August 2006 - Boost-users - lists.boost.org

Re: [Boost-users] Using regex on a file
by Todd Greer 31 Aug '06

31 Aug '06

Dave wrote: > My first thought was to use regex_search and pass it two stream iterators, > one pointing to the beginning of the file and one representing the end of > the file. Alas, this won't work since regex_search requires bidirectional > iterators, which stream iterators are not. You might consider using a file_iterator from Spirit. It is a random-access iterator for files, which sounds like exactly what you need. Check it out at http://www.boost.org/libs/spirit/doc/file_iterator.html. -- Todd

1 0

Problem building 1.33.1 for Intel Mac -- Seeking patches if they exist.
by Bennett Smith 30 Aug '06

30 Aug '06

Hi, I am attempting to build Boost 1.33.1 for an Intel Mac. Some of the libraries do not build successfully, and I think it is related to the different architecture. One of the errors I get is: ld: Undefined symbols: __Z9test_mainiPPc Does anyone have patches for Boost 1.33.1 that would make it build everything successfully on an Intel Mac? BTW, I also tried doing a build of the 1.34.0 release candidate files and everything there seems to build fine. I just don't really want to base our project on 1.34 until it is officially released. Thanks for any help you can offer. Cheers -- Bennett

1 0

Re: [Boost-users] Mismatch and regex newbie problem still problem -- FINAL NOTE
by james.jones＠firstinvestors.com 30 Aug '06

30 Aug '06

From: "david v" <danova_fr(a)hotmail.com> > Thanks to all for your comments.... > It's a pity this class of functionality is not implemented in boost. I'm in > the bioinformatics field and starting to work with boost. The use of regex > in general as boost implements is quite common in the field and boost is > very handy.!!! However at some point some advanced functionalities with > insertion,deltions and repeats are very handy to use. Maybe there's a chance > that some of the work out there get into boost as a boost_bio module ??? > Note that biocpp.sourceforge.net has done some work on bioinformatics and > that maybe it would be interesting to know if they were willing to move > their codeinto boost. This will avoid people from having to install > thousands of libraries to do their work. > > What do you think ?? I think if this functionality were added, it should NOT be called Boost.Bio or refer to bioinformatics in any way, because this isn't bioinformatic-specific. But in general adding these string algorithms would be useful - they are used in many fields. - James Jones Administrative Data Mgmt. Webmaster 375 Raritan Center Pkwy, Suite A Data Architect Edison, NJ 08837

1 0

is mpl::for_each depricated?
by Ovanes Markarian 30 Aug '06

30 Aug '06

Hi! Unfortunately I can not find online documentation for the mpl::for_each algorithm. Which I know from C++ Metaprogramming book. Is it depricated? I need at least the header file name. Would be nice if docs would be updated, at least stating, that this function is depricated. With Kind Regards, Ovanes Markarian

2 1

Documentation Suggestion.
by Minkoo Seo 30 Aug '06

30 Aug '06

Hi. I enjoy boost library these days, and really appreciate it. However, sometimes, I found it very hard to even compile a program that uses boost libraries. As an example, serialization library needs libboost_serialization.so. Hence, I have to compile a program like the following: g++ -lboost_serializatino foo.cpp Nevertheless, AFAIK, nowhere in the document, how to compile a program is not presented. Some of the docs even mention the header file. I don't know why this is so common in the documentation of boost, and I think mentioning how to compile will be very useful for users. So, please, please put some 'how to compile' section especially if the library needs -lxxx_library flag. I really thank boost developers' time and efforts. Sincerely, Minkoo Seo -- View this message in context: http://www.nabble.com/Documentation-Suggestion.-tf2189683.html#a6058183 Sent from the Boost - Users forum at Nabble.com.

3 4

[iostreams] Auto flush required
by Roman Perepelitsa 30 Aug '06

30 Aug '06

Hi. I've implemented output stream using boost.iostreams library. It requires call to flush in order to actually write data to sink. But I want to flush data to sink automatically after I write any object with operator<<. struct stream_sink : public boost::iostreams::sink { stream_sink(std::ostream & strm) : strm_(strm) {} std::streamsize write(const char * s, std::streamsize n) { if (strm_.write(s, n)) return n; else throw std::runtime_error("error"); } private: std::ostream & strm_; }; std::stringstream s; boost::iostreams::stream<stream_sink> strm(s); strm << "abc"; strm.flush(); std::cout << "String: '" << s.str() << "'" << std::endl; This test outputs "String: 'abc'" but if you comment strm.flush() then it outputs "String: ''". I want my stream to work like the stringstream does (it does not require call to flush). std::stringstream strm; strm << "abc"; std::string s; strm >> s; std::cout << "String: '" << s << "'" << std::endl; It outputs "String: 'abc'". I need it to create stream which can be used to convert wide char data (wchar_t *) to multibyte (char *). It can be used this way: // wcstombsstream is wostream and istream // "utf-8" is multibyte encoding wcstombsstream strm("utf-8"); strm << L"abc"; std::string s; strm >> s; This code does not work at the moment because it requires to call strm.flush() before reading from stream. Thus I need some auto flush facility :-) Thank you for any help, Roman Perepelitsa.

1 1

Boost Serialization lib:Problem in object tracking and Exporting classes
by mahesh prabhu 30 Aug '06

30 Aug '06

Hi, I'm new to Boost Serialization library . I wanted to use the library for some of my full fledge so I thought I might as well test out the features that the library documentation claimed it had . Many of the basic features did work but some of the advanced features that I wanted most like object tracking and serializing derived objects through a pointer to the base class gave me some problems . I faced the following two problems ( in order of decreasing priority ): (1) Object Serialization : I have 4 pointers to 3 objects of the same class , On deserializing these the pointers are mixed up . If while writing out pointers 1 & 2 pointed to objects 1 and 2 , and pointers 3 and 4 pointed to objects 1 and 3 then on restoration pointers 1 & 2 pointed to objects 1 & 2 but pointers 3 & 4 pointed to objects 2 and 3 instead of 1 and 3 . (2) Writing derived objects through pointer to base class : This feature worked if I did a registration through ar.register_type(static_cast<gps_moreAcc1 *>(NULL)) but did not work if I did BOOST_CLASS_EXPORT( gps_moreAcc1) . I get a "DoneAbort" message i.e a core dump . I don't know too much about templates , so I was not able to debug the problem . I have attached a program which exhibit both problems , but the code that causes the second problem is commented out . This can be reproduced by just commenting out ar.register_type calls and uncommenting BOOST_CLASS_EXPORT macros . I have added more comments in the source file . Please let me know if I am doing something wrong or if these are know problems with the serialization library or if the serialization library cannot be handled by serialization . +Mahesh

1 0

Re: [Boost-users] timed_mutex::scoped_timed_lock
by Andrew Wingorodov 30 Aug '06

30 Aug '06

10x, sorry for my english > I am not sure what this delay function does. this function from ${BOOSTDIR}/libs/thread/test/util.inl > scoped_time_lock takes boost::xtime or bool. 4 here is interpreted as > "true", not as time. oops. okey, the new version of my cool program (how i can see event of timeout?) are ///------- COOLPRG timed_mutex mut; #define TIMEOUT 1 #define AUTHTIME 4 void timeout () { timed_mutex::scoped_timed_lock lk (mut, delay(TIMEOUT)); if ( lk.locked ()) cout << "how i can see event of timeout?" << endl; else cout << "okey!" << endl; } std::auto_ptr<timed_mutex::scoped_timed_lock> lock; void auth () { cout << "auth" << endl; boost::thread::sleep ( delay(AUTHTIME)); cout << "auth end" << endl; lock->unlock(); } int main () { lock = (std::auto_ptr<timed_mutex::scoped_timed_lock>) new timed_mutex::scoped_timed_lock (mut,10); thread_group run; run.create_thread ( &auth); run.create_thread ( &timeout); run.create_thread ( &timeout); run.create_thread ( &timeout); run.join_all(); if ( !lock->locked ()) cout << "unlock!" << endl; } ///EOF And result (how i can see event of timeout?): andr@andr> ./coolprg auth okey! okey! okey! auth end unlock! andr@andr> -- andr.ru

1 0

[BGL] access order in outEdgeList containers ?
by Manuel Yguel 30 Aug '06

30 Aug '06

Hello, Does the order of the add_edge() calls for an adjacency_list graph with vecS or listS OutEdgeList containers defines the access order through out_edge_iterators ? If true, how to define some add_edge_after() method with the same semantic than the insert_after method defined for slist in http://www.sgi.com/tech/stl/Slist.html Thanks a lot. P.S. thanks Doug, for your answer concerning graph properties.

1 0

[regex] - Don't give wrong documentation in boost site itself.
by kiran 30 Aug '06

30 Aug '06

Hi This mail is to say that wrong documentation in a site like boost in no worser than no documentation at all. http://www.boost.org/libs/regex/doc/syntax_perl.html In this the second line saying that "Perl regular expressions are the default behavior in Boost.Regex" and then when using boost::regex i found that "dot" in perl and boost::regex have different meaning by default. I am not speaking aganist the guy who made the documentation. He did a great job by creating the doc which help newbies like me to great extent. The aim of this mail is to get the doc corrected. Could anyone tell what are the differences between perl regex and boost::regex by default . Thanks Kiran.

1 0