- Boost-users - lists.boost.org

[regex] How to pass BOOST_REGEX_RECURSIVE to bjam build process (win32)
by Christian Hoffmann 06 Mar '09

06 Mar '09

Hi, we used to compile boost regex in boost version 1.36 with the following command: bjam.exe -sHAVE_ICU=1 -sICU_PATH=../../../../icu -sBOOST_REGEX_RECURSIVE toolset=msvc-7.1 threading=multi python-debugging=on link=shared variant=debug --without-mpi libs\regex\build When switching to boost 1.38, we realized that BOOST_REGEX_RECURSIVE is not passed/picked up anymore when using the same command (apart compiler version to compile (verified with some added pragma messages): How should this be configured nowadays? Or did it never work that way but the default behaviour changed? Maybe the user-config.jam file should be used, but I didn't find an example of how to do this. Thanks a lot, Chris

3 2

BoostCon Registration
by Robert Jones 06 Mar '09

06 Mar '09

BoostCon registration seems to be broken. I started the reg. process, received the confirmation email, followed the link in the email, which gave "Page Not Found". Is there anyone on here who can help? Thanks, Rob.

2 1

[Proto] Annoying MSVC bug
by Dave Jenkins 06 Mar '09

06 Mar '09

Almost every useful Proto program has workarounds for an MSVC bug that causes syntax errors on valid nested function types. Eric Niebler filed a bug report with Microsoft on 6/14/2007 and they closed it on 6/19/2007 with status "Closed(Won't Fix)". This is back in the days of Visual Studio 2005. I'd like to encourage Microsoft to reopen this bug report and fix it. Could any of you with logins to Microsoft Connect go here https://connect.microsoft.com/VisualStudio/feedback/ViewFeedback.aspx?Feedb… and answer their 3 question survey expressing your dissatisfaction and encourage them to fix this. Thanks, Dave Jenkins

3 3

boost library advice wanted
by Sean Farrow 06 Mar '09

06 Mar '09

Hi: I am in the process of implementing a program that turned commands from voice recognition software in to code that are entered in to an application. there are approximately 200 codes available in this particular sofware I am automating. Wile I could use boost::stringAlgo along with the functions available in the stl there must be a better alternative. To give an idea of howthis would work the user might say: conditional new line, and [l] would be placed in the current document onscreen. the user could say skip 5 lines and [sk5] would be entered.

1 0

xcode iphone vs regular project issues
by Tom Delorenzi 06 Mar '09

06 Mar '09

Hi. I have worked on boost before using Visual Studios and loved using it. I am using a mac for some development now with xcode. My question is 2 fold. First off, on a regular c++ project I can use boost no problem, but I haven't seen how to get the auto complete to list functions through the -> operator. Has anyone had luck? In visual studios i could include the header and the non template header in my project to get this functionality to work. Second question is, when developing a project that was created using the iphone template the program fails to compile when I include any boost.h files, even though I am including them in the exact same way that was succesfull in the c++ project. Has anyone tried this?

2 1

xcode smart pointers
by Tom Delorenzi 06 Mar '09

06 Mar '09

I figured out both my problems. Boost works fine in iphone projects, you just need to have all your source code end in .mm to enable objective-c++ and the other problem is caused because xcode cannot handle templates when it comes to code insight(or namespaces supposidly). Thats really unfortunate.

1 0

Nstate: dynamic_bitset generalized for bounded integer values?
by Hostile Fork 06 Mar '09

06 Mar '09

Hello boost-users! I've written a class in the same spirit as the dynamic_bitset. But instead of being restricted to inserting values that range [0..1], the packed values can range from [0..radix-1]. Nothing that fancy, but a few people on #boost IRC thought it looked potentially useful, so I decided to try announcing it here. The library is called "Nstate", and is released under the boost software license (of course!) http://hostilefork.com/nstate/ As an example, I point out that it can fit 20 "Tristates" (a.k.a. Nstate<3>) into 32 bits, as opposed to 20 bytes (if one used a packing approach with 2 bits per tristate, you would only get 16 of them out of 32 bits). I'm guessing this would be low on the radar in terms of getting vetted for inclusion in boost. But if anyone out there is looking for something *like* this, here's a start. I'd certainly appreciate any peer review or wisdom to help make it better, faster, smarter...boostier? So if you're bored and just like making free libraries have more "Quality" then please send feedback! Files are: http://github.com/hostilefork/nocycle/blob/master/Nstate.hpp http://github.com/hostilefork/nocycle/blob/master/Nstate.cpp Regards, Brian

1 0

Re: [Boost-users] [Units] implicit conversion of quantities
by Christopher Bruns 05 Mar '09

05 Mar '09

Steven Watanabe wrote: > In this particular case, I would suggest writing the function as a template > and casting. In general you can write a forwarding overload that casts as > appropriate. Thanks Steven for the advice. I consulted Matthias Schabel off list to help figure out how to get this working. In case others have similar needs, I am attaching my modified test program. /* =================================== */ #include <boost/units/systems/si.hpp> #include <boost/units/base_units/metric/angstrom.hpp> #include <iostream> namespace boost { namespace units { typedef metric::angstrom_base_unit::unit_type angstrom_unit; BOOST_UNITS_STATIC_CONSTANT(angstrom, angstrom_unit); typedef scaled_base_unit< si::meter_base_unit, scale<10, static_rational<-9> > > nanometer_base_unit; typedef nanometer_base_unit::unit_type nanometer_unit; BOOST_UNITS_STATIC_CONSTANT(nanometer, nanometer_unit); } } // namespace boost::units using namespace boost::units; using namespace std; // First version of showBondLength takes a specific quantity type (nanometers) std::ostream& showBondLength(quantity<nanometer_unit> len, std::ostream& os) { // This implementation could be compiled into a library. os << len << std::endl; } // Second version of showBondLength can take any length type, and casts it to // the specific type. The parameter type is a bit long winded for a public API // method, but not unreadable. template<class Y> std::ostream& showBondLength(quantity<unit<length_dimension, Y> > len, std::ostream& os) { // This (simple, forwarding) implementation must be fully defined in a header. return showBondLength( (quantity<nanometer_unit>)len, os); } int main() { // These constructors all work; conversions are automatic quantity<nanometer_unit> len1(0.152 * nanometer); quantity<angstrom_unit> len2(1.52 * angstrom); quantity<nanometer_unit> len3(1.52 * angstrom); quantity<angstrom_unit> len4(0.152 * nanometer); // These three calls are OK, nanometer quantity is passed showBondLength(0.152 * nanometer, cout); showBondLength(len1, cout); showBondLength(len3, cout); // Explicit casts to nanometer are OK for angstrom quantities... // but yuck showBondLength( (quantity<nanometer_unit>)(1.52 * angstrom), cout ); showBondLength( (quantity<nanometer_unit>)len2, cout ); showBondLength( (quantity<nanometer_unit>)len4, cout ); // These all (used to) fail to compile before templated version was added. // Users want implicit conversion here from angstroms to nanometers showBondLength(1.52 * angstrom, cout); showBondLength(len1, cout); showBondLength(len2, cout); showBondLength(len3, cout); showBondLength(len4, cout); // Conversion from meters works too showBondLength(3.0 * si::meters, std::cout); // Non-length quantities correctly fail to compile // showBondLength(3.0 * si::seconds, std::cout); return 0; } /* =================================== */

1 0

[boost] compose ArgumentPacks using a comma
by er 05 Mar '09

05 Mar '09

Hi All, I have a structure containing a fusion sequence which is constructed recursively. BOOST_PARAMETER_KEYWORD(tag,kwd_struct) // This element depends on others, hence fetches the structure elem_x(const Args& args){ ...args[kwd_struct]...; } structure(Args const& args) :elements_( make_elements( ((kwd_struct = (*this)),args) ) ) {}; error: in constructor elem_x(const Args&) [with Args = ...] no match for operator[] in args[kwd_struct] If I reverse the order of the arguments (args,(kwd_struct = (*this))) it compiles. Why the discrepancy?

1 1

[accumulator] extract_result during initialization
by er 05 Mar '09

05 Mar '09

Hi All, I have an accumulator (a) that takes an initial value (x_), and another (b), whose initial value (x_) depends on that of (a). I would have thought that this is feasible because accumulator_set initializes (a) before (b), but apparently b.x_ is initialized with a garbage value. namespace impl { template<typename Sample,typename Id> class accu_a_impl : public accumulator_base { public: typedef Sample result_type; template<typename Args> accu_a_impl(Args const &args) :x_( args[kwd<Id>::value] ) { } result_type result(dont_care) const { return x_; } private: result_type x_; }; } namespace tag { template <typename Id = mpl::void_> struct accu_a : depends_on<> { typedef accumulators::impl::accu_a_impl<mpl::_1,Id> impl; }; } namespace impl { template<typename Sample,typename Id0,typename Id1> class accu_b_impl : public accumulator_base { typedef tag::accu_a<Id1> a_type; public: typedef Sample result_type; template<typename Args> accu_b_impl(Args const &args) :x_( extract_result<a_type>(args[accumulator]) // faulty? ) { } template<typename Args> void operator ()(Args const &args) { // overrides initialization for the sake of testing x_ = extract_result<a_type>(args[accumulator]); } private: result_type x_; }; } namespace tag { template <typename Id0,typename Id1=mpl::void_> struct accu_b : depends_on<tag::accu_a<Id1> > { typedef accumulators::impl::accu_b_impl<mpl::_1,Id0,Id1> impl; }; } *.cpp typedef mpl::size_t<0> id0; typedef mpl::size_t<1> id1; typedef double value_type; typedef tag::accu_a<id0> a_type; typedef tag::accu_b<id1,id0> b_type; typedef accumulator_set< value_type, stats< b_type > > set_type; set_type set( ( kwd<id0>::value = 0.9 ) ); std::cout << "a.x_=" << extract_result<a_type>(set) << std::endl; std::cout << "b.x_=" << extract_result<b_type>(set) << std::endl; set(0); std::cout << "b.x_=" << extract_result<b_type>(set) << std::endl; //Output: //a.x_=0.9 //b.x_=0.1 // where 0.9 is expected //b.x_=0.9

2 3