- Boost-users - lists.boost.org

[Boost-Users][serialization] Floating point precision
by jsanga＠cox.net 28 Feb '07

28 Feb '07

Is there a way to specify the precision of doubles when writing to xml_oarchive? Refer to the output snippet below. <Control class_id="3" tracking_level="0" version="1"> <Type>0</Type> <LowMZ>5.0499999999999998</LowMZ> <HiMZ>5.0999999999999996</HiMZ> I would like 5.05 and 5.1 for Low and High. Stepping through the code with the debugger I ended up in basic_text_oprimitive::save( const double t). void save(const double t) { if(os.fail()) boost::throw_exception(archive_exception(archive_exception::stream_error)); os << std::setprecision(std::numeric_limits<double>::digits10 + 2); os << t; } where it seems the precision is fixed. The only solution I can seem to think of is to split the serialization, and on the output convert my double to string (with the correct precsion) and then serialize the string. Is there a different or better way?

2 1

[Boost.MultiIndex] Problem using dataelement from derived class as key
by Manuel Jung 28 Feb '07

28 Feb '07

Hi, Im having a problem. I derive class B from class A. class a has the member A::id. I want to use this one as key for my MultiIndexContainer. But i get the following compiling errors: test.cpp:41: Fehler: Templateargument »&A::id« konnte nicht in »unsigned int B::*« umgewandelt werden test.cpp:42: Fehler: Templateargument 2 ist ungültig test.cpp:43: Fehler: Templateargument 1 ist ungültig test.cpp:44: Fehler: Templateargument 2 ist ungültig Sorry it on german. A testfile for reproducing my problem is attached. Can someone help? thanks.. Greetings Manuel Jung

2 3

[boost][circular_buffer] problem using copy contructor
by Andreas Rose 28 Feb '07

28 Feb '07

Hello, I'm new to circular_buffer. After reading the documentation i expect the following should work: const int c = 5; boost::circular_buffer<int> cb1(c); assert(cb1.capacity() == c); boost::circular_buffer<int> cb2(cb1); assert(cb2.capacity() == c); //failed, capacity() returns 0 But it does'nt work. After looking into the code of circular_buffer, i think the copy c'tor should be changed into something like this: circular_buffer(const circular_buffer<T, Alloc>& cb) : m_size(cb.size()), m_alloc(cb.get_allocator()) { initialize(cb.capacity()); BOOST_TRY { cb_details::uninitialized_copy(cb.begin(), cb.end(), m_buff, m_alloc); } BOOST_CATCH(...) { deallocate(m_buff, cb.capacity()); BOOST_RETHROW } BOOST_CATCH_END } Or do i missunderstand the documentation? I'm using vs2003 and the latest version of circular_buffer from boost-sandbox in ordinary debug mode. Thanks. Andreas Rose

1 0

Problem using regex library in Windows/cygwin
by Diego Molla 28 Feb '07

28 Feb '07

Hi, I have a problem using the Boost regex library. I have isolated the problem to the attached piece of code. For some reason the output returned by the function that uses regex is garbage. I have tested the same program on a Linux version of boost and it works fine. I wonder if the problem is on the Windows installation. Could someone using Windows or Cygwin test this and tell me if it works? I've just re-installed the cygwin version (1.33.1-3) and I still get the same error. Here is a typical interaction. It should extract the XML attribute and value but instead it returns garbage: [diego@valencia ~...af/src]$ g++ -o regexError.exe regexError.cpp -lboost_regex-gcc-mt [diego@valencia ~...af/src]$ regexError Write a fragment of XML: this <test arg="a1">is a test</test> yeah Write the XML tag to search: test Found element is a test Searching arg="a1" Found attribute 1g with value a1 Write a fragment of XML: this <test a2="more">is another test</test> Write the XML tag to search: test Found element is another test Searching a2="more" Found attribute or with value morY Write a fragment of XML: end [diego@valencia ~...af/src]$ -- Do you want to know what is going on in Language Technology in Australia and New Zealand? Join ALTA -- http://www.alta.asn.au/ This message is intended for the addressee named and may contain confidential information. If you are not the intended recipient, please delete it and notify the sender. Views expressed in this message are those of the individual sender, and are not necessarily the views of Macquarie University. --------------------------------------------------------------------- Diego MOLLA ALIOD diego(a)ics.mq.edu.au Department of Computing http://www.ics.mq.edu.au/~diego Macquarie University

2 1

[graph] patch to automatically build libbgl-viz.so and libbgl-viz.a
by David Baird 28 Feb '07

28 Feb '07

Hi, I'm interested in helping make the BGL easier to use right out-of-the-box. But, I'm not an expert at Boost (or Jamfiles) yet, so please guide me if this is wrong. I tried to patch the Jamfile so that it automatically builds the Graphgiz parser libraries (libbgl-viz.so and libbgl-viz.a). I only tested Jamfile (not Jamfile.v2) on some Gentoo boxes with gcc-4.1.1 and gcc-3.3.5 and it worked nicely. Some other future improvements I would like to do are to improve the Graphviz parser (to be able to parse more varieties of .dot files) and to review the documentation and check it for correctness (I already submitted some patches about half a year ago for a couple mistakes that I found). Please let me know what you think. Without further ado, here's the patch: Index: boost_1_33_1/libs/graph/build/Jamfile =================================================================== --- boost_1_33_1.orig/libs/graph/build/Jamfile +++ boost_1_33_1/libs/graph/build/Jamfile @@ -33,6 +33,21 @@ GRAPHVIZ_DEFS(graphviz_digraph_parser) = } local gGENERATOR_FUNCTION(LIB) = bgl-lib-file ; + dll bgl-viz + : ../src/$(SOURCES).cpp + : <sysinclude>$(BOOST_ROOT) <define>YY_NO_UNISTD_H + <runtime-link>dynamic + <define>BOOST_ALL_DYN_LINK=1 + # Darwin compiler ICEs if we turn optimization on + <darwin><*><debug-symbols>off + <darwin><*><optimization>off + # Intel compiler ICEs if we turn optimization on + <intel-win32><*><optimization>off + <vc-7_1><*><cxxflags>-GR- + : debug <inlining>on #inlining prevents linker name clashes due to truncation! + <debug-symbols>off # Debugging symbols become HUGE + ; + lib bgl-viz : ../src/$(SOURCES).cpp : <sysinclude>$(BOOST_ROOT) <define>YY_NO_UNISTD_H @@ -47,4 +62,7 @@ GRAPHVIZ_DEFS(graphviz_digraph_parser) = ; + install bgl-viz lib + : <dll>bgl-viz <lib>bgl-viz + ; } Index: boost_1_33_1/libs/graph/build/Jamfile.v2 =================================================================== --- boost_1_33_1.orig/libs/graph/build/Jamfile.v2 +++ boost_1_33_1/libs/graph/build/Jamfile.v2 @@ -24,3 +24,5 @@ obj graphviz_digraph_lex : graphviz_lex. <implicit-dependency>graphviz_digraph_parser ; obj graphviz_digraph_parser : graphviz_parser.yy : <bison.prefix>bgl_dir_ <define>GRAPHVIZ_DIRECTED=1 ; + +boost-install bgl-viz ;

1 1

[program_options] unsigned variables silently cast
by Torrey Pines 28 Feb '07

28 Feb '07

Hi With the current implementation, unsigned variables accept negative values, that get silently cast. E.g. on my platform, if "--length" expects size_t, --length -32 assigns it a value 4294967264 without warnings or errors. This behavior surprised me, and I'm using a workaround (parsing an int, and checking its range manually). However, is this an intended feature of the library? Thanks

2 1

[program_options] disable partial matches
by Torrey Pines 28 Feb '07

28 Feb '07

Hi With the current implementation, the user does not have to supply the full option names, e.g. "--he" is as good as "--help", unless there is an ambiguity. Is there any way to disable this, i.e. allow only full matches? Thanks

2 1

[program_options] checking the presence of required arguments
by Torrey Pines 28 Feb '07

28 Feb '07

Hi My program has a bunch of required arguments. I grouped them into one options_description: options_description required("Required options"); required.add_options() /* .... */ ; // ... options_description desc("Allowed options"); desc.add(required).add(other); I'd like to check that all required arguments are in fact supplied without having to do if(vm.count("arg1") == 0) { cerr << "usage error: arg1 is required\n"; return 1; } for each of them. Any suggestions? Thanks

2 1

boost::any template argument deduction
by Digvijoy Chatterjee 28 Feb '07

28 Feb '07

Hi, I have a function template like : template <class A> A what_type ( const boost::any &operand ){ A a; a = boost::any_cast<a>(operand); return a; } The idea is to provide the client code an easy way extract the type hidden by the boost::any value, Now there is no way to make the compiler deduce the type to return from an expression like below: std::string s = "hello"; boost::any a = s; some_op( what_type(a) ); //calls some_op(string &); Instead now the client has to write code using type() which returns std::typeinfo & if ( a.type().name() == "Ss" ) some_op(what_type<string> a); else if ( a.type().name() == "d" ) some_op(what_type<double> a); Or use functions to test type and cast: if ( is_string (a) ) some_op( what_type<string>(a)); else ( is_int(a) ) some_op ( what_type<int>(a)); ….. Is there any elegant way to handle this problem ?? --Digz

1 0

[MPL] Problem with vector
by Daniel 28 Feb '07

28 Feb '07

Hi, I have few Adapter classes. Each of them provide typedef operations which specifies which operations that Adapter supports (vector_c, with 0 or 1 at appropriate positions). I want to assert that two Adapters support different operations. However, my code doesn't compile. Any ideas? Daniel template<unsigned int N1, unsigned int N2> struct Adapter { typedef vector_c<unsigned int, N1, N2> operations; }; template<class TAdapter1, class TAdapter2> struct Check { typedef typename transform< typename TAdapter1::operations, typename TAdapter2::operations, plus<_1, _2> >::type sum_of_operations; typedef typename max_element<sum_of_operations>::type max_of_operations; BOOST_MPL_ASSERT((less_equal<max_of_operations, int_<1> >)); }; int main() { Check< Adapter<0, 1>, Adapter<1, 0> >(); return 0; }

2 1