November 2009 - Boost-users

[Spirit] poor performance
by HT4N - 20 Nov '09

20 Nov '09

I just shocked to find out that my parser with Spirit-classic is 4x slower than my original ad-hoc parser. I do generate parsetree with pt_parse(). So does my adhoc parser but my adhoc one generates its own custom parse tree. My question are: 1. Does spirit classic with generation of parse tree is known to be slow? 2. Does anyone have ever done perf study and compare Spirit.Classic with Spirit.Qi (assuming to generate custom parse tree in Spirit.Qi)? 3. If the bottleneck is in the Spirit.Classic parse tree generation, has anyone ever done a custom parse tree generation in Spirit.Classic? For now it seems I am hitting a roadblock to use Spirit as performance is my main issue. Your insight is highly appreciated. Thanks HT _________________________________________________________________ Bing brings you maps, menus, and reviews organized in one place. http://www.bing.com/search?q=restaurants&form=MFESRP&publ=WLHMTAG&crea=TEXT…

3 2

Boost.Fusion 1.41 issue ?
by Surya Kiran Gullapalli 20 Nov '09

20 Nov '09

Hello, he following test program fails to compile with Boost-1.41 on msvc8.0 SP1. It was working fine with Boost-1.40. Am I missing something here ? Thanks, Surya #include <iostream> #include <boost/preprocessor.hpp> #include <boost/fusion/container.hpp> using namespace std ; using namespace boost ; #define MAX_ARGS 5 template <int id, BOOST_PP_ENUM_PARAMS(MAX_ARGS, typename T)> void func (const boost::fusion::vector<BOOST_PP_ENUM_PARAMS(MAX_ARGS, T)> p_args) { cout << "Multiple Arguments: id " << id << endl ; } template<int id> void func (void) { cout << "No arguments: id " << id << endl ; func <id>(boost::fusion::make_vector()) ; } int main (void) { func<10> (boost::fusion::make_vector(1, 2)) ; func<6> () ; return 0 ; }

2 3

multi-index container random access index ordered
by Vladimir Voronkov 20 Nov '09

20 Nov '09

Hi, I defined a container like this: typedef boost::multi_index_container< std::pair<int, int>, boost::multi_index::indexed_by< boost::multi_index::random_access<>, boost::multi_index::ordered_unique< boost::multi_index::member<std::pair<int, int>, int, &std::pair<int, int>::first> > > > IdContainer; typedef IdContainer::nth_index<0>::type IdsArr; typedef IdContainer::nth_index<1>::type IdsSet; and then I insert a data into it: a_set.insert(std::make_pair(1, 0)); a_set.insert(std::make_pair(3, 0)); a_set.insert(std::make_pair(2, 0)); a_set.insert(std::make_pair(5, 0)); a_set.insert(std::make_pair(4, 0)); in this case elements will be “pushed back” in terms of random access index, i. e. for (size_t i = 0; i < a_arr.size(); i++) { std::cout << a_arr[i].first << std::endl; } will return 1 3 2 5 4 I would like to have them inserted ordered way, and as long as we have ordered index I suppose it’s possible. How can I do that? I suspect I could do like this: a_arr.insert(a_set.find(1), std::make_pair(1, 0)); but it errors about incompatible types of iterators for what a_set.find returns and what a_arr.insert accepts. Or… may be there are some kind of definition of random access index with respect to another index which is ordered??? I don’t know. Your help is appreciated.

2 1

Re: [Boost-users] [Filesystem] API enhancement request
by Alan M. Carroll 20 Nov '09

20 Nov '09

I would like to have an extension added to Boost.Filesystem, which I have been hand patching along the way. For version 1.41.0, in boost/filesystem/path.hpp, at line 274, add public: typedef boost::BOOST_FILESYSTEM_NAMESPACE::basic_path<String, Traits> path_type; What this does is provides access to the path type for a path iterator from just the iterator. I find it very handy when writing generic path handling code. Otherwise you have to hard code for a particular type of path, or pass the type along even though the iterator depends on that type.

1 0

libboost_filesystem-vc90-xxx-1_39.lib requirement in boost 41?
by Archie14 20 Nov '09

20 Nov '09

My project (windows, vs2008, debug) was linking successfully under boost-39. When I switched to boost 41 version the linker started to complain about missing lib (fatal error LNK1104: cannot open file 'libboost_filesystem-vc90- mt-gd-1_39.lib'). My linker properties does not specify the library explicitly, just the path to boost libraries. The path had been changed to boost 41 (I have $BOOST system var that both compiler and linker properties refer to) and there are corresponding libraries under the same relative path in boost 41 (libboost_filesystem-vc90-mt-gd-1_41.lib). Specifying new library explicitly in the linker properties did not help. I renamed old "39" directory and both compiler and linker are referring to boost via system variable. I double checked that this system variable points to the new "41" directory. Compiling completes sucessfully, but linker fails. How can I correct the problem?

3 2

different behavior for boost::filesystem with VC9 in debug and release
by frederic cervenansky 20 Nov '09

20 Nov '09

Hello, Sorry to ask this question but I investigate and I didn't find any previous message. I try to use boost::filename::is_directory( path) in my code developed on visual studio 2008. In debug mode, I have an error on this function and specificaly at file_status result( detail::status_api( ph.external_file_string(), ec ) ); in operations.hpp. (Detail status crashes, it seems). In release mode, no problem everything is fine, my path is valid and my program is running.... And yes, my path is valid. I have boost 1.40, the compilation option for boost and my program are the same. The linkage is valid (I checked that my compiled libraries are used). So anyone has a clue for this bug? Because I'd like to understand and stay in debug mode to track bugs. Thanks by advance

1 0

Newbie looking for example of try...catch for Date_Time
by Ramon F Herrera 20 Nov '09

20 Nov '09

I have only cursory experience with exemption catching. I already looked in the documentation but didn't find an example. My program keeps on crashing when a bad month or date is used. TIA, -RFH

1 0

posix_time_zone::calc_zone probably buggy
by Sorin Manolache 20 Nov '09

20 Nov '09

Hello, The Posix specification says that the offset in the TZ variable gives the time to add to the local time in order to get UTC. Thus, western hemisphere zones have positive offsets and eastern hemisphere zones have negative offsets. time_zone_base::base_utc_offset gives the offset of the timezone from UTC. Thus, western hemisphere zones have negative base_utc_offset_ and eastern hemisphere zones have positive base_utc_offset_. Posix specifies that the offset specified in TZ must be added to the localtime in order to obtain UTC. base_utc_offset_ must be added to UTC in order to obtain local times. Thus, base_utc_offset_ must be minus the TZ offset. So I think a minus should be added in front of date_time::str_from_delimited_time_duration<time_duration_type,char_type> in the calc_zone method of date_time/local_time/posix_time_zone.hpp. Regards, Sorin

1 0

[function] Using boost::function when rtti is not available
by David Genest 20 Nov '09

20 Nov '09

Hi, I am in a gaming build environment that has traditionally turned of rtti because of space and portability issues. But, I would like to use the very useful boost::function. On MSVC boost::function works, but fails on g++ for the ps3. I have read on the lists that it is possible to implement a rtti-less version, and that this is even wanted in the future. See: http://article.gmane.org/gmane.comp.lib.boost.devel/192378 And http://article.gmane.org/gmane.comp.lib.boost.devel/191945 I know of a lot of people that would like this dependency on rtti removed. My question to the list is: did anyone have a try at an implementation and could share it ? I have toyed with boost::function and sp_typeinfo.hpp and tried to replace occurrences of type_info in function_base.hpp, but I don't have enough understanding to finish the work and my early attempts have failed. Thank you, David.

1 0

boost pointer serialisation broken on IBM AIX
by Avi Bahra 20 Nov '09

20 Nov '09

Serialisation of pointers is ok linux, hpux, but has the following exception: archive_exception::unregistered_class on IBM AIX v10.1 using boost 1.39 Trivial case can be made to work. However as the number of files grows, pointer serialisation start/stops working. There seems to undefined behaviour. There is a _known_ problem with serialisation via base pointer, however this problem is exhibited with normal pointer serialisation. It is extremely difficult to create small enough test case which shows the problem. Sometimes a test case will work and at other times a seemly innocuous change (like adding/removing more hierarchy) will show the problem again. There does not appear to be any rhyme or reason why, Could it be a compiler issue ? template<class T> struct polymorphic { static void save( Archive &ar, T & t ){ ........ const boost::serialization::extended_type_info * true_type = i.get_derived_extended_type_info(t); // note:if this exception is thrown, be sure that derived pointer // is either registered or exported. if(NULL == true_type){ boost::serialization::throw_exception( archive_exception(archive_exception::unregistered_class) <<<<<<<<<<<<<<<<<<<<<<<< ); } Has any one else come across this weird behaviour, in fact I would like to here from anyone who has got pointer serialization working consistently on AIX. Any help greatly appreciated Ta, Avi

1 0