think the topological_sort() algorithm begs for an algorithm that would allow iterating over all unique orderings. As I stated above, I think something of the sort of std::next_permutation() would make a lot of sense, especially because graph algorithms, and finding all different toposorts in particular, tend to be expensive operations. Simply asking for one ordering at a time gives a lot of flexibility to the user with the added benefit of efficiency and integration with the STL's way of doing things. -- Matan

1 0

All orderings of boost::graph::topological_sort()
by Matan Nassau 22 Feb '08

22 Feb '08

Hello, In implementing a dependency graph subsystem, one of the tasks is figuring out all the possible ways of traversing the graph with respect to the dependencies. boost::graph::topological_sort() looks like a reasonable place to start; however, topological_sort() only gives /one/ linear ordering, not all possible orderings. I digged into the code and found that the algorithm is in fact a call to depth_first_search with a custom visitor. The 0-in-degree vertices are not mentioned in the algorithm, and I suspect manipulating the order of these vertices would be a good start in generating different orderings. I also digged into the depth_first_search() code, but there too it doesn't seem to be straightforward to change the order in which those start-nodes are visited. I am also aware that even if I would find a way to change the order in which the 0-in-degree vertices are visited it will still not give me all the various orderings; I think that repeating the procedure with a bredth_first_search() will give me yet more different orderings, and with I'll have all the different orderings. I may have missed something obvious but I really don't know how to tackle the problem of generating various orderings. Any advice, please? Thank you! -- Matan

2 1

void_cast patch
by VennGrayDog 22 Feb '08

22 Feb '08

I found #pragma warning(push) without #pragma warning(pop) difs attached Index: void_cast.hpp =================================================================== --- void_cast.hpp (revision 43370) +++ void_cast.hpp (working copy) @@ -221,6 +221,10 @@ } // namespace serialization } // namespace boost +#ifdef BOOST_MSVC +# pragma warning(pop) +#endif + #include <boost/config/abi_suffix.hpp> // pops abi_suffix.hpp pragmas #endif // BOOST_SERIALIZATION_VOID_CAST_HPP

1 0

Re: [Boost-users] Problem using boost::bind with VC2005 (compiler crashes)
by Daniel Lidström 22 Feb '08

22 Feb '08

> > The compiler seems to crash when trying to compile the > > RailCaterpillarControllerGUI constructor. > > Any ideas why? > > I'm using 8.0.50727.867 (vsvista.050727-8600), boost-1.34.1 > > Try adding the argument placeholders to your bind call: > > // cbToggleDeadZone should be called by widget's callback > CBF_Callback callback( > boost::bind( > &RailCaterpillarControllerGUI::cbToggleDeadZone, > this, > _1, > _2 > )); > > > HTH / Johan Of course, forgot the placeholders... Thanks! -- Daniel

1 0

boost::filesystem::detail::not_found_error with Boost installer on VC2005
by Rainer Thaden 22 Feb '08

22 Feb '08

Hi all, yes, I know, there has been a thread some days ago, but I just registered and don't know how to reply to a mail that I don't have. Anyway, there seems to be a bug in boost 1.34.1 with VC 2005 when using directory_iterator. I used the filesystem before and everything worked fine. Now (since I use the iterator), I get 1>HD2.obj : error LNK2019: Verweis auf nicht aufgelöstes externes Symbol ""unsigned int boost::filesystem::detail::not_found_error" (?not_found_error@detail@filesystem@boost@@3IA)" in Funktion ""private: unsigned int __thiscall boost::filesystem::basic_directory_iterator<class boost::filesystem::basic_path<class std::basic_string<char,struct std::char_traits<char>,class std::allocator<char> >,struct boost::filesystem::path_traits> >::m_init(class boost::filesystem::basic_path<class std::basic_string<char,struct std::char_traits<char>,class std::allocator<char> >,struct boost::filesystem::path_traits> const &)" (?m_init@?$basic_directory_iterator@V?$basic_path@V?$basic_string@DU?$char_traits@D@std@@V?$allocator@D@2@@std@@Upath_traits@filesystem@boost@@@filesystem@boost@@@filesystem@boost@@AAEIABV?$basic_path@V?$basic_string@DU?$char_traits@D@std@@V?$allocator@D@2@@std@@Upath_traits@filesystem@boost@@@23@@Z)". I looked into the filesystem dll with depends and found the function there with ordinal 48 (0x30) (in both the debug and the release Version for multithreading). Also in the lib, I found it. Rather strange. The signatures of the linker and the ones found in the DLL also match. I used the installer from Boost-Consulting. In the thread some time ago, a responder claimed, he could compile successfully. Maybe he used a self-compiled boost version? I am going to test that, but maybe somebody has an answer right now? Best regards, Rainer Thaden

1 1

Problem using boost::bind with VC2005 (compiler crashes)
by Daniel Lidström 22 Feb '08

22 Feb '08

Hello! I'm trying to use boost::bind and boost::function to perform a callback mechanism. However, the VC2005 compiler crashes for some reason. Let me show the code right away: #include <boost/bind.hpp> #include <boost/function.hpp> #include <boost/any.hpp> #include <boost/noncopyable.hpp> #include <iostream> #include <tchar.h> struct Fl_Widget; typedef boost::function<void (Fl_Widget&, const boost::any&)> CBF_Callback; struct Fl_Widget { CBF_Callback cbf_callback_; boost::any user_data_; //Three calls to set or probe that functor callback variable. CBF_Callback cbf_callback() const {return cbf_callback_;} void cbf_callback(CBF_Callback c, const boost::any& p) {cbf_callback_=c; user_data_=p;} void cbf_callback(CBF_Callback c) {cbf_callback_=c;} //Calls to do the functor callback void do_cbf_callback() {cbf_callback_(boost::ref(*this),user_data_);} // override user_data_ void do_cbf_callback(const boost::any& user_data) {cbf_callback_(boost::ref(*this),user_data);} }; struct GUI : private boost::noncopyable { static GUI& Instance() { static GUI theGUI; return theGUI; } void print() { std::cout << "Inside GUI" << std::endl; } private: GUI(){} }; struct RailCaterpillarControllerGUI { Fl_Widget widget; RailCaterpillarControllerGUI() { // cbToggleDeadZone should be called by widget's callback CBF_Callback callback(boost::bind(&RailCaterpillarControllerGUI::cbToggleDeadZone, this)); widget.cbf_callback(callback); } void cbToggleDeadZone(Fl_Widget& widget, const boost::any& data) { // assume data is reference to the GUI instance boost::any_cast<boost::reference_wrapper<GUI> >(data).get().print(); } }; int _tmain(int argc, TCHAR** argv) { RailCaterpillarControllerGUI controller; controller.widget.do_cbf_callback(boost::ref(GUI::Instance())); } Compiling gives this: 1>------ Build started: Project: test, Configuration: Debug Win32 ------ 1>Compiling... 1>test.cpp 1>c:\users\daniel.sbg\programming\boost_1_34_1\boost\bind.hpp(1575) : warning C4180: qualifier applied to function type has no meaning; ignored 1> c:\users\daniel.sbg\programming\boost_1_34_1\boost\bind.hpp(1609) : see reference to class template instantiation 'boost::_bi::add_cref<Pm,I>' being compiled 1> with 1> [ 1> Pm=void (__thiscall RailCaterpillarControllerGUI::* )(Fl_Widget &,const boost::any &), 1> I=1 1> ] 1> c:\users\daniel.sbg\programming\test\test.cpp(48) : see reference to class template instantiation 'boost::_bi::dm_result<Pm,A1>' being compiled 1> with 1> [ 1> Pm=void (__thiscall RailCaterpillarControllerGUI::* )(Fl_Widget &,const boost::any &), 1> A1=RailCaterpillarControllerGUI * 1> ] 1>c:\users\daniel.sbg\programming\boost_1_34_1\boost\mem_fn.hpp(318) : warning C4180: qualifier applied to function type has no meaning; ignored 1> c:\users\daniel.sbg\programming\boost_1_34_1\boost\bind\bind_template.hpp(22 5) : see reference to class template instantiation 'boost::_mfi::dm<R,T>' being compiled 1> with 1> [ 1> R=void (Fl_Widget &,const boost::any &), 1> T=RailCaterpillarControllerGUI 1> ] 1>Project : error PRJ0002 : Error result 255 returned from 'C:\Program Files\Microsoft Visual Studio 8\VC\bin\cl.exe'. The compiler seems to crash when trying to compile the RailCaterpillarControllerGUI constructor. Any ideas why? I'm using 8.0.50727.867 (vsvista.050727-8600), boost-1.34.1 Thanks in advance! -- Daniel

2 1

regexp - how to do an anti-match on a word
by David Ward 22 Feb '08

22 Feb '08

Boosters: I'm having trouble determining a regular expression that will do the following: 1. Match the following strings: TermGeMstreamIPTriplePlay_11.0.IpTxMstream_0.DefaultGateway TermGeMstreamIPTriplePlay_11.0.VplsTxMstream_0.CurMbps TermGeMstreamIPTriplePlay_11.0.MplsTxMstream_0.AvgMbps TermGeMstreamIPTriplePlay_11.0.MacTxMstream_0.Address 2. But not match the string: TermGeMstreamIPTriplePlay_11.0.TrafficTxMstream_0.RampStart So, essentially match the strings with IpTx, VplsTx, MplsTx and MacTx, but not match strings with TrafficTx. Right now, I have: .*\.(IpTx|MacTx|MplsTx|VplsTx)Mstream_0\..* This works, however, it is not very robust because there may be more entries added to the matching words over time. Is there someway to do an anti-match on the word "TrafficTx"? Something like: .*\.(!TrafficTx)Mstream_0\..* Thanks, David

2 1

"Re: Boost, HP-UX and gcc"
by stacie＠dcindigos.com 22 Feb '08

22 Feb '08

I'm on 11.23 itanium using gcc 4.1.1 and the latest build of 1.35 beta (feb 21) With my setup, c++config,h indicates that the header wchar.h does exist, however gcc does *not* set _GLIBCPP_USE_WCHAR_T nor_GLIBCXX_USE_WCHAR_T. My build keeps throwing No WCHAR_MIN and WCHAR_MAX errors, and suggests that I adjust integer_traits<>. I was going to just let it slide but unfortunately I received this error while trying to build using the Unit Test framework. Since unit testing is my friend, I would really like to get this working. I tried a work-around from the boost mailing list, but the code that I found in my files was not quite like what he modified: ***** Added code at top of bad header file: #ifdef hpux #include <wchar.h> #undef WCHAR_MIN /* (wchar_t)0 */ #define WCHAR_MIN 0 #endif * error simply changed to: /home/mlinck/develop/api/tp-sources-trunk/boost_1_33_1/boost/regex/v4/regex_traits_defaults.hpp:94:28: error: missing binary operator before token "(" * tried a different workaround at the code where the error is thrown. Noticed that broken code gets skipped for HP_aCC compiler. Figure we can maybe get away with same for gcc compiler: => #if !defined HP_aCC && !defined hpux * seems to compile. Had to apply to boost/regex/v4/perl_matcher.hpp as well ****** How would I modify integer_traits<> ? Since that is what is suggested...... Thanks!

3 2

Re: [Boost-users] "Re: Boost, HP-UX and gcc"
by stacie＠dcindigos.com 22 Feb '08

22 Feb '08

Hi Boris, No help there. I'm still getting the same errors. I'm willing to look at it some more if you have any more suggestions. How can I turn off the building of the python and serialization libraries? There are other libraries that may be broken with HP, too and I'd like to exclude those from the build. Stacie > -------- Original Message -------- > Subject: Re: [Boost-users] "Re: Boost, HP-UX and gcc" > From: "Boris Gubenko" <Boris.Gubenko(a)hp.com> > Date: Thu, February 21, 2008 4:19 pm > To: <boost-users(a)lists.boost.org> > Cc: Boris Gubenko <Boris.Gubenko(a)hp.com> > > > I don't think John's fix in http://svn.boost.org/trac/boost/changeset/42890 made > it to 1.35. You migh want to apply it locally and see if it fixes your problem. > > Thanks, > Boris > > ----- Original Message ----- > From: <stacie(a)dcindigos.com> > To: <boost-users(a)lists.boost.org> > Sent: Thursday, February 21, 2008 3:36 PM > Subject: [Boost-users] "Re: Boost, HP-UX and gcc" > > > > I'm on 11.23 itanium using gcc 4.1.1 and the latest build of 1.35 beta > > (feb 21) > > > > With my setup, c++config,h indicates that the header wchar.h does exist, > > however gcc does *not* set _GLIBCPP_USE_WCHAR_T nor_GLIBCXX_USE_WCHAR_T. > > > > My build keeps throwing No WCHAR_MIN and WCHAR_MAX errors, and suggests > > that I adjust integer_traits<>. > > I was going to just let it slide but unfortunately I received this error > > while trying to build using the Unit Test framework. Since unit testing > > is my friend, I would really like to get this working. > > > > I tried a work-around from the boost mailing list, but the code that I > > found in my files was not quite like what he modified: > > > > ***** > > Added code at top of > > bad header file: #ifdef hpux #include <wchar.h> #undef > > WCHAR_MIN /* (wchar_t)0 */ #define WCHAR_MIN 0 #endif > > * error simply changed > > to: > > /home/mlinck/develop/api/tp-sources-trunk/boost_1_33_1/boost/regex/v4/regex_traits_defaults.hpp:94:28: > > error: missing binary operator before token "(" > > * tried a different workaround at the code where the error > > is thrown. Noticed that broken code gets skipped for > > HP_aCC compiler. Figure we can maybe get away with same > > for gcc compiler: > > => #if !defined HP_aCC && !defined hpux > > > > * seems to compile. Had to apply to > > boost/regex/v4/perl_matcher.hpp as well > > ****** > > > > > > How would I modify integer_traits<> ? Since that is what is > > suggested...... > > > > > > Thanks! > > > > > > > > _______________________________________________ > > Boost-users mailing list > > Boost-users(a)lists.boost.org > > http://lists.boost.org/mailman/listinfo.cgi/boost-users > > > > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users

2 1

BGL: Finding a vertex by the vertex_index_t property
by Torsten Sadowski 21 Feb '08

21 Feb '08

Hi, is there a simple possibility to find a vertex in a adjacency_list graph by the vertex_index_t property? My uninspired way would be to build a map<int, vertex_descriptor> but there might be something more elegant. Cheers, Torsten

2 1

[variant] ideas needed
by ope 21 Feb '08

21 Feb '08

Hi, I've the following code: ---8<--- #include <boost/variant.hpp> #include <iostream> #include <string> struct available : public boost::static_visitor<bool> { template <typename T, typename U> bool operator()( const T&, const U& ) const { return false; // cannot compare different types } template <typename T> bool operator()( const T& lhs, const T& rhs ) const { static const std::string NA( "NA" ); return lhs != NA; //??? } }; using namespace boost; using namespace std; int main() { variant< double, std::string > R; variant< double, std::string > L; R = 3.14; L = "NA"; cout << R << endl; cout << L << endl; bool r = boost::apply_visitor(available(), R ); cout << ( r ? "true" : "false" ) << endl; bool l = boost::apply_visitor(available(), L ); cout << ( l ? "true" : "false" ) << endl; } --->8--- which does not compile. Anyway, from a data parser I get a double or a "NA" (not available) string. This test shall handle both cases. To process using the double I have to check if it available (than use defaults). Unfortunately, the static_visitor<bool> needs two arguments. How can I get the syntax (or simpler) as used in the example? Thanks Olaf

4 5

A request for help using the Boost Unit Test Library
by Tim Allison 21 Feb '08

21 Feb '08

I am working on a project that uses the Boost Multi-Index library and Boost shared_ptr, so I know I can use the Boost libraries on our project. However, I am unable to build so much as an example from the Boost Unit Test Library, and need help. (I assume the problem is a conflict with compiler switches, or something along those lines.) I set up a test directory, into which I will link the source code I wish to test. The Make file I am using is a modified form of one we use to build my project, so I know it will build both the module using some Boost libraries, and our old code, some of which is more than 10 years old. I am using boost 1.33.1-42@586 on Suse 10.2 (If it matters, my computer is a Latitude D510) and Gnu compiler 4.1.3-29@586 (as listed by Smart Package Manager). When I run the make file, it echos out the following line: g++ -Wno-deprecated -frepo -m32 -g -I../lib -I/usr/X11R6/include -I../../../source/ecs/shared -DOS_IS_LINUX -c FuserTests.cc FuserTests.cc is the file into which I have attempted to copy the examples. (The examples are both from the patmedia.net/rogeeff site, and the official boost site, as corrected within the discussion forum.) The case I am most interested in running uses BOOST_AUTO_TEST_CASE( functionName ). This line produces the error: FuserTests.cc:16: error: expected constructor, destructor, or type conversion before ‘(’ token Is there something special I need to do to get this to compile in C++? -- Tim Allison Chief Engineer MIDCON Millennium Engineering and Integration (719) 264-4302 Work (719) 264-4318 Fax

2 1

[has_xxx] How can it work at all???
by Ovanes Markarian 21 Feb '08

21 Feb '08

Hello *! As already asked many times in the list about the broken HAS_XXX ( BOOST_MPL_HAS_XXX_TRAIT_DEF<http://www.boost.org/libs/mpl/doc/refmanual/has-xxx-trait-def.html>) mpl I would like to ask following: If I use this code (on MSVC 7.1 compiler) and I have a type like: struct test { int xxx; }; and use the macro namespace x { BOOST_MPL_HAS_XXX_TRAIT_DEF(xxx); } This will result in the following code fragment: -------------------------------------------------------------------------------------------- /// included code from has_xxx.hpp namespace boost { namespace mpl { namespace aux { template< typename T > struct msvc71_sfinae_helper { typedef void type; }; } } } namespace sn { template< typename T, typename U = void > struct has_xxx_impl_ { static const bool value = false; typedef boost::mpl::bool_<value> type; }; template< typename T > struct has_xxx_impl_< T , typename boost::mpl::aux::msvc71_sfinae_helper< typename T::xxx >::type > { static const bool value = true; typedef boost::mpl::bool_<value> type; }; template< typename T, typename fallback_ = boost::mpl::bool_<false> > struct has_xxx : has_xxx_impl_<T> { }; } -------------------------------------------------------------------------------------------------- For me this will clearly fail, since T::xxx is not a type but an integer member of the struct. If xxx would be a member function it would not be a type as well, so has_xxx can only determine if there is some nested class type or a typedef within the inspected type. Is this correct? Many thanks, Ovanes

4 7

[bind, function] bind a cast
by Alexander Gutenev 21 Feb '08

21 Feb '08

Hi Is there any easy way to bind an explicit type conversion ? For example, I want to supress compiler warning here with static_cast: #include <boost/bind.hpp> #include <boost/function.hpp> int main(void) { boost::function<long (void)> foo; boost::function<short (void)> bar; bar = foo; // compiler warning. bar = boost::bind<short>(foo); // the same warning //bar = boost::bind(static_cast<short>, foo); // does not compile return 0; } If not, I think it might be added. Casts are useful sometimes not only to avoid warning, but also to get code compiled.

2 3

[Thread] Error compiling from svn trunk (tss.hpp)
by Luiz Marques 21 Feb '08

21 Feb '08

Hi, I'm trying to compile my code with boost.thread from SVN trunk (gcc 4.1.2, Fedora 7). I get the following error : /include/boost/thread/pthread/tss.hpp:55: error: invalid conversion from void* to MyThread* The code being compiled creates a thread_specific_ptr with a custom cleanup function, and is OK with boost 1.34 . Looking at the specified header, in line 55 there is a call to the custom cleanup_function: void operator()(void* data) { cleanup_function(data); } It seems like a static_cast<T*> is missing in the parameter. I changed it locally and the problem was solved. Can anyone confirm this problem ? Thanks, Luiz Marques

1 0

Re: [Boost-users] [regex] Why pattern anchored at buffer end ($)alsomatch for (\x0c)FF, (\x0d)CR
by Chandan Nilange 21 Feb '08

21 Feb '08

I am restarting this old thread. My Question: How to modify Boost regex library, to match perl behavior, such that, I only get pattern matches after/before \n and ignore (\x0c)FF, (\x0d)CR for $ & ^ anchored patterns ? Regards, Chandan John Maddock wrote: >Chandan Nilange wrote: > > >>Hello John, >> >>I not very clear. >> >>So do you mean that - >> >>$ & ^ matching pattern after/before any (\x0c)FF, >>(\x0d)CR is right PCRE /m modifier behaviour. >> >> > >I believe so, but deciding what the right behaviour should be is not >necessarily easy: Perl has the luxury that it controls File IO as well as >regexes: in contrast Boost.Regex is designed to accept text from all kinds >of sources and platforms. Most applications will in any case regard an >isolated \r or \n as a line break, as well as \r\n. Likewise in what sense >does a form-feed *not* start a new line? :-) > >HTH, John. > >_______________________________________________ >Boost-users mailing list >Boost-users(a)lists.boost.org >http://lists.boost.org/mailman/listinfo.cgi/boost-users > > ____________________________________________________________________________________ Be a better friend, newshound, and know-it-all with Yahoo! Mobile. Try it now. http://mobile.yahoo.com/;_ylt=Ahu06i62sR8HDtDypao8Wcj9tAcJ

2 4

which toolset for VC8 on XP64
by Andreas Fabri 21 Feb '08

21 Feb '08

Hello, With the precompiled boost libs from boost-consulting I get linker errors error LNK2019: unresolved external symbol "private: void __cdecl boost::detail::tss::init(class boost::function1<void,void *,class std::allocator<class boost::function_base> > *)" (?init@tss@detail@boost@@AEAAXPEAV?$function1@XPEAXV?$allocator@Vfunction_base@boost@@@std@@@3@@Z) referenced in function "public: __cdecl boost::detail::tss::tss(class boost::function1<void,void *,class std::allocator<class boost::function_base> > *)" (??0tss@detail@boost@@QEAA@PEAV?$function1@XPEAXV?$allocator@Vfunction_base@boost@@@std@@@2@@Z) I started a shell such that Microsoft Visual Studio 8/VC/BIN/amd64/CL.EXE is the compiler on the $PATH I then run bjam --toolset=msvc" but the problem persists. Is there another toolset I have to choose? thanks in advance, andreas

4 4

Link errors using VC++ 9.0, with Boost libs compiled with VC++ 8.0
by David Hartman 21 Feb '08

21 Feb '08

I'm compiling and linking using VC++ 9.0, and using Boost libraries built with VC++ 8.0. Unresolved external link errors are returned. Has anyone else had this problem? I'm going to build the Boost libraries using VC++ 9.0, for use with VC++ 9.0, but I was wondering if anyone else has tried to migrate to VC++ 9.0, and how they handled it. An example of the link error is below. 28>libboost_unit_test_framework-vc80-mt-gd.lib(xml_report_formatter.obj) : error LNK2019: unresolved external symbol "__declspec(dllimport) public: __thiscall std::_Container_base::~_Container_base(void)" (__imp_??1_Container_base@std@@QAE@XZ) referenced in function "protected: void __thiscall std::vector<struct std::pair<char,char const *>,class std::allocator<struct std::pair<char,char const *> > >::_Tidy(void)" (?_Tidy@?$vector@U?$pair@DPBD@std@@V?$allocator@U?$pair@DPBD@std@@@2@@std@@IAEXXZ) David Hartman

3 2

[program_options] Unicode problem
by Sergei Politov 21 Feb '08

21 Feb '08

The following example could not be compiled: ================================================================= #include <string> #include <boost/program_options.hpp> namespace po = boost::program_options; int main() { std::wstring test; po::options_description desc("Allowed options"); desc.add_options() ("test", po::wvalue<std::wstring>(&test)->default_value(L"value"), "description"); } ================================================================= It is due to wvalue::default_value tries to do lexical_cast<std::string>(std::wstring). Also I cannot change default_value to default_value("value") as soon as it accepts only something convertible to std::wstring. In other words it is impossible to specify default_value for wstring option.

3 8