February 2012 - Boost-users - lists.preview.boost.org

[PREDEF] Review ends today
by Joel Falcou 29 Feb '12

29 Feb '12

The review of Boost.PRedef by Rene Rivera is ending today, February 29th. I recall to all people wanting to submit a formal review to do so. Regards

1 0

Metaprogramming Question
by John M. Dlugosz 28 Feb '12

28 Feb '12

I've not had much opportunity to do much template metaprogramming in my work, but I've kept up with my reading. So, might I get some advice on how to approach this problem? I'm thinking of an assert-like statement that follows the syntax of Boost.Format, thus avoiding the variadic argument list. MY_FANCY_ASSERT (cond, "format string") % arg % arg; works if the macro expands to an unnamed temporary, whose destructor does the interesting work if the first argument is false. my_assert_formatter_type (cond, "cond", __FILE__, __LINE__, "format string") This is easy to do, simply deriving from boost::formatter, I would think. But, I want to avoid doing any work except on failure. This means remembering the arguments including the following %arg stuff, and passing everything through to Boost.Format based on the condition. So I'm thinking that template metaprogramming can let me collect all the arguments (any number of them) into one object, efficiently at compile time. So, what approach should I use? Plain MPL and classic metaprogramming? Fusion? Phoenix? and any concrete suggestions on how to proceed, or pointers to similar works? Thanks, —John

6 16

boost::python handles boost::shared_ptr specially...
by Neal Becker 28 Feb '12

28 Feb '12

Will it just work if I use std::shared_ptr instead?

2 1

WIN64 vs _WIN64
by Roberto Fichera 28 Feb '12

28 Feb '12

Hi All, I got a compilation error while compiling a library under Win64 arch that includes the boost/numeric/interval/msvc_rounding_control.hpp c:\boost_1.48\include\boost/numeric/interval/detail/msvc_rounding_control.hpp(33) : error C4235: nonstandard extension used : '_asm' keyword not supported on this architecture c:\boost_1.48\include\boost/numeric/interval/detail/msvc_rounding_control.hpp(33) : error C2065: 'FLD' : undeclared identifier c:\boost_1.48\include\boost/numeric/interval/detail/msvc_rounding_control.hpp(34) : error C4235: nonstandard extension used : '_asm' keyword not supported on this architecture So I checked the code under boost/numeric/interval/detail/msvc_rounding_control.hpp and it seems that it does a preprocessor check against WIN64 instead of _WIN64 namespace boost { namespace numeric { namespace interval_lib { namespace detail { #if BOOST_MSVC < 1400 || *defined(WIN64)* extern "C" { double rint(double); } #else inline double rint(double x) { _asm FLD [x] ; _asm FRNDINT ; //_asm RET ; } #endif Is there any reason why do not check against _WIN64 instead of WIN64 since the boost does always check against _WIN64? _WIN64 is also already pre-defined as normal MSVC definition when targeting x86_64 archs. [roberto@roberto include]$ grep -R WIN64 * boost/serialization/force_include.hpp:#elif ! defined(_WIN32) && ! defined(_WIN64) boost/detail/win/synchronization.hpp: #if defined(_WIN64) boost/detail/win/basic_types.hpp:# ifdef _WIN64 boost/intrusive/detail/mpl.hpp:#if defined(_MSC_EXTENSIONS) && !defined(__BORLAND__) && !defined(_WIN64) boost/python/converter/builtin_converters.hpp:# if defined(_MSC_VER) && defined(_WIN64) && PY_VERSION_HEX < 0x03000000 boost/xpressive/xpressive_fwd.hpp:# if (defined(_WIN32) || defined(_WIN64) || defined(_WINCE)) \ boost/thread/win32/thread_primitives.hpp:# ifdef _WIN64 boost/regex/config.hpp:# if(defined(_WIN32) || defined(_WIN64) || defined(_WINCE)) \ boost/signals2/detail/lwm_win32_cs.hpp:#if defined(_WIN64) boost/config/compiler/intel.hpp:#if (!(defined(_WIN32) || defined(_WIN64)) && defined(__STDC_HOSTED__) && (__STDC_HOSTED__ && (BOOST_INTEL_CXX_VERSION <= 1200))) || defined(__GXX_EXPERIMENTAL_CPP0X__) boost/config/compiler/intel.hpp:#if defined(_WIN32) || defined(_WIN64) boost/smart_ptr/detail/lwm_win32_cs.hpp:#if defined(_WIN64) *boost/numeric/interval/detail/msvc_rounding_control.hpp:#if BOOST_MSVC < 1400 || defined(WIN64)*

2 2

C++ guru required!
by Robert Ramey 27 Feb '12

27 Feb '12

I've got a really dumb question. My question is illustrated by the following code snippit. template<class T, class U> T inline operator%(const T & lhs, const U & rhs) { if(0 == rhs) throw std::domain_error("Divide by zero"); if(boost::is_same<T, boost::uintmax_t>::value && boost::numeric::is_signed<U>::value ){ if(1 == rhs || -1 == rhs) // is this dropped? overflow("unsigned type can hold this result"); } return lhs % rhs; } I would like to think that the second if is always dropped from the compile for a particular pair of types since this can be evaluated at compile time. If I'm correct, I can replace some tedious template metaprogramming for some straightforward and transparent code inserted in a convenient code inserted in the most convenient place. Andre Alex... gave a talk at "Going Native" proposing a "static if" for this case. But I don't see the necessity for for this since I would assume that the compiler just optimises away the "dead" code. I've compiled the above and it seems to do what I want but still I wonder. Basically I see lots of applications of variations on this idea to get the benefits of tmp without the attendent pain. Am I missing anything here? Robert Ramey

8 20

[quickbook] how to use
by Nathan Ridge 27 Feb '12

27 Feb '12

Hello, I am trying to use QuickBook to write documentation for a project (that's not a Boost library), and I have a basic question about how to use it. So far I've followed the setup instructions on this page, built the quickbook executable by running bjam in BOOST_ROOT\tools\quickbook, written a .qbk file in QuickBook format, and ran the quickbook executable on my .qbk file to produce an XML file which looks like BoostBook XML. Now how do I go about generating HTML or PDF from that BoostBook XML file? Thanks, Nate

2 1

[build] Is there any interest in a configure auto generator for boost-build?
by Tianjiao Yin 27 Feb '12

27 Feb '12

Hello all: Boost build configure (site-config.jam) is not easy to write for me, especially when I install some new compilers or tools. So I wrote a script to generate the configure in "/etc/site-config.jam" automatically for myself. You can get the whole source code here: http://ideone.com/kkI2G When I run this script by using argument "--setup", it will generate a new configure in "/etc/site-config.jam" like this: http://ideone.com/4yWyI I am using archlinux, not tested in any other platform. For archers, this script is part of boost-build-nightly in AUR, it works for me. BTW, boost-build 2.0_m12 is in the archlinux official repository. But this version is too outdated to build many examples in the boost 1.48.0. When m13 release? Regards, Tianjiao

1 0

[Container] comment on slist/forward_list documentation
by Brian Wood 26 Feb '12

26 Feb '12

Iiuc, Boost.Container implements slist but not forward_list. I'd like to suggest tweaking the following sentence. *"Boost.Container* updates the classic slist container with C++11 features like move semantics and placement insertion and implements it a bit differently for the standard C++11 forward_list." ^^ It might be better to say "and implements it a bit differently than the standard C++ forward_list." ? I wondered if you had both an slist and a forward_list implementation for a while and did some other checking to figure out what was in the library. http://www.boost.org/doc/libs/1_49_0/doc/html/container/non_standard_contai… Shalom Brian Wood Ebenezer Enterprises http://webEbenezer.net

2 1

[Iterators] How to implement proxy iterators?
by Oswin Krause 26 Feb '12

26 Feb '12

Hi Everybody, I'm trying to implement a proxy_iterator for ublas::matrix which should behave as an iterator over the matrix rows. My "obvious" solution does not seem to work: typedef boost::numeric::ublas::matrix<double> Matrix; typedef boost::numeric::ublas::matrix_row<Matrix> MatrixRow; typedef boost::numeric::ublas::vector<double> Vector; struct MatrixIterator: public boost::iterator_facade< MatrixIterator, Vector,//the semantic value type is a Vector boost::random_access_traversal_tag, MatrixRow//reference is a row of a matrix >{ //snip constructors private: friend class boost::iterator_core_access; //snip iterating stuff //fun part MatrixRow dereference() const { return row(*m_matrix,m_element); } std::size_t m_element; Matrix* m_matrix; }; which reuslts in an error that MatrixRow can't be implicitely casted to Vector. Which is correct. At the moment I'm using a 90% hacked workaround which changes the ValueType of the Iterator to MatrixRow. This seems to work as long as I don't use algorithms which rely on the actual value_type of the iterator (for example iter_swap as implemented in the gnu libstdc++ will fail miserably even though a correct swap is implemented for the MatrixRow...different story ) Since I need std::random_shuffle I am out of luck. I tried to find the actual requirements of iterator_facade with respect to ValueType/ReferenceType in the documentation, but was not able to find them. Does someone now a solution to this problem which works and results in an iterator which can safely be used in random_shuffle? Or can point me to the correct requirements? Greetings from Germany, Oswin Krause

2 5

Linking problem
by Johan Råde 26 Feb '12

26 Feb '12

I built Boost with the command b2 toolset=msvc-10.0 address-model=32 link=static runtime-link=shared threading=multi stage I'm linking statically with Boost and dynamically with the CRT (compiler option /MD). I get the error message LINK : fatal error LNK1104: cannot open file 'libboost_iostreams-vc100-mt-1_49.lib' What might the problem be?

3 4