January 2008 - Boost-users - lists.boost.org

iostreams::stream: flush doesn't work but endl does
by Rob Desbois 23 Jan '08

23 Jan '08

Hi all, I've created my own iostreams device (for a serial port) which for debugging purposes prints out calls to read()/write(). I'm constructing a stream on the device like this: serial_device serial("/dev/ttyS3", serial_device::Baud38400); stream<serial_device> s(serial); If I do this: s << "hello world" << endl; serial.write() is called and does its debug output, but with this: s << "hello world\r" << flush; I get nothing despite using 'flush'. I've also tried calling s.flush() manually but get the same lack of effect. How come flush isn't working as it should here? TIA, --rob -- Rob Desbois

2 2

Solve linear system (tridiagonal matrix)
by nisha kannookadan 23 Jan '08

23 Jan '08

Hi everybody I need to solve a linear system Ax = b, where A is a tridiagonal matrix, is there an easy way to do it? I programmed an algo, which computes the inverse of the tridiagonal matrix with the usmani-algo and uses prod then, but this is quite time intense. I want to use the lapack bindings, but I dont get along with it. Im bit confused, if I need to install something additional to it, the way I tried out gesv, its not working and I get error msgs. I searched through out the www, but its hard to find good info. If somebody has already the code to solve linear equations with lapack, I would thankful, if he could post it here. Thank you Nisha K _________________________________________________________________ Express yourself instantly with MSN Messenger! Download today it's FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/

2 1

Property Tree
by Glasgow, Craig C 23 Jan '08

23 Jan '08

Hello I'm using boost property_tree with the variant data type: typedef boost::variant<std::string, std::vector<std::string> > registry_entry; i.e I want to store either strings or vectors of strings at each key node. I'm using the inserters and extractors at the bottom of this email (similar to those included in the custom_data.cpp in the property_tree/examples folder). Does anyone know how I can write the tree out in xml format? The provided write_xml method can only handle property trees of string data type. Is there anyway to either tweak the code (e.g. modify the inserter and extractors) or wrap the variant class such that property_tree::write_xml will correctly render the tree as xml? Regards Craig // Custom extractor - converts data from boost::any to Type template<class Type> struct pt_extractor { inline bool operator()(const boost::any &data, Type &extracted, const std::locale &loc) const { extracted = boost::any_cast<Type>(data); return true; // Success } }; // Custom inserter - converts data from Type to boost::any template<class Type> struct pt_inserter { inline bool operator()(boost::any &data, const Type &to_insert, const std::locale &loc) const { data = to_insert; return true; // Success } }; // Custom property_tree traits - using custom extractor and inserter template<class Ch> struct pt_traits { // Character type to be used by ptree keys typedef Ch char_type; // Key type to be used by ptree typedef std::basic_string<Ch> key_type; // Data type to be used by ptree typedef boost::any data_type; // Extractor to be used by ptree template<class Type> struct extractor: public pt_extractor<Type> { }; // Inserter to be used by ptree template<class Type> struct inserter: public pt_inserter<Type> { }; // Key comparison function inline bool operator()(const key_type &key1, const key_type &key2) const { return _stricmp(key1.c_str(), key2.c_str()) < 0; } }; typedef boost::property_tree::basic_ptree<pt_traits<char> > pt_rep; typedef boost::variant<std::string, std::vector<std::string> > registry_entry; ***************************************************************************** This communication is sent by the Standard Bank Plc or one of its affiliates The registered details of Standard Bank Plc are: Registered in England No. 2130447, Registered Office 25 Dowgate Hill London EC4R 2SB Authorised and Regulated by the Financial Services Authority. More information on Standard Bank is available at www.standardbank.com Everything in this email and any attachments relating to the official business of Standard Bank Group Limited and any or all subsidiaries, the Company, is proprietary to the Company. It is confidential, legally privileged and protected by relevant laws. The Company does not own and endorse any other content. Views and opinions are those of the sender unless clearly stated as being that of the Company. The person or persons addressed in this email are the sole authorised recipient. Please notify the sender immediately if it has unintentionally, or inadvertently reached you and do not read, disclose or use the content in any way and delete this e-mail from your system. The Company cannot ensure that the integrity of this email has beenmaintained nor that it is free of errors, virus, interception or interference. The sender therefore does not accept liability for any errors or omissions in the contents of this message which arise as a result of e-mail transmission. If verification is required please request a hard-copy version. This message is provided for informational purposes and should not be construed as a solicitation or offer to buy or sell any securities or related financial instruments. *****************************************************************************

1 0

asio socket options
by Eugene M. Kim 23 Jan '08

23 Jan '08

Greetings, On UNIX-like platforms, boost::asio::detail::socket_option::{boolean,integer} templates are very useful when defining socket option classes for non-standard options that are supported by the platform, e.g.: namespace so = boost::asio::detail::socket_option; typedef so::boolean<SOL_SOCKET, SO_REUSEADDR> reuse_addr; typedef so::boolean<SOL_SOCKET, SO_REUSEPORT> reuse_port; typedef so::boolean<SOL_SOCKET, SO_ERROR> pending_error; typedef so::integer<SOL_SOCKET, SO_SNDLOWAT> send_low_watermark; typedef so::integer<SOL_SOCKET, SO_RCVLOWAT> recv_low_watermark; typedef so::boolean<IPPROTO_IP, IP_HDRINCL> include_header; typedef so::boolean<IPPROTO_IP, IP_DONTFRAG> do_not_fragment; typedef so::integer<IPPROTO_IP, IP_TOS> type_of_service; typedef so::integer<IPPROTO_IP, IP_TTL> time_to_live; ... and so on. It is not suitable to include these non-standard options in the asio library itself, but it seems justifiable to move the boolean/integer templates out of boost::asio::detail and into boost::asio. Also, many socket options take a C struct as an argument, so a template for such options might be useful as well: template <int Level, int Name, class Value> class typed : public Value { typedef Value value; template <typename Protocol> int level(const Protocol &) const { return Level; } template <typename Protocol> int name(const Protocol &) const { return Name; } template <typename Protocol> value *data(const Protocol &) { return (value *)this; } template <typename Protocol> std::size_t size(const Protocol &) const { return sizeof(Value); } template <typename Protocol> void resize(const Protocol &, std::size_t s) const { if (s != sizeof(Value)) throw std::length_error("resizing fixed-length socket option"); } }; Then it becomes trivial to define and use such socket options: typedef typed<SOL_SOCKET, SO_SNDTIMEO, struct ::timeval> send_timeout; typedef typed<SOL_SOCKET, SO_RCVTIMEO, struct ::timeval> recv_timeout; send_timeout sndtimeo; sndtimeo.tv_sec = 1; sndtimeo.tv_usec = 0; Would this template be appropriate for inclusion in a future version of asio? Just my $.02. Cheers, Eugene

1 0

meta-programming library
by noman javed 23 Jan '08

23 Jan '08

Hi I need to work with boost-mpl library. Should I install the whole boost library or is there any possibility to just install boost-mpl ? Thanks in anticipation Noman Javed --------------------------------- Be a better friend, newshound, and know-it-all with Yahoo! Mobile. Try it now.

2 1

The Scanner Business
by jono 23 Jan '08

23 Jan '08

hello possums! i have the following rules declared in a spirit grammar definition (using boost 1.8.4 )..... //----------------------- typedef typename lexeme_scanner<ScannerT>::type lexeme_scanner_t; rule<lexeme_scanner_t> legal_initial_p ,legal_p ; //----------------------- and initialized in definition::definition() .... //----------------------- legal_initial_p = alpha_p | ch_p('_') |'<'|'>'|'='|'?'|'/'|'*'|'&'|'-'|'+'|'%' |'!'|'$'|':'|'@' ; legal_p = legal_initial_p | digit_p | ch_p('.') ; //----------------------- the above code compiles and functions as expected under MSVC 7.1 but causes g++ 4.1.1 to hang. if i use type rule<ScannerT> for legal_p and legal_initial_p, the code compiles on both, but doesn't work properly. i'm using |dynamic_distinct_parser ||so i can handle the '.' character as a syntactic token in certain contexts. |||//---------------------- |some declarations ... || struct keywords : symbols<Object> { ... } keywords_p; || ||dynamic_distinct_parser<ScannerT> keyword_p; ||||| //----------------------|some definitions ...| | ||chset<> distinct(chset<>(legal_ident) | chset<>("0-9")); distinct_directive<> keyword_d(distinct); distinct_parser<> keyword_p(distinct); |||| || ||rule<ScannerT, object_closure::context_t> ||Symbol_p = lexeme_d [ ( legal_initial_p >> *legal_p ) - keywords_p ] [ Symbol_p.value_ = intern_(arg1, arg2) ] ; || ||| dot_p = keyword_p("."); //---------------------- |// .... ||it's quite possible there's a better way to do the parsing, but it was fiddly, and i don't have any more time to mess with it. (the rest of the grammar is a recursive scheme reader, hence the closures) i would be very grateful for any suggestions as to what might be causing g++ to go walkabout! cheers, jono jp(a)day-one.com | | |

1 0

Regex newbie problem
by Barberi 23 Jan '08

23 Jan '08

Newbie to boost. I used the MSVC 8.0 installer, and got the first couple of example programs to work, but failed on the regex example. I'm using MSVC 2008 (aka MSVC-9.0) (Express Edition) IDE, in Debug configuration. The linker can't resolve several symbols with libboost_regex_vc80-mt-gd-1_34_1.lib. Stripped of the qualifiers these are: _Orphan_all(void) const, ~_Container_base(void), _Swap_all(...), and insert(...). I switched to release configuration, now the linker looks in libboost_regex_vc80-mt-1_34_1.lib and can't find insert(...). Any suggestions? Should I wait for a MSVC-9.0 installer, and go back to MSVC 2005 in the meantime? Stephen B.

3 8

Boost, HP-UX and gcc
by Martin_Thomas＠McAfee.com 23 Jan '08

23 Jan '08

Hello, I am trying to use use Boost::Regex on HP 11i with gcc 4.2.2 and I notice that in config/platform/hpux.hpp there is a define to turn off wide-char support unless using aCC and have a1 source eg: #if !(defined(__HP_aCC) || !defined(_INCLUDE__STDC_A1_SOURCE)) # define BOOST_NO_SWPRINTF # define BOOST_NO_CWCTYPE #endif This will never allow gcc to use wregex even if wchar support is available. Is this because there are so few people using Zgcc oin HP-ux or am I missing something? On the assumption that it is related to usage, I have added the following to allow gcc 4 to use wregex: #if defined(__GNUC__) # if (__GNUC__ >3) # undef BOOST_NO_CWCTYPE # endif #endif Is there anything obviously wrong with this? Thanks // Martin

4 8

[program_options] crash on application exit
by ope 22 Jan '08

22 Jan '08

Hi, to get the main() more clean, I moved the command line options parser to an own class (some stuff missing): ---8--- class CommandLineOptions { public: CommandLineOptions( int argc, char** argv ); private: po::options_description generic_options; po::options_description qt4_common; #if defined( Q_OS_UNIX ) || defined( Q_OS_MAC ) po::options_description qt4_x11; #endif #if defined( Q_OS_WIN ) po::options_description qt4_win; #endif #if defined DEBUG po::options_description qt4_dbg; #endif po::options_description hidden_options; po::options_description usage_options; po::options_description cmdline_options; po::positional_options_description pos_options; po::variables_map vm; }; --->8--- with ---8<--- CommandLineOptions::CommandLineOptions( int argc, char** argv ) : generic_options("Generic options"), qt4_common("Qt options"), #if defined( Q_OS_UNIX ) || defined( Q_OS_MAC ) qt4_x11( "Qt X11 specific options" ), #endif #if defined( Q_OS_WIN ) qt4_win( "Qt Windows specific options" ), #endif #if defined( DEBUG ) qt4_dbg( "Qt debug options" ), #endif hidden_options("Hidden options") { generic_options.add_options() ( "version,v", "print version string") ( "help,h", "produce help message") ; hidden_options.add_options()(...); qt4_common.add_options()(... #if defined( Q_OS_UNIX ) || defined( Q_OS_MAC ) qt4_x11.add_options()(... #if defined( Q_OS_WIN ) qt4_win.add_options()(... #if defined( DEBUG )(... usage_options .add( generic_options ) .add( qt4_common ) #if defined( Q_OS_UNIX ) || defined( Q_OS_MAC ) .add( qt4_x11 ) #elif defined( Q_OS_WIN ) .add( qt4_win ) #endif #if defined( DEBUG ) .add( qt4_dbg ) #endif ; cmdline_options.add( usage_options ).add( hidden_options ); pos_options.add( "input-file", -1 ); po::store(po::command_line_parser(argc, argv). options(cmdline_options).positional(pos_options).run(), vm); po::notify(vm); if (vm.count( "version" )) { std::cout << Version() << "\n"; } ... } --->8--- and the main(): ---8<--- int main( int argc, char **argv ) { try { logging::init(); CommandLineOptions cmdline( argc, argv ); QT_REQUIRE_VERSION( argc, argv, QT_VERSION_STR ); Q_INIT_RESOURCE( workbench ); QApplication app( argc, argv ); QMainWindow wb; wb.show(); // using boost.logging v2 L_ << "execute application."; return app.exec(); ... --->8--- It works fine until the app exits. My gdb session wasn't very useful to me: (gdb) target exec ./edas (gdb) run Starting program: /.../edas2/edas Using host libthread_db library "/lib/i686/cmov/libthread_db.so.1". (no debugging symbols found) (no debugging symbols found) (no debugging symbols found) (no debugging symbols found) [Thread debugging using libthread_db enabled] [New Thread 0xb70506c0 (LWP 13445)] [New Thread 0xb6aa5b90 (LWP 13450)] 28:31 [1] execute application. Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 0xb70506c0 (LWP 13445)] 0x08051980 in ?? () (gdb) backtrace #0 0x08051980 in ?? () If I comment out my CommandLineOptions class instance it works. Thanks, Olaf

1 0

boost::filesystem::rename - behavior/code vs. docs
by Andriy Gapon 22 Jan '08

22 Jan '08

This is what documentation on boost::filesystem::rename states: http://www.boost.org/libs/filesystem/doc/tr2_proposal.html template <class Path1, class Path2> void rename(const Path1& from_p, const Path2& to_p); Requires: Path1::external_string_type and Path2::external_string_type are the same type. Effects: Renames from_p to to_p, as if by POSIX rename(). Postconditions: !exists(from_p) && exists(to_p), and the contents and attributes of the file originally named from_p are otherwise unchanged. [Note: If from_p and to_p resolve to the same file, no action is taken. Otherwise, if to_p resolves to an existing file, it is removed. A symbolic link is itself renamed, rather than the file it resolves to being renamed. -- end note] Actual effect is that an exception is thrown when renaming a file to an existing file on the same fs. rename(2) performs an 'atomic' rename in this case without problems, but in the boost code there is a specila check against "too permissive" posix rename(): http://boost.cvs.sourceforge.net/boost/boost/libs/filesystem/src/operations… 1126 rename_api( const std::string & from, const std::string & to) 1127 { 1128 // POSIX is too permissive so must check 1129 fs::system_error_type dummy; 1130 if ( fs::exists( status_api( to, dummy ) ) ) 1131 return EEXIST; 1132 return std::rename( from.c_str(), to.c_str() ) != 0 1133 ? errno : 0; 1134 } And rename testing code also checks for the above behavior. I think that this is a huge difference between the docs and the code, and this is very counter-intuitive too. -- Andriy Gapon

1 0