December 2009 - Boost-users

[mpl] Strange integral constant overflow warning in mpl::less_equal
by Edward Diener 19 Dec '09

19 Dec '09

This simplified version of some TMP code I am writing using mpl and integer_traits is giving me a strange warning. I'm using VC9 and Boost 1.40: ------------------------------------------------------------------------------------- struct ResultTypeIntegerValues { typedef boost::mpl::long_<0L> lzero; typedef boost::mpl::long_<boost::integer_traits<long>::const_min> lmin; typedef boost::mpl::less_equal<lmin,lzero>::type type; }; ------------------------------------------------------------------------------------- with warning message: ------------------------------------------------------------------------------------- c:\utilities\boost\boost_1_40_0\boost\mpl\aux_\integral_wrapper.hpp(73) : warning C4307: '-' : integral constant overflow 1> c:\utilities\boost\boost_1_40_0\boost\mpl\aux_\preprocessed\plain\less_equal.hpp(60) : see reference to class template instantiation 'boost::mpl::long_<N>' being compiled 1> with 1> [ 1> N=-2147483648 1> ] 1> c:\utilities\boost\boost_1_40_0\boost\mpl\aux_\preprocessed\plain\less_equal.hpp(70) : see reference to class template instantiation 'boost::mpl::less_equal_tag<T>' being compiled 1> with 1> [ 1> T=nmspace::ResultTypeIntegerValues::lmin 1> ] 1> c:\programming\programs\xxx\yyy.h(64) : see reference to class template instantiation 'boost::mpl::less_equal<N1,N2>' being compiled 1> with 1> [ 1> N1=nmspace::ResultTypeIntegerValues::lmin, 1> N2=nmspace::ResultTypeIntegerValues::lzero 1> ] ------------------------------------------------------------------------------------- As far as I can see I am doing a perfectly valid comparison. Is this a minor bug in mpl or am I doing something wrong or should this warning just be expected ? Warning C4307 gives: ------------------------------------------------------------------------------------- Error Message 'operator' : integral constant overflow The operator is used in an expression that results in an integer constant overflowing the space allocated for it. You may need to use a larger type for the constant. A signed int holds a smaller value than an unsigned int because the signed int uses one bit to represent the sign. The following sample generates C4307: // C4307.cpp // compile with: /W2 int i = 2000000000 + 2000000000; // C4307 int j = (unsigned)2000000000 + 2000000000; // OK int main() { } -------------------------------------------------------------------------------------

2 2

Re: [Boost-users] [Boost.Test] valgrind complains about invalid reads
by Scott Banachowski 18 Dec '09

18 Dec '09

--- Steven Watanabe wrote: > joel falcou wrote: > > I'm jumping in the wagon but this very piece of code > leads to a > > "incorrect free" on Snow Leopard using g++ 4.2.1 > Apple. > > Maybe it is related to the virtual matter > > We've already discussed this many times. The code above > is wrong for reasons unrelated to the use of a virtual > destructor, > although adding a virtual destructor masks the problem. > This has been fixed in the trunk. Thanks Steven, with some searching I think I found the patch you are referring to: https://svn.boost.org/trac/boost/changeset/56467 Seeing this patch, I now see how the virtual function is a red herring (although I don't know why it happens to work, in this case). Nevertheless, I still like the idea of making it virtual, just to avoid the necessity of switching on type and casting to delete, but that's my opinion. Scott

1 0

sizeof of boost::int8_t
by eial＠cs.bgu.ac.il 18 Dec '09

18 Dec '09

running sizeof on a boost::int8_t returns 1, but what is the type? int/char? thanks.

8 7

[containers] defining the insertion point of non-unique keys
by Olaf Krzikalla 18 Dec '09

18 Dec '09

Hi @boost, I'm looking for the solution to the following problem: there is a collection of objects and I want to find a particular object by two different keys. The first key is unique, so there is no problem. The second key however is non-unique and if I search for an object using this key I want to find the element that was inserted last in that sequence. Neither boost::multindex nor boost:intrusive seems to have solutions right at hand. If using boost::multindex I could try to insert a new element in the non-unique sequence using the lower_bound iterator of that sequence as a hint. However it is not documented that in that case the element is always inserted before the hint. Same is true for boost:intrusive::multi_set/multi_map. And I'm not able to swap the position of two adjacent elements in a multi_set or multi_map, if they have equal keys. To illustrate the problem, here is some code: struct A { int a; // non-unique key void* b; // unique key // stuff }; using namespace boost::multi_index; typedef multi_index_container<A, indexed_by< ordered_non_unique<member<A,int,&A::a> >, ordered_unique<member<A,void*,&A::b> > > > tCollection; void foo(tCollection& collection, int key) { typedef tCollection::nth_index<0>::type tItByInt; tItByInt i = collection.get<0>().lower_bound(key); // use the current lower bound as a hint: tItByInt insertPos = collection.get<0>.insert(i, getObject(key)); // the big question: does the assertion holds? assert(collection.get<0>().lower_bound(key) == insertPos); } Best regards Olaf Krzikalla

4 9

Re: [Boost-users] Boost Graph and GraphViz - Names and Subgraphs
by Graham Hemingway 18 Dec '09

18 Dec '09

I am building graphs with boost::graph and want to generate GraphViz output. Most everything works as expected, but I am having trouble getting two things to work correctly: 1) My subgraphs are not visualized in the GraphViz output 2) I am setting the graph_name property of both the root graph and all subgraphs, but the names never appear in the GraphViz output. Any help would be appreciated. Cheers, Graham

1 0

[phoenix] Staus of phoenix
by Edward Diener 18 Dec '09

18 Dec '09

Is phoenix an official Boost library ? I do not see it listed among the Boost libraries for 1.40 or 1.41 although I do see it in the pdf documentation for Boost libraries put together by John Maddock.

6 11

Template type questions
by Glenn Macgregor 18 Dec '09

18 Dec '09

Hi All, I am using the boost::variant with the binary visitor algorithm, apply_visitor. struct UnknownValue {}; typedef boost::variant<UnknownValue, long, unsigned long, double, string, char> ValueType; I am using apply_visitor for binary visitation as below: class AddVisitor : public boost::static_visitor<> { public: template <typename T> void operator()(const T &lhs, const T &rhs) { T val = lhs + rhs; } template <typename T, typename R> void operator()(const T &lhs, const R &rhs) { // do nothing...different types } template <typename T> void operator()(const T &lhs, const UnknownValue &rhs) { // do nothing...we have an unknown value } template <typename T> void operator()(const UnknownValue &lhs, const T &rhs) { // do nothing...we have an unknown value } void operator()(const UnknownValue &lhs, const UnknownValue &rhs) { // do nothing...we have an unknown value } }; The above is not complete because I can do addition if the types are long and unsigned long or long and double or long and char...etc. What is the preferred method for doing this? Do I need to create an overridden operator() for every argument combination that is valid or can I get the argument types in the function? Thanks Glenn

2 1

[spirit] help parsing data file into a struct
by Kenny Riddile 18 Dec '09

18 Dec '09

This is my first time using Spirit and I haven't messed with writing EBNF grammars since college, so I'm sorry if this is a simple question. Lets say I have a structure that looks like this: struct Foo { std::string someString; std::string someOtherString; std::vector<std::string> stringList; }; BOOST_FUSION_ADAPT_STRUCT( Foo, (std::string, someString) (std::string, someOtherString) (std::vector<std::string>, stringList) ) and I have a data file that looks like this: SOMESTRING hello SOMEOTHERSTRING goodbye STRINGLIST { string1 string2 string3 } I have written the following grammar to parse such a file into the struct above: template< typename Iterator > struct FooDefinition : spirit::qi::grammar< Iterator, Foo() > { FooDefinition() : FooDefinition::base_type( start ) { using namespace spirit::qi; start = someString >> +space >> someOtherString >> +space >> stringList; someString = lit("SOMESTRING") >> +space >> value; someOtherString = lit("SOMEOTHERSTRING") >> +space >> value; stringList = lit("STRINGLIST") >> *space >> lit('{') >> *space >> (value % +space) >> *space >> lit('}'); value = +char_("a-zA-Z_0-9"); space = lit(' ') | lit('\n') | lit('\t'); } spirit::qi::rule< Iterator, Foo() > start; spirit::qi::rule< Iterator, std::string() > someString; spirit::qi::rule< Iterator, std::string() > someOtherString; spirit::qi::rule< Iterator, std::vector<std::string>() > stringList; spirit::qi::rule< Iterator, std::string() > value; spirit::qi::rule< Iterator > space; }; However, I would like the order of the three statements in the data file (SOMESTRING, SOMEOTHERSTRING, and STRINGLIST) to be irrelevant. Right now, they must be in the order shown above for parsing to succeed. How would I go about accomplishing this with Spirit?

2 2

inheritance, input streams, parsing and arithmetic range of values
by Hicham Mouline 18 Dec '09

18 Dec '09

Hello I would like your opinions about design/implementation choices for an application I am writing. I have a list of models. Each model is a kind of black box that takes inputs and produces outputs. All the inputs are either of boolean or arithmetic type. (bool, int, double) A subset of inputs is valid for all models. Then different models have different number of other inputs. For a fixed set of inputs, the model produces 1 fixed set of outputs. It is desired to vary the values of some/all (at least 1) of the inputs and get different outputs sets, and then plot that. The inputs are specified in files (considered as std::istream in the code), the syntax is trivial: input1 = value1 ... inputn = valuen When it is desired to vary the input values, the syntax changes a little: input1 = value1 input2 = value21..value22 ... inputn = valuen in this example only input2 changes, and the idea is to plot one of the outputs over the differnt values of input2. Each derived model has a concrete struct input_i_s { ... } that holds its inputs. class model { public: virtual fixed_inputs get_fixed_inputs( std::istream& ) const =0; /// throws if syntax error, or not all inputs are specified, or some values are not valid virtual variable_inputs get_variable_inputs( std::istream& ) const =0; /// the same, allows the '..' notation though for ranges, at least of the inputs need to be a range }; // fixed_inputs is a pointer to a base class of all inputs for all models 1 implementation pseud-code looks like: fixed_inputs model_i::get_fixed_inputs( std::istream& input) const { size_t inputs = 7; std::string buffer; while (inputs>0 && input.good()) { input >> buffer; if ( buffer.substr(0, sizeof("input1")) == "input1=") { std::istringstream is(buffer.substr(0, sizeof("input1"))); is >> .... ; /// pseudo-code : stores the value into a field in fixed_input1s // if wrong value, throws an exception } else if ... ... } return fixed_inputs( new input_i_s() ); // pseudo-code, say for model number i } I would like then to write the implementation for the syntax which accepts ranges as well. (the get_variable_inputs function) I hope to be able to factorize code between the 2 implementations. 1. boost::program_options can read options from input streams. Given that I am interested in ranges as well, is it adviced to go down that road? 2. Is there a library that helps to generate the expected format of the input stream from the struct of inputs for each model, and reads the stream automatically. Can it do range values as well? 3. is it possible then to factorize code between the get_fixed_inputs and get_variable_inputs? Regards,

1 0

(no subject)
by Michael Fowler 18 Dec '09

18 Dec '09

Hello. I tried using the code provided here: http://www.boost.org/doc/libs/1_41_0/doc/html/boost_asio/example/echo/block… to echo text back to a client but i haven't been successful. It seems to hang at size_t length = sock->read_some(boost::asio::buffer(data), error); what is wrong? Thanks

1 0