February 2009 - Boost-users - lists.preview.boost.org

Re: [Boost-users] metafunction invocation performance
by Hicham Mouline 26 Feb '09

26 Feb '09

> ----- Original Message ----- > From: "David Abrahams" <dave(a)boostpro.com> > To: boost-users(a)lists.boost.org > Subject: Re: [Boost-users] metafunction invocation performance > Date: Thu, 26 Feb 2009 10:02:20 -0500 > > > > On Jul 4, 2008, at 7:36 AM, Hicham Mouline wrote: > > > The solution to this one is provided by Joaquin, in his reply below: > > > > I tried compiling your code with MSVC++ 8.0 and it is indeed incredibly > > slow. After playing a little with the code I realized that MPL views are > > the culprit here: for some reason it is much faster to fold on a > > vector-based computation rather than a view: > > > > I learned an interesting and somewhat depressing fact from Doug Gregor > recently. Most compilers don't use a canonical representation of types. > That is, in most compilers, int* and add_pointer<int>::type occupy separate > pieces of storage, and the complier jumps through some complicated hoops when > it needs to evaluate them for sameness. Of course that also means that > vector<int*> and vector<add_pointer<int>::type> are represented separately, > etc. That can totally skewer the gains of memoization, among other things.f > > Apparently recent versions of GCC have canonical type representation, and so > does Clang (for what that's worth -- no C++ yet), but I betcha MS doesn't, > and that would help explain why views cost more. I can try to experiment with this. I could try the view version versus the non-view version provided my Joaquin in g++. What minimum version of g++ ? regards,

1 0

Boost.Python linking error on Scratchbox
by Taneli R. 26 Feb '09

26 Feb '09

I have a program that uses Boost.Python and it compiles & links in normal desktop environment at x86 architecture (Ubuntu Linux) without any problems. I've compiled the same Boost libraries (1.37) in Scratchbox ARM environment, since my goal is to get my program running in Nokia N800 Internet Tablet as well. Compiling is successful, but when I try to link my code to Python libraries I get the following error: /usr/local/lib/libboost_python-gcc34-mt-1_37.so: undefined reference to `PyUnicodeUCS2_AsWideChar' /usr/local/lib/libboost_python-gcc34-mt-1_37.so: undefined reference to `PyUnicodeUCS2_FromEncodedObject' collect2: ld returned 1 exit status I haven't had any problems with other Boost libraries I've used in Scratchbox. Should I compile the Boost.Python libraries with some specific parameter? -- View this message in context: http://www.nabble.com/Boost.Python-linking-error-on-Scratchbox-tp22176960p2… Sent from the Boost - Users mailing list archive at Nabble.com.

2 1

Dynamic function call question
by Perry, Daniel 26 Feb '09

26 Feb '09

Hello, I've been trying to design a function dispatch system and I was wondering if the following is possible with the current boost libraries... I currently parse json data into an vector of variants, I would like to be able to take those variants and invoke an arbitrary boost function object with a consistent return type. Is it possible to determine the function argument types, get the typed value from the proper index in the array of variants, and call the function with arguments? This sounds like something the boost fusion library might be able to do, but I can't find any example code using the library to see if this is the right fit for the capabilities of the library or if what I'm looking to do is not possible without tons of boiler plate. Any insight would be greatly appreciated, Regards, Daniel Perry

3 4

Re: [Boost-users] Dynamic function call question
by Alan M. Carroll 26 Feb '09

26 Feb '09

1 0

accumulator taking a deduced argument - endless compilation time
by er 26 Feb '09

26 Feb '09

Hello, Comments are in the code. Any help appreciated. *.hpp template<typename RealType,typename Id> class mu : public accumulator_base { typedef mu<RealType,Id> self_type; typedef parameter::deduced<self_type> par0_type; typedef parameter::parameters<par0_type> params; public: typedef RealType result_type; mu(result_type value):value_(value){} template<typename Args> mu(const Args& args):value_((result_type)(0)){ // compile bottleneck (caused by the assigment, not the rhs) const self_type& input = params()(args); value_ = input.value_; } private: result_type value_; }; *cpp: typedef mpl::int_<0> id0_t; typedef accumulators::tag::mu<id0_t> mu0_t; typedef boost::mpl::vector< mu0_t > explicit_entities; typedef double value_type; typedef accumulators::accumulator_set< value_type, explicit_entities > acc_type; typedef mpl::apply<mu0_t::impl,value_type>::type input0_type; input0_type input0(0.5); // acc_type acc; // compiles fast acc_type acc((input0)); // compiles forever

2 4

C_time.hpp fails to get a vaild time
by Lukas.Orlowski＠dlr.de 26 Feb '09

26 Feb '09

Dear Boost users I'm experiencing problem with the latest Boost version 1.3.8. When ever I try to get the current system time or do anything that needs the system time like: boost::xtime xt; boost::xtime_get(&xt, boost::TIME_UTC); The function c_time.hpp "static std::tm* gmtime(const std::time_t* t, std::tm* result) " Throws it's "could not convert calendar time to UTC time" exception. This might be thread specific but since xtime.hpp uses date_time functions I'm certain it is not. Version 1.3.6 does work fine. Am I missing some API changes? Regards Luke

1 0

best way to remove duplicates from forwardIterator [startIter, endIter]
by Sandeep Gupta 26 Feb '09

26 Feb '09

Hi, Given a collection in terms of [startIter, endIter] how can obtain [uniqBeginIter, uniqEndIter] that represents unique elements of the original container. I can always explicitly instantiate a set S of [startIter, endIter] and use [S.begin(), S.end()]. Is it possible to express this in terms of algorithms. Please note that I won't be able to use std::sort as startIter does not conform to RandomAccessIterator. Its a forward iterator. Thanks Sandeep

3 5

Boost.Test on Non Exported Dll Classes
by Jon Black 26 Feb '09

26 Feb '09

I have a bunch of dll projects that make up an application. I'd like to unit test these projects, but I have two problems: 1) The classes I'd like to test are in the dll and not all of them are exported; 2) Boost.Test doesn't have a main entry point, so if I solve 1 by packing my tests in the dll, how can I run them? If there is a better solution to 1, I'd love to hear it. Otherwise, how can I solve 2?

2 1

[TR1] How to say BOOST_USE_BOOST_TR1
by Markus Werle 26 Feb '09

26 Feb '09

Hi! I'd like to #include <boost/tr1/*.hpp> while at the same time state that I am *not* interested in the compiler's TR1 implementation (vc9 ...) Any MACRO BOST_HAS_NO_TR1 I missed? The opposite (BOOST_HAS_TR1) exists. Markus

6 9

Re: [Boost-users] Boost-users Digest, Vol 1918, Issue 3
by Stefan Funck 26 Feb '09

26 Feb '09

I'm not sure how that could work, since I want different access rights for the very same object? Stefan > Though I have not used it myself, there is enable_if or disable_if in BOOST. That might help. > > -dhruva > > > > ----- Original Message ---- >> From: Stefan Funck <stefanfunck(a)web.de> >> To: boost-users(a)lists.boost.org >> Sent: Thursday, 26 February, 2009 7:34:01 PM >> Subject: [Boost-users] make member functions (in)accessible >> >> Dear all, >> >> I'm trying to find some magic that makes a member function of some object x >> accessible if and only if it were called from within a class that has object x >> as a member. Is there a (boost)-way of doing this? >> >> As this may seem a bit obscure, I'm trying to give an example: >> >> template <typename T, bool Writable> >> class Property >> { >> public: >> T get() const; >> void set(const T& val); >> private: >> T m_val; >> }; >> is a simple class holding a value of T. >> >> I have some classes holding potentially many Properties. I'd like to have these >> properties publically available, some as only readable, others writable as well, >> so that others (e.g. gui-classes) can access them: >> >> class A >> { >> public: >> Property propA; >> Property propB; >> }; >> >> so I may do: >> >> A a; >> a.propA.get(); >> a.propA.set(0.5); >> a.propB.get(); >> >> but NOT: >> >> a.propB.set(1); >> >> because propB was defined with Writable=false. BUT, from within class A, I would >> like to do >> >> propB.set(1); >> >> because otherwise this property is useless. (How) can this be achieved (with >> boost)? >> >> I could make set() private in the Writable=false - specialisation of Property >> and declare A as a friend, but that's prohibitive since Property shouldn't know >> about A and it won't work for derived classes of A. Another option would be >> >> class A >> { >> public: >> A() : propA(propA_priv), propB(propB_priv) {} >> >> PropertyWrapper propA; >> PropertyWrapper propB; >> private: >> Property propA_priv; >> Property propB_priv; >> }; >> >> so the actual properties are private and the wrapper provides the actual >> interface to the outside world. But as there may be many properties I'd like to >> get away without the wrapper objects ... >> >> Thanks for any ideas! >> Stefan

1 0