- Boost-users - lists.boost.org

[signals] dropping ThreadingModel template parameter
by Frank Mori Hess 01 Mar '07

01 Mar '07

It seems that now that we drop locks before running slots, we can get by with using boost/detail/lightweight_mutex.hpp for our mutex. Since it is header-only, and apparently compiles to a null mutex if the compiler is in single-threaded mode, I don't see much need for a ThreadingModel template parameter on the signal class any more. I'm dropping it from thread_safe_signals. -- Frank

5 6

[signals] scoped_blocker for connections?
by Frank Mori Hess 01 Mar '07

01 Mar '07

I wonder if it might be convenient to provide a scoped_blocker object that lets a connection be blocked/unblocked in an exception-safe way, similar to how a scoped_lock locks a mutex, or a scoped_ptr owns allocated memory. A fancier scheme would be to provide a shared_blocker, where when you request a shared_block from the connection it does reference counting so the connection remains blocked until the last shared_blocker (requested by any connection handle object referring to the same connection body) goes out of scope. This could be implemented with a shared_ptr that has a custom deleter, and a weak_ptr kept in the connection body to allow different connection handles to get copies of the same shared_ptr. -- Frank

1 0

boost::variant question
by Digvijoy Chatterjee 01 Mar '07

01 Mar '07

Hi, i have a boost::variant holding some C++ types which have to be inserted into the database : boost::variant<int, std::string, double > dbColumnValueVar; before i insert the value_type into the database I need to call a processing function which is overloaded for every datatype used in the variant ie. each of these functions are defined process ( int ); process ( std::string ); process ( double); template <class T> process(T); ..... what i do not understand is how do i use apply_visitor or some other tactic to automatically dispatch the call to the right process function as in : boost::variant<int, std::string, double > dbColumnVar; dbColumnVar = "hello world"; process ( dbColumnVar ); // should call process std::string ....but is always calling the generic template version ?? dbColumnVar = 1.3434; process ( dbColumnVar ); //again calls generic template not specialization cout << dbColumnVar is somehow able to get the correct overloaded << operator for the hidden datatype irrespective of datatype ?? Any help will be appreciated Thanks Digz

2 1

iostreams::mapped_file optimal_buffer_size error
by Benjamin A. Collins 01 Mar '07

01 Mar '07

I'm trying to use iostreams::mapped_file as a source for a iostreams::filtered_istream, and it's not quite working out. I create my chain like this: boost::iostreams::filtered_istream in; in.push(boost::iostreams::zlib_decompressor()); in.push(boost::iostreams::mapped_file("some file name")); // error here I haven't dug too deeply yet because I'm hoping this is not new or obscure. gcc 3.3.5 on OpenBSD 3.9 reports the following error: [03.01.07 12:13:28 bc@qui-gon libgit++]$ scons tools Skipping checks... g++ -Wall -Werror -fno-builtin -g -O0 -fno-default-inlines -DVERSION="0.0.4" -I/usr/local/include/boost-1_33_1 -Isrc -c -o src/myprogram.o src/myprogram.cpp /usr/local/include/boost-1_33_1/boost/iostreams/optimal_buffer_size.hpp: In function `streamsize boost::iostreams::optimal_buffer_size(const T&) [with T = boost::iostreams::mapped_file]': /usr/local/include/boost-1_33_1/boost/iostreams/chain.hpp:236: instantiated from `void boost::iostreams::detail::chain_base<Self, Ch, Tr, Alloc, Mode>::push_impl(const T&, int, int) [with T = boost::iostreams::mapped_file, Self = boost::iostreams::chain<boost::iostreams::input, char, std::char_traits<char>, std::allocator<char> >, Ch = char, Tr = std::char_traits<char>, Alloc = std::allocator<char>, Mode = boost::iostreams::input]' /usr/local/include/boost-1_33_1/boost/iostreams/chain.hpp:179: instantiated from `void boost::iostreams::detail::chain_base<Self, Ch, Tr, Alloc, Mode>::push(const T&, int, int, typename boost::disable_if<boost::iostreams::is_std_io<T>, void>::type*) [with T = boost::iostreams::mapped_file, Self = boost::iostreams::chain<boost::iostreams::input, char, std::char_traits<char>, std::allocator<char> >, Ch = char, Tr = std::char_traits<char>, Alloc = std::allocator<char>, Mode = boost::iostreams::input]' /usr/local/include/boost-1_33_1/boost/iostreams/chain.hpp:484: instantiated from `void boost::iostreams::detail::chain_client<Chain>::push_impl(const T&, int, int) [with T = boost::iostreams::mapped_file, Chain = boost::iostreams::chain<boost::iostreams::input, char, std::char_traits<char>, std::allocator<char> >]' /usr/local/include/boost-1_33_1/boost/iostreams/chain.hpp:472: instantiated from `void boost::iostreams::detail::chain_client<Chain>::push(const T&, int, int, typename boost::disable_if<boost::iostreams::is_std_io<T>, void>::type*) [with T = boost::iostreams::mapped_file, Chain = boost::iostreams::chain<boost::iostreams::input, char, std::char_traits<char>, std::allocator<char> >]' src/myprogram.cpp:27: instantiated from here /usr/local/include/boost-1_33_1/boost/iostreams/optimal_buffer_size.hpp:38: error: ' struct boost::iostreams::detail::optimal_buffer_size_impl<boost::iostreams::mapped_file> ' has no member named 'optimal_buffer_size' scons: *** [src/myprogram.o] Error 1 -- Benjamin Collins ben.collins(a)acm.org

1 0

[the Boost Graph library] no matching function : vertices(const Graph& g)
by Gerd Menche 01 Mar '07

01 Mar '07

Dear Boost users, I have come across this strange problem: After designing an interface to a library dealing with molecules (OpenBabel). I get the following compiler error: /usr/include/boost/graph/breadth_first_search.hpp: In function void boost::breadth_first_search(const VertexListGraph&, typename boost::graph_traits<G>::vertex_descriptor, Buffer&, BFSVisitor, ColorMap) [with VertexListGraph = OpenBabel::OBMol, Buffer = boost::queue<OpenBabel::OBAtom*, std::deque<OpenBabel::OBAtom*, std::allocator<OpenBabel::OBAtom*> > >, BFSVisitor = boost::VerboseBFSVisitor, ColorMap = std::vector<boost::default_color_type, std::allocator<boost::default_color_type> >]: bfs_example.cpp:73: instantiated from here /usr/include/boost/graph/breadth_first_search.hpp:96: error: no matching function for call to vertices(const OpenBabel::OBMol&) Easy enough you say -- go ahead and write that function. But it is allready in place. After modiying breath_first_search.hpp on line 96 from ... vertices(g) ... to ... gvertices(g) ... and changing the name of my vertices() function to gvertices(). It works. Is there a different vertices() function defined anywhere, that gets in the way? ____________________________________________________________________________________ Looking for earth-friendly autos? Browse Top Cars by "Green Rating" at Yahoo! Autos' Green Center. http://autos.yahoo.com/green_center/

2 1

boost 1.33.1 + qt 3.3.7 free on linux
by misiu 01 Mar '07

01 Mar '07

Hello, I'm developing standard c++ application with boost libs: any, thread, file_system, lambda, regex and everything is fine. Then I'm adding Qt for GUI and on MS Windows still fine but on Linux I got very strange problem: compilation and linking without problems (even no warnings) but just after start program hangs (even not showing GUI window) but not crushes (no core dumped). Without boost dependent parts Qt program works fine and boost parts without Qt work fine as well. What do you thing could make such a problem? What kind of checks can I do? If it is important I compiled Qt and boost by myself. I have to say I'm not so experienced with developing application on Linux, everything I'm preparing on Windows and from time to time I'm checking whether it works on Linux. Regards, Bartek

3 3

Performance of object_pool
by tricycle 01 Mar '07

01 Mar '07

The most important reason to use a memory pool is performance. BUT object_pool doesn't! When I use object_pool to construct a lot of objects, the destroy method become very slow. And I found object_pool use ordered_free to free the memory, and it amortized O(N). It is a very serious problem, I hope it can be fixed in the next release. thanks.

1 0

Re: [Boost-users] Problem using regex library in Windows/cygwin
by Diego Molla 01 Mar '07

01 Mar '07

Thanks John, spot on. I've used a new iterator variable for the result of regex_search and it works now. Now the code is: if (regex_search(what[1].str(),what2,attrMarkup)) { > There's at least one issue with your code, in: > if (regex_search(what[1].str(),what,attrMarkup)) { > the result of what[1].str() is a temporary string object: by the time the > call returns the object has gone out of scope, and the iterators stored > in > "what" are invalid. > HTH, John. Diego -- 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

1 0

[filesystem] ..\foo\bar - invalid path syntax?
by n.torrey.pines 01 Mar '07

01 Mar '07

Hi As you know, on Windows, it is conventional to use backslashes in paths. I am not out to teach Windows users a different convention, let alone "generic syntax". So I'd like to construct paths from strings like ..\foo\bar. The constructors like path(const std::string&) treat this as an invalid syntax however (changing '\' to '/' works fine). Is there a different procedure I should invoke that respects the native platform syntax? I've looked in the docs, but I didn't find it. There are only procedures for native printing. I could of course manually replace '\' with '/' in the string, but such solution probably wouldn't be very robust. Thanks

1 0

[type_traits] is_function
by Cromwell Enage 28 Feb '07

28 Feb '07

Assuming a compiler supports is_function, will this metafunction check for the presence of member functions in a class? For example: struct Foo { void bar(); }; struct Baz { }; Given the right template argument, will is_function detect the bar() function in Foo but return false_ for Baz? If so, what is the right template argument? If not, are there any known workarounds? Cromwell D. Enage ____________________________________________________________________________________ Finding fabulous fares is fun. Let Yahoo! FareChase search your favorite travel sites to find flight and hotel bargains. http://farechase.yahoo.com/promo-generic-14795097

2 2