- Boost-users - lists.preview.boost.org

iterator_adaptor problem (mpl?)
by Angus Leeming 01 Jul '04

01 Jul '04

Consider a container such as std::vector<std::pair<int, std::string> > my_container; I want to be able to iterate over the ints of this container so, following the tutorial to the iterator_adaptor docs, came up with template <typename BaseIterator> class first_iter : public boost::iterator_adaptor< first_iter<BaseIterator>, BaseIterator, boost::use_default, boost::use_default, typename BaseIterator::value_type::first_type & > { // snip body }; This works fine so long as BaseIterator::value_type is not const, but falls over when it is because I've defined the 'super_t::reference' type to be mutable. I guess that I need a piece of metafunction magic, but my naive implementation fails typename boost::mpl::apply_if< boost::is_const<typename BaseIterator::value_type> , typename BaseIterator::value_type::first_type const & , typename BaseIterator::value_type::first_type & >::type trial.C:38: error: no type named `type' in `struct boost::mpl::apply_if< boost::is_const<std::pair<int, std::string> >, const int&, int&>' I must admit I'm a bit baffled. There plainly is a 'type' in the apply_if class template. Any pointers? I attach the code as I have it. While I'm at it, is there not something in the MPL that will return the appropriate const-qualified reference for me? None of the header file names in boost/mpl jump out and grab me, but I may well be looking in the wrong place. Regards, Angus

2 4

RE: [Boost-users] [pool] Speed of pool_allocator<>
by Klaus Nowikow 01 Jul '04

01 Jul '04

Peter Dimov wrote: > Klaus Nowikow wrote: > > > > I changed the test program (see below). Now I use object_pool<> > > to create and destroy a number of elements (I use a vector<C*> to > > store them in the meantime). The results are even worse than before: > > > > new/delete: 0.08s > > object_pool: 25.20s > > [...] > > > template <template <class, class> class Pool> > > double Test() > > { > > Pool<C> P(CHUNK_SIZE); > > boost::timer Timer; > > > > std::vector<C*> V; > > for(int i = 0; i < NUM_ELEMENTS; ++i) > > { > > V.push_back(P.construct()); > > } > > for(std::vector<C*>::iterator pos = V.begin(), end = V.end(); pos != > > end; ++pos) { > > P.destroy(*pos); > > } > > > > return Timer.elapsed(); > > } > > Your test is still very unrealistic. You create a large > number of objects in > one batch, then destroy them afterwards. A more realistic > scenario would > have creation and destruction interleaved. > > The most accurate test would be to try Boost.Pool with your actual > application and see how it affects its real-world > performance. It's still > possible for it to be slower, of course, but at least you'd > measure the real > thing. Of course you're right. Obviously I broke the "don't optimize what does not have to be optimized" rule ;-). When creating and deleting 100000 Objects with new/delete takes only 0.1 seconds I won't have to look for something more efficient. I will do what you suggest: write my code and see if it needs optimization. Regards, Klaus

1 0

[mpl] Alexandrescu's hierarchy generators
by Klaus Nowikow 01 Jul '04

01 Jul '04

Greetings! Does boost.mpl provide type hierarchy generators like GenScatterHierarchy and GenLinearHierarchy as described in "Modern C++ Design"? I must confess I do not understand the mpl completely yet ;-) Best regards, Klaus

2 1

RE: [Boost-users] [pool] Speed of pool_allocator<>
by Klaus Nowikow 01 Jul '04

01 Jul '04

Peter Dimov wrote: > Klaus Nowikow wrote: > > I am in the process of writing a program which has to > > handle large numbers (~200000 - 400000) objects of the same class. > > These objects will be created and deleted frequently, so I thought > > some kind of specialized memory management would be handy. > > > > I wrote a test program (see below) to check what boost.pool > would gain > > me. The results make me think that I have done something wrong: > > > > [...] > > > > i.e., std::allocator was more than 3 times faster than > > boost::pool_allocator (Win2k, 256MB). Any ideas what i have done > > wrong? > > [...] > > > template <class Allocator> > > double Test() > > { > > std::vector<Allocator::value_type, Allocator> Vector; > > boost::timer Timer; > > > > for(int i = 0; i < NUM_ELEMENTS; ++i) > > { > > Vector.push_back(i); > > } > > > > return Timer.elapsed(); > > } > > If I read the code correctly, you don't create or delete > objects. You create > and delete (large) arrays of objects. Hmm. I'll have to think about this. Maybe the test was a bit over-simplified. The real program would have a vector<C*> where C is my class. I'll change the example (using object_pool) and post the results.

1 0

[pool] Speed of pool_allocator<>
by Klaus Nowikow 01 Jul '04

01 Jul '04

Hi! I am in the process of writing a program which has to handle large numbers (~200000 - 400000) objects of the same class. These objects will be created and deleted frequently, so I thought some kind of specialized memory management would be handy. I wrote a test program (see below) to check what boost.pool would gain me. The results make me think that I have done something wrong: debug mode: std::allocator 2.17s boost::pool_allocator 7.18s release mode: std::allocator 0.11s boost::pool_allocator 0.36s i.e., std::allocator was more than 3 times faster than boost::pool_allocator (Win2k, 256MB). Any ideas what i have done wrong? See the test program below. // MemoryPoolTest.cpp // Boost includes ------------------------------------------------------------- #include <boost/timer.hpp> #include <boost/pool/pool_alloc.hpp> // Std includes --------------------------------------------------------------- #include <vector> #include <iostream> enum { NUM_ELEMENTS = 3000000 }; template <class Allocator> double Test() { std::vector<Allocator::value_type, Allocator> Vector; boost::timer Timer; for(int i = 0; i < NUM_ELEMENTS; ++i) { Vector.push_back(i); } return Timer.elapsed(); } int main(int argc, char* argv[]) { std::cout << "Starting memory pool test" << std::endl; double Time; std::cout << "Allocating with boost::pool_allocator<>: "; Time = Test<boost::pool_allocator<int> >(); std::cout << Time << "s" << std::endl; std::cout << "Allocating with std::allocator<> : "; Time = Test<std::allocator<int> >(); std::cout << Time << "s" << std::endl; return 0; }

2 1

Newbie question... (sorry)
by Chris 30 Jun '04

30 Jun '04

Hi all, sorry for the newbie question... I have been trying to use the boost libraries with no luck I will briefly as possible list the steps I took, so that you can let me know where I went wrong. (please bear with me) 1) I downloaded the latest version and decompressed it so that i have a "C:\boost_1_31_0 directory". 2) I built Boost.Jam by starting the Visual Studio.Net command prompt, cd'ed to the "C:\boost_1_31_0\tools\build\jam_src" directory and invoked the build.bat .... The build was succesfull. 3) Copied the bjam.exe from the "C:\boost_1_31_0\tools\build\jam_src\bin.ntx86" directory to the Vc7/bin directory. 4)Added the VC71_ROOT environment variable to windows with the "C:\Program Files\Microsoft Visual Studio .NET 2003\Vc7" path as the value. 5) Made sure that the PATH environment variable had the Vc7\bin path in it. 6) reset. 7) Started the Visual Studio.Net command prompt, cd'ed to the "C:\boost_1_31_0" directory and invoked bjam "-sTOOLS=vc7.1" install ... the whole process started.. building was succesfull (I believe) and I had a new directory c:\Boost with the \include and \lib sub dirs as mentioned in the "Getting Started" page. 8) Started visual studio.NET, went into the menu "Tools->Options->Projects->VC++ Directories" where I set "C:\boost_1_31_0" in the list of INCLUDE directories. 9) Copied a simple example from the boost website to test the libraries, compiled and VC doesn't find one of the header files.. 10) After many, many attempts I lost it and added the "C:\boost_1_31_0\lib" and "C:\boost_1_31_0\bin" to the other list of directories in the LIBRARY and EXECUTABLE list of directories in VC. After loosing it some more (after more attempts) I added even "C:\boost" in the same places as I had done in the previous steps (just in case). Now I get this fatal error "fatal error LNK1104: cannot open file 'libboost_filesystem-vc71-sgd-1_31.lib' " It doesn't seem to find the .lib but I have checked in the boost directories and that .lib is there!!! Any suggestions.... please... I also feel really bad, 'cause it seems I am the only idiot who hasn't been able to do this :(((( If someone can help me I will gladly return the favor (if I can) Thank you very very much in advance, Chris ------------------------ "Programming is Understanding" - Kristen Nygaard (I just wish I could understand more)

4 5

[admin] Wiki Spam -- IP's are being banned
by Jeff Garland 28 Jun '04

28 Jun '04

All - I'm cross-posting this to boost-users and the developer list since it pertains to both. Over the last couple months there has been a rise in 'spam attacks' on the Wiki. Basically people are inserting links to commercial sites on wiki pages. These sites are totally unrelated to boost, c++, etc. I thank the many boosters that have spent their precious time removing the defacements. But the time has come to crack down. The following IP ranges have been banned from the Wiki. 61.55.32.* 61.55.33.* 69.63.103.* 210.82.76.* 210.82.106.* 219.233.195.* 218.5.65.* I will be monitoring and banning additional IP addresses as needed. If I have accidentally banned someone incorrectly, please let me know offlist: jeff at crystalclearsoftware.com Thanks, Jeff

1 0

singleton_pool
by . . 28 Jun '04

28 Jun '04

hi all, i want to use singleton_pool, but on vc6.0 i get the following error (i have vc6.0 sp5): boost\pool\simple_segregated_storage.hpp(51) : error C2248: 'object_creator' : cannot access private struct declared in class 'boost::details::pool::singleton_default<struct boost::singleton_pool<class CMate rialProxy,4,struct boost::default_user_allocator_new_delete,class boost::details::pool::win32_mutex,32>::pool_type>' e:\documents\develop\include\boost-1.30.2\boost\pool\detail\singleton.hpp(63) : see declaration of 'object_creator' e:\documents\develop\include\boost-1.30.2\boost\pool\simple_segregated_storage.hpp(51) : while compiling class-template static data member 'struct boost::details::pool::singleton_default<struct boost::singleton_pool<class CMaterialProxy,4,st ruct boost::default_user_allocator_new_delete,class boost::details::pool::win32_mutex,32>::pool_type>::object_creator boost::details::pool::singleton_default<boost::singleton_pool<CMaterialProxy,4,boost::default_user_allocator_new_delete,boost::deta ils::pool::win32_mutex,32>::pool_type>::create_object' thanks in advance Fabio --------------------------------- Do you Yahoo!? New and Improved Yahoo! Mail - Send 10MB messages!

1 0

ternary search trie
by Sybrand Badenhorst 28 Jun '04

28 Jun '04

Hi, Does the BOOST library (somewhere) contain a ternary search trie container type (as described by Robert Sedgewick in his "Algorithms - 3ed edition" or here: http://www.cs.princeton.edu/~rs/strings/) that can by used as a standalone container (such as for example STL's map) ? I don't find something like this listed on the Boost library page (http://www.boost.org/libs/libraries.htm) Any help will be greatly appreciated, Thanks Sybrand

3 3

Spirit problem
by Michael Orlov 28 Jun '04

28 Jun '04

Hi, I am new to Spirit, and I am writing a reader for s-expressions, and so far has encountered the following problems, which seem to me to be contrarry to the documentation (Spirit 1.8.0 in Boost 1.31.0). Consider the following (small) C++ program: --------------------------------------------- #include <iostream> #include <iterator> #include <string> #include <boost/spirit/core.hpp> #include <boost/spirit/iterator/multi_pass.hpp> using namespace std; using namespace boost::spirit; struct Sexp : public boost::spirit::grammar<Sexp> { template <typename ScannerT> struct definition { typedef rule<ScannerT> rule; definition(const Sexp &self) { first = ( sexp = '(' >> *sexp >> ')' ); } subrule<0> sexp; rule first; const rule& start() const { return first; } }; }; int main() { typedef std::istream_iterator<char> iterator_t; typedef boost::spirit::multi_pass< iterator_t, boost::spirit::multi_pass_policies::input_iterator, boost::spirit::multi_pass_policies::first_owner > iterator_wrapper_t; Sexp sexper; iterator_t under_first(cin); iterator_t under_last; iterator_wrapper_t first(under_first); iterator_wrapper_t last(under_last); while (true) { parse_info<iterator_wrapper_t> info = parse(first, last, sexper); first.clear_queue(); if (info.hit) std::cout << "Hit: " << info.length << std::endl; else break; }; } --------------------------------------------- The purpose is to read s-expression from input, delivering them to the user the moment they are ready. Since whitespace skipper tries to consume as much space as it can, it is not suitable here, so I don't want to use it. Here comes the first problem: even when I don't pass space_p to parse(), the skipper is still present: --- ( ) ^D Hit: 2 --- --- (())( Hit: 4 ) ^D Hit: 2 --- Another problem is when I try to use istreambuf_iterator instead of istream_iterator (this is done in Spirit examples, too). When substituting istreambuf_iterator in the first typedef in main(), the program no longer compiles - neither with Intel Compiler 8.0.055, nor with GCC 3.3.2. Intel complains about /usr/local/include/boost/spirit/iterator/multi_pass.hpp(537): error: initial value of reference to non-const must be an lvalue return **input; and GCC's messages are too long to parse. :) Any help? Thanks, Michael

2 1