January 2005 - Boost-users - lists.preview.boost.org

enable_if for function templates
by Agoston Bejo 23 Feb '05

23 Feb '05

Hi all, I am trying to enable only two version of a function. The caller may choose one of the two versions by specifying a template parameter. I've got two versions for the problem, both of them failing the same way. -------------------------- Version A): integral template parameter -------------------------- #include <iostream> #include <boost/utility/enable_if.hpp> #include <boost/type_traits.hpp> using namespace std; using namespace boost; template<unsigned int N, typename T> enable_if_c<N==1> f(T t) { cout << 1 << endl; } template<unsigned int N, typename T> enable_if_c<N==2> f(T t) { cout << 2 << endl; } int _tmain(int argc, _TCHAR* argv[]) { f<1>(10); // ERROR f<2>(5.5); // ERROR f<3>(7); // ERROR return 0; --- The error message on all 3 lines is: error C2668: 'f' : ambiguous call to overloaded function The compiler's "explanation" for the first one: (for the other two it is much the same except the type) could be 'boost::enable_if_c<B,T> f<1,int>(int)' with [ B=false, T=void ] or 'boost::enable_if_c<B,T> f<1,int>(int)' with [ B=true, T=void ] while trying to match the argument list '(int)' -------------------------- Version B) : type template parameter -------------------------- #include <iostream> #include <boost/utility/enable_if.hpp> #include <boost/type_traits.hpp> using namespace std; using namespace boost; struct A1 {}; struct A2 {}; template<typename U, typename T> enable_if< is_same<U, A1> > g(T t) { cout << 1 << endl; } template<typename U, typename T> enable_if< is_same<U, A2> > g(T t) { cout << 2 << endl; } int _tmain(int argc, _TCHAR* argv[]) { g<A1>(10); // ERROR g<A2>(5.5); // ERROR g<int>(7); // ERROR return 0; } ---- Error messages: error C2668: 'g' : ambiguous call to overloaded function (for all three of them) "Explanation" sample: could be 'boost::enable_if<Cond,T> g<A1,int>(int)' with [ Cond=boost::is_same<A1,A2>, T=void ] or 'boost::enable_if<Cond,T> g<A1,int>(int)' with [ Cond=boost::is_same<A1,A1>, T=void ] while trying to match the argument list '(int)' ----------------- enable_if should actually prevent the compiler from trying to compile the 'false' version, shouldn't it? Thx, Agoston

3 2

using boost::format with STLPort
by Alexey Bakhvalov 23 Feb '05

23 Feb '05

Hello, I have a problem when I try to include "boost/format.hpp" from Boost 1.32 in a project that uses STLPort library. I am using MSVC 7.1 and the compiler prints the following error: \boost\format\format_implementation.hpp(216) : error C2244: 'boost::basic_format<Ch,Tr,Alloc>::size' : unable to match function definition to an existing declaration \boost\format\format_class.hpp(58) : see declaration of 'boost::basic_format<Ch,Tr,Alloc>::size' definition 'basic_format<Ch,Tr,Alloc>::size_type boost::basic_format<Ch,Tr,Alloc>::size(void) const' existing declarations 'boost::basic_format<Ch,Tr,Alloc>::size_type boost::basic_format<Ch,Tr,Alloc>::size(void) const' If I switch back to Boost 1.31 or don't use STLPort then the problem goes away. I just wonder what can be the cause of such strange error because I can't see any difference between function's definition and declaration. Also I use many other Boost libraries and don't have this problem with either of them.

3 2

boost::format bug?
by Orhun Birsoy 23 Feb '05

23 Feb '05

Hi, I am getting boost::io::too_many_args exceptions for the following line. cout << format(" %1% %3% \n") % 1 % 2 % 3 % 4 % 5; but not for this one cout << format(" %1% %5% \n") % 1 % 2 % 3 % 4 % 5; If the last argument is used in the format, I do not get the exception. Is this a bug? Thanks. Orhun Birsoy -- _____________________________________________________________ Web-based SMS services available at http://www.operamail.com. >From your mailbox to local or overseas cell phones. Powered by Outblaze

2 1

"template with C linkage"
by Chris Coleman 14 Feb '05

14 Feb '05

Hi, I'm struggling with Boost (1.32) on Solaris 8 at the moment - I do not have root access to the machine. The default compiler is gcc-2.95, and so has no <ostream> <locale> etc so the -sTOOLS=gcc toolset won't work. I then thought that the solution would be to install STLPort, but can't get this to build cleanly. Since then I have discovered that there is another version of gcc (gcc-3.2) hidden in /opt/sfw/gcc-3.2. With this I then built the boost libraries again, and installed them with the line: bjam "--prefix=/staff/cqc/ -sGCC_ROOT_DIRECTORY=/opt/sfw/gcc-3.2 -sTOOLS=gcc" install This seemed to work fine with no problems at all. I then moved the headers, and the libs to the approprite places, and set about compiling my project. The result below is repeated many many times about various files: /opt/sfw/gcc-3.2/bin/gcc -DHAVE_CONFIG_H -I. -I. -I.. -D_REENTRANT -c -o zz_broker.o `test -f zz_broker.cpp || echo './'`zz_broker.cpp In file included from /opt/sfw/gcc-3.2/include/c++/3.2/utility:66, from /staff/cqc/include/boost/config/select_stdlib_config.hpp:20, from /staff/cqc/include/boost/config.hpp:40, from /staff/cqc/include/boost/smart_ptr.hpp:14, from zz_broker.h:28, from zz_broker.cpp:24: /opt/sfw/gcc-3.2/include/c++/3.2/bits/stl_relops.h:91: template with C linkage /opt/sfw/gcc-3.2/include/c++/3.2/bits/stl_relops.h:104: template with C linkage /opt/sfw/gcc-3.2/include/c++/3.2/bits/stl_relops.h:117: template with C linkage /opt/sfw/gcc-3.2/include/c++/3.2/bits/stl_relops.h:130: template with C linkage I have searched the list archives and googled around quite alot, but can't seem to find any info. Any help will be greatly recieved, Cheers Chris This message has been scanned but we cannot guarantee that it and any attachments are free from viruses or other damaging content: you are advised to perform your own checks. Email communications with the University of Nottingham may be monitored as permitted by UK legislation.

6 8

Re: Compiling boost on tru64 (Richard Hadsell)
by Mark Helzer 06 Feb '05

06 Feb '05

>Mark Helzer wrote: >>I'm getting this error trying to build boost >>tru64cxx65-C++-action >>bin/boost/libs/date_time/build/libboost_date_time.so/tru64 >>cxx65/debug/greg_month.o >>cxx: Error: /adp/local/src/boost_1_32_0/boost/cstdint.hpp, line 87: #5 >>could >> not open source file "inttypes.h" >> >>Quite true, it doesn't exist. >>Here's my command line: >>bjam -sTOOLS=tru64cxx65 install >> >>Compiler is >> >> >>cxx -V >> >> >>Compaq C++ V6.5-041 for Digital UNIX V4.0F (Rev. 1229) >>Compiler Driver V6.5-041 (cxx) cxx Driver >> >>Is this a configuration issue? >> >> >For now, remove '|| defined(__osf__)' from line 85 of cstdint.hpp. >There may be a more general fix in the current CVS version, but this >change works for me (and my 4.0F system). Kewl, that works, I got a little further down the path. Next problem: cxx: Error: /adp/local/src/boost_1_32_0/boost/bind.hpp, line 206: #266 "value" is ambiguous detected during: instantiation of class "boost::_bi::list1<A1> [with A1=boost::_bi::list_av_1<boost::arg<1>>::B1]" at line 161 of "/adp/local/src/boost_1_32_0/boost/bind/bind_template.hpp " instantiation of class "boost::_bi::bind_t<R, F, L> [with R=std::wstring, F=std::wstring (*)(const std::string &), L=boost::_bi::list1<boost::_bi::list_av_1<boost::arg<1>>: :B1>]" at line 71 of "/adp/local/src/boost_1_32_0/libs/program_options/build/. ./src/parsers.cpp" template<class T> T & operator[] (value<T> & v) const { return v.get(); } --------------------------------------^ I see value defined as a class template at line 90 of boost/bind.hpp

3 3

[variant] is_a/can_be
by Yuval Ronen 05 Feb '05

05 Feb '05

Hi. I encountered the need to query a boost::variant to which type it currently contain. I'm talking about something like: variant<int, string> v(3); assert(v.is_a<int>()); assert(!v.is_a<string>()); This can be done using the which() method, but its drawback is that I rely on the order of the types in the variant declaration rather on the type, which is really what I'm interested in. I can implement the variant_is_a<>() feature using a visitor such as: template <typename T> struct is_a_visitor : public boost::static_visitor<bool> { template <typename U> bool operator()(const U &) const { return false; } bool operator()(const T &) const { return true; } }; template <typename T, class Variant> bool variant_is_a(const Variant &v) { return v.apply_visitor(is_a_visitor<T>()); } but this way has a problem: calling variant_is_a<double>() on a variant of type variant<int, string> (no double for this variant) will simply return false, rather than just not compile, which is what I prefer. Making this not compile will require a new feature from variant: a boolean constant 'can_be'. With this, the variant_is_a() function could be written with additional line: BOOST_STATIC_ASSERT(Variant::can_be<T>::value); IMO, the is_a/can_be features are general and useful enough to be included in the library. Wouldn't you agree? Yuval

4 7

[mpl::transform] How to make metafunction from transform?
by Vyacheslav E. Andrejev 04 Feb '05

04 Feb '05

Hello All! I want to parametrize mpl::transform to form a new metafunction. But I have confused with mpl syntax and notations. Imagine, I have following vector typedef boost::mpl::vector<Type1, Type2, Type3, Type4> types; and following transformation: typedef boost::mpl::transform< types , boost::is_convertible<Type3, boost::mpl::_1> >::type result; The result of the tansformation is a bool vector that indicates whether Type3 is convertible to any of type from types vector. Now I want to form metafunction that will construct such array not from Type3, but from its argument. Of course, I may use straightforward method: struct WhoConverted { template <typename ConversionType> struct apply { typedef typename boost::mpl::transform< types , boost::is_convertible<ConversionType, boost::mpl::_1> >::type type; }; }; But it seems more like workaround rather than solution. Can it be done in more convenient and short manner by means of all mpl machinery like mpl::lambda, mpl::bind, placeholders, etc... ? -- Vyacheslav E. Andrejev System Architect, Excelsior, LLC

2 2

accessing the underlying C array of a boost::array
by Frank Hess 02 Feb '05

02 Feb '05

Does anyone know what the preferred method of accessing the underlying C array of a boost::array is? In boost 1.32 there is a data() member which returns a const pointer, and a c_array() member which returns a pointer. Also, the elems member itself is public. But at http://www.josuttis.com/cppcode/array.html, the author states that data() can be used for both const and non-const pointers, and indeed it is so in the version he posts on his website http://www.josuttis.com/cppcode/array.hpp.html (which doesn't provide the c_array() member but overloads data() instead). -- Frank

2 1

boost-users: cant i use mpl::contains to find members of list_c ??
by suresh.t＠philips.com 02 Feb '05

02 Feb '05

hi, I was trying to use mpl::contains with list_c and it always seems to return mpl::false_. For example typedef mpl::list_c< int ,16,8 > mylist; typedef mpl::contains<mylist, mpl::int_<16> >::type tt; "tt::value" is zero though i thought 16 being in the mylist it would be 1. Can someone please let me know whether the usage is improper and if so what should be used to find elements in list_c? Thanks in Advance S.T

2 1

Exercices 3-3 and 3-4 (C++ Templates Metaprogramming)
by Philippe Mori 02 Feb '05

02 Feb '05

I'm not sure of the meaning of "twice twice" and "twice on itself" respectively. My guess would be for exercice 3 that I should do something like : typedef twice<to_be_filled_1>::type T1; // T1 would be T ** // The second time, I replace T (in to_be_fiiled_1) by // T1 (in to_be_filled_2) , thus T2 is T1** which is also T**** typedef twice<to_be_filled_2>::type T2; while for exercice 4, we want something that would look like : typdef twice<twice<to_be_filled_3a>to_be_filled_3b>::type T3; Does I understand the questions correctly? For twice, I use the definition from section 3.3.2. Philippe

2 1