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

Spirit 1.6 with VC6
by Mark Storer 25 Jun '04

25 Jun '04

I'm getting several compiler errors when trying to use the Spirit library. When compiling one of the samples from the v1.6 documentation, I get a linker error: SpiritTest.obj : fatal error LNK1179: invalid or corrupt file: duplicate comdat "?do_parse@?$parse_subrule@V?$match@Unil_t@spirit@boost@@@spirit@boost@@U?$subrules_scanner@V?$scanner@PBDU?$scanner_policies@U?$skipper_iteration_policy@Uiteration_policy@spirit@boost@@@spirit@boost@@Umatch_policy@23@Uaction_policy@23@@spirit@boos67ac75e0" LNK1179 suggested that the /H _compiler_ option might be involved. It (supposedly) controls how long a symbol is allowed to be when the compiler is generating symbol names. I've tried several different variations on the /H option, /H512, /H1024, etc. They all give the same error message, and the symbol "?do_parse@..." doesn't get any longer (or shorter for that matter). I also get a number compiler errors when I #include <boost/spirit.hpp>: c:\sdks\boost-1.30.2\boost\spirit\utility\impl\chset\basic_chset.ipp(32) : error C2955: 'arg' : use of class template requires template argument list c:\sdks\boost-1.30.2\boost\bind\arg.hpp(26) : see declaration of 'arg' ----------basic_chset.ipp, lines 31-33 template <typename CharT> inline basic_chset<CharT>::basic_chset(basic_chset const& arg) : rr(arg.rr) {} ----------arg.hpp lines 24-26 template<int I> class arg { }; The boost/bind include is in the core/composite/actions. Trying to mess with the order of inclusion seems Fraught With Peril. The basic_chset template is in "utility", and everything in "utility" depends on various bits and pieces of "core". basic_chset is derived from unary, which is defined in core/composite. A Gordian knot... and me without a sword. --Mark Storer Senior Software Engineer Verity, Inc.

2 1

RE: [Boost-users] protassign
by Nat Goodspeed 24 Jun '04

24 Jun '04

> -----Original Message----- > From: Nat Goodspeed > Sent: Wednesday, June 23, 2004 12:17 PM > To: boost-users(a)lists.boost.org > Subject: RE: [Boost-users] protassign > > In the forthcoming Boost serialization library: > > http://www.rrsd.com/boost/ > > there's a state_saver template class that does what you're talking > about: > > http://www.rrsd.com/boost/libs/serialization/doc/state_saver.html > > This seems useful enough to promote to a library more general than the > serialization library. Hmm! I was suggesting that the state_saver template from the serialization library be elevated. If the protassign template makes it into Boost sooner, may I suggest that the serialization library be revised to use protassign instead of state_saver? We certainly wouldn't want two redundant Boost templates.

1 0

RE: [Boost-users] protassign
by Nat Goodspeed 24 Jun '04

24 Jun '04

> -----Original Message----- > From: Erik Thiele [mailto:erik@thiele-hydraulik.de] > Sent: Wednesday, June 23, 2004 11:28 AM > To: boost-users(a)lists.boost.org > Subject: [Boost-users] protassign > > now i can do: > > void foo() > { > protassign<bool> foorunning_assigner(foorunning, true); > dojobs(); > dofoooo(); > } > > and that's all. is there one such thing in boost already??? [Nat] In the forthcoming Boost serialization library: http://www.rrsd.com/boost/ there's a state_saver template class that does what you're talking about: http://www.rrsd.com/boost/libs/serialization/doc/state_saver.html This seems useful enough to promote to a library more general than the serialization library.

3 3

failures and featurerequest in/for numeric_cast
by Peter Piehler 24 Jun '04

24 Jun '04

failures and featurerequest in/for numeric_cast I use boost in a little project for my study. In some cases the software must convert numeric datatypes. So I tried to convert with boost-library numeric_cast and found the following problems (gcc version 3.2.3 20030422): -the maximum could not be handled correctly, for example: +numeric_cast<int>(numeric_limits<unsigned int>::max()); doesn't throw a exception -no precession handling, example: +numeric_cast<float>(numric_limits<double>::epsilon()+1); doesn't throw a exception -no NaN, bounded/unbounded and exact handling. -no specialization. I need numeric_cast specializiation for: +numeric_limits-interface can't use for datatypes with caracteristics only known at runtime. +datatyps without static_cast support (example: gmp-library) So I have extended the boost::numeric_cast for my special requirements. You can download it under the following link: http://www.pienet.de/source/numeric_cast.tar.gz . Compile and run numeric_cast_tescases.cpp with USE_BOOST_CAST to test ::boost::numeric_cast. The output is: Running 2 test cases... numeric_cast_testcases.cpp(72): error in "standard_cast_test": exception bad_numeric_cast is expected numeric_cast_testcases.cpp(91): error in "standard_cast_test": exception bad_numeric_cast is expected numeric_cast_testcases.cpp(93): error in "standard_cast_test": exception bad_numeric_cast is expected numeric_cast_testcases.cpp(135): error in "standard_cast_test": exception bad_numeric_cast is expected *** 4 failures detected in test suite "numeric_cast Test" I think the line static_cast<X>(static_cast<Y>(x)) != x; is a tricky code. For X/Y==char/unsigned char,short/unsigned short etc. is always false. The compiler convert it only if the datatypes have not the same size/bytes or are between integers and floating points. sorry for my bad english and code, it's the first time that i ever have published. Peter Bestellen Sie Y! DSL und erhalten Sie die AVM "FritzBox SL" für 0. Sie sparen 119 und bekommen 2 Monate Grundgebührbefreiung. http://de.adsl.yahoo.com

2 1

bug in date_time? cannot initialize ptime to not_a_date_time
by Erik Thiele 24 Jun '04

24 Jun '04

hi. #include <boost/date_time/posix_time/ptime.hpp> int main() { boost::posix_time::ptime x(boost::date_time::not_a_date_time); } # g++ --version 2.95.4 #g++ -Wall -c foo.cpp -Iboost_1_31_0 (compiling into cache) foo.cpp: In function `int main()': foo.cpp:5: no matching function for call to `boost::posix_time::ptime::ptime (boost::date_time::special_values)' boost/boost/date_time/posix_time/ptime.hpp:39: candidates are: boost::posix_time::ptime::ptime(boost::gregorian::date, boost::posix_time::time_duration) boost/boost/date_time/posix_time/ptime.hpp:42: boost::posix_time::ptime::ptime(boost::gregorian::date) boost/boost/date_time/posix_time/ptime.hpp:45: boost::posix_time::ptime::ptime(const boost::posix_time::int64_time_rep &) boost/boost/date_time/posix_time/ptime.hpp:48: boost::posix_time::ptime::ptime(const boost::posix_time::ptime &) foo.cpp:5: warning: unused variable `class boost::posix_time::ptime x' -- Erik Thiele

2 1

protassign
by Erik Thiele 23 Jun '04

23 Jun '04

hi in some cases i do assign values to variables just for the life of my function: void foo() { bool oldfoorunning = foorunning; foorunning = true; dojobs(); dofoooo(); foorunning = oldfoorunning; } now the problem arises when dojobs throws an exception. in order to be safe i have to do: void foo() { bool oldfoorunning = foorunning; foorunning = true; try { dojobs(); dofoooo(); } catch (...) { foorunning = oldfoorunning; throw; } foorunning = oldfoorunning; } this is ugly and it can get more ugly by having more than one such variable. then you stack try/catch into each other and code gets really ugly and error prone. for that reason i created a class as follows: template<class X> class protassign : public boost::non_copyable { public: inline protassign (X &var_, const X &newvalue) : var(var_) { oldvalue = var; var = newvalue; } inline ~protassign () { try { var = oldvalue; } CATCHDESTRUCTOR; //<-- my special destructor error catching macro, //forget it. (is there an alternative in boost?) } private: X &var; X oldvalue; }; now i can do: void foo() { protassign<bool> foorunning_assigner(foorunning, true); dojobs(); dofoooo(); } and that's all. is there one such thing in boost already??? cu erik -- Erik Thiele

1 0

Re: Re: compiling boost with mingw in winxp
by Alexis H. Rivera-Rios 23 Jun '04

23 Jun '04

Hi, 1) g++ is in the path. Doing the g++ --version displays the version information: C:\opt\boost_1_31_0>g++ --version g++ (GCC) 3.2.3 (mingw special 20030504-1) Copyright (C) 2002 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. 2) Mingw is installed in the default directory: c:\mingw Does the fact that I got a warning about spawnvp function when I built bjam has something to do with this? C:\opt\boost-jam-3.1.10>.\bootstrap.mingw\jam0 -f build.jam --toolset=mingw "--t oolset-root= " ...found 83 targets... ...updating 4 targets... .cc. bin.ntx86\yyacc.exe .cc. bin.ntx86\mkjambase.exe .cc. bin.ntx86\jam.exe execnt.c: In function `execcmd': execnt.c:651: warning: passing arg 3 of `spawnvp' from incompatible pointer type execnt.c:682: warning: passing arg 3 of `spawnvp' from incompatible pointer type .link. bin.ntx86\bjam.exe 1 file(s) copied. ...updated 4 targets... I read somewhere something about different signatures for the spawn function and this indicates the one used by bjam is not the one that mingw supports. Why would spawn fail? Is this an OS issue or a compiler thing? Is bjam the program that fails? Or is it a python script that fails? How can I trace this error? I really appreciate your help, Alex ===== Programming Tutorial: In Python: To do this, do this In Perl: To do this, do this or this or this or this... In C: To do this, do this, but be careful In C++: To do this, do this, but don't do this, be careful of this, watch out for this, and whatever you do, don't do this __________________________________ Do you Yahoo!? Yahoo! Mail - 50x more storage than other providers! http://promotions.yahoo.com/new_mail

3 3

Re: [Boost-users] Re: compiling boost with mingw in winxp
by Alexis H. Rivera-Rios 22 Jun '04

22 Jun '04

Hi, This is the output of running the command line with the d+4 and d+2 option respectively. I'm running the command from the dos shell (cmd.exe). As far as the filesystem. The filesystem is NTFS. Thanks, Alex C:\opt\boost_1_31_0>bjam -sTOOLS=mingw -sPYTHON_ROOT=c:\Python23 -sPYTHON_VERSIO N=2.3 -d+4 Executing raw command directly Executing command [g++] [-Wl,--exclude-symbols,_bss_end__:_bss_start__:_data_end __:_data_start__] [-Wl,--out-implib,bin\boost\libs\date_time\build\boost_date_ti me.dll\mingw\debug\boost_date_time-mgw-d-1_31.lib] [-g] [-shared] [-mno-cygwin] [-o] ["bin\boost\libs\date_time\build\boost_date_time.dll\mingw\debug\boost_date _time-mgw-d-1_31.dll"] ["bin\boost\libs\date_time\build\boost_date_time.dll\ming w\debug\greg_month.obj"] ["bin\boost\libs\date_time\build\boost_date_time.dll\mi ngw\debug\greg_weekday.obj"] ["bin\boost\libs\date_time\build\boost_date_time.dl l\mingw\debug\date_generators.obj"] spawn: Invalid argument C:\opt\boost_1_31_0>bjam -sTOOLS=mingw -sPYTHON_ROOT=c:\Python23 -sPYTHON_VERSIO N=2.3 -d+2 gcc-Link-action bin\boost\libs\date_time\build\boost_date_time.dll\mingw\debug\b oost_date_time-mgw-d-1_31.dll bin\boost\libs\date_time\build\boost_date_time.dll \mingw\debug\boost_date_time-mgw-d-1_31.lib g++ -Wl,--exclude-symbols,_bss_end__:_bss_start__:_data_end__:_data_start_ _ -Wl,--out-implib,bin\boost\libs\date_time\build\boost_date_time.dll\mingw\debu g\boost_date_time-mgw-d-1_31.lib -g -shared -mno-cygwin -o "bin\boost\libs\dat e_time\build\boost_date_time.dll\mingw\debug\boost_date_time-mgw-d-1_31.dll" "bin\boost\libs\date_time\build\boost_date_time.dll\mingw\debug\greg_month.obj" "bin\boost\libs\date_time\build\boost_date_time.dll\mingw\debug\greg_weekday.obj " "bin\boost\libs\date_time\build\boost_date_time.dll\mingw\debug\date_generator s.obj" spawn: Invalid argument ===== Programming Tutorial: In Python: To do this, do this In Perl: To do this, do this or this or this or this... In C: To do this, do this, but be careful In C++: To do this, do this, but don't do this, be careful of this, watch out for this, and whatever you do, don't do this __________________________________ Do you Yahoo!? Yahoo! Mail - 50x more storage than other providers! http://promotions.yahoo.com/new_mail

4 4

Dynamic bitset help
by Michael Short 21 Jun '04

21 Jun '04

Hello please anyone some one must have a example or something using the dynamic bitset? Has anyone got a few good examples of the dynamic bitset been used with member functions like append and block type? If so could you please send in some thanks. Michael

1 0

Re: [Boost-users] Newbie: What is the actual return type?
by Doug Gregor 21 Jun '04

21 Jun '04

[Sorry about the top-post; switching e-mail accounts] On Sun, 20 Jun 2004, Tim Rowe wrote: > > You're actually getting back an (Lvalue) property map. Think of it > > like an std::map<vertex_descriptor, std::string>, because it maps > > from the vertices in the graph g to the vertex name (which is an > > std::string here). The difference is that the property map is fast > > (O(1)) whereas the std::map would be slow (O(log |V|)). > > > > Interior properties will get a bit easier with the next Boost > > release. > > > > > > Doug > > Thanks. The property_map documentation says that "the get() function > takes a property map object [...] and a key object"; is that the get() > that's used there? Well, no :(. The get() being used there is actually part of the PropertyGraph concept, documented here: http://www.boost.org/libs/graph/doc/PropertyGraph.html When you have an interior property p (like vertex_name_t in the example graph) and graph g, the expression get(p, g) gives you a property map for that particular property and graph, that can be "indexed" with the appropriate key (e.g., a vertex or edge descriptor). > So vertex_name is a map of maps, with a Graph as a key? This analogy actually works (much to my surprise!), so long as you don't push it too far :) So yes, get(vertex_name, g) returns a map from vertices in g to the name of the vertex. > And where is the interface to property_map defined? > boost_1_31_0\libs\graph\doc\using_property_maps.html says that it's > defined in boost_1_31_0\libs\property_map\property_map.html, but that > just > contains another example, and I can't find which header declares it -- I > can't see it in property_map.hpp which is where I expected it to be! The actual struct template property_map is used (only!) to get the return type of, e.g., get(vertex_name, g). It is defined in boost/graph/properties.hpp, but the definition won't help much because I don't think that's quite what you're asking for. When you use get(vertex_name, g) you get back a property map. There's no single type that defines a property map, because it's a concept. To find out what you can do with a property map, check out: http://www.boost.org/libs/property_map/property_map.html Specifically, you want the links to the ReadablePropertyMap, WriteblePropertyMap, and LvaluePropertyMap concepts. Doug

1 0