December 2007 - Boost-users - lists.preview.boost.org

[statechart] history transitions make state-local-storage "un-cool"
by Mick Hollins 26 Jun '08

26 Jun '08

Apologies if this has been discussed before, but I couldn't find any mention of it: I am currently using the history facility of boost::fsm. I've discovered that when I re-enter a state, X, via a history transition I don't actually re-enter the same instance of X that I was in before, but instead enter a newly constructed instance of X. I just checked the UML spec regarding history transitions and it says that "Any necessary entry actions are performed", so given that boost::fsm uses constructors to represent entry actions I guess a new instance of X has to be constructed to ensure UML compliance. My actual concern is the effect this has on the "coolness" of state-local-storage (see http://boost-sandbox.sourceforge.net/libs/statechart/doc/faq.html#Whats_so_…) It seems to me that state-local-storage, which I agree is generally cool, is particularly un-cool (and bug-inducing) in the presence of history transitions. I'm not sure what a good alternative is, but just thought I'd raise the discussion. cheers, mick PS: does the history facility display the same semantics in boost::statechart? I plan to move to statechart real soon now ...

4 19

[foreach][range] How would BOOST_REVERSE_FOREACH be implemented?
by Pete Bartlett 05 Mar '08

05 Mar '08

Boost.Foreach is such a great piece of kit that once you start using it you just can't stop. Indeed I like working directly with values rather than iterators so much that is feels normal for me to write something like BOOST_FOREACH( value_type x , make_iterator_range( rbegin( seq ) , rend( seq ) ) whenever I wish to reverse iterate over a sequence. My co-workers are not big fans of typing though, so they want a REVERSE_FOREACH. At first naïve glance it looks easy to implement in terms of the original FOREACH: template<typename C> inline iterator_range</* something */> make_reverse_iterator_range(C& c) { return make_iterator_range( rbegin(c) , rend(c) ); } #define REVERSE_FOREACH( x , container ) BOOST_FOREACH( x , make_reverse_iterator_range( container ) ) This works fine when container is an lvalue. But then co-workers, used to the wonders of the original FOREACH, inevitably start to use the new macro with rvalues and get a crash due to dangling references (- the iterator_range's members point to departed quantities.....) and if you could somehow work around that, there is the issue hidden inside the "/* something */" that you have to cope with fact that MSVC has no proper const rvalue detection. Finally the #define MACRO(x,y) ANOTHER_MACRO( f(x) , g(y) ) style is not recommended because they are brittle in the face of x and/or y themselves being macros. ... but all these issues are exactly the ones that BOOST_FOREACH itself solves! Thus surely it is technically possible to write a REVERSE_FOREACH that works with rvalues as well as lvalues, e.g. by essentially copying and pasting the FOREACH implementation and replacing "begin" and "end" by "rbegin" and "rend" in the vital places. However one feels there may be a better way... has anyone had a go at creating a "fully-functional" REVERSE_FOREACH and could offer me any advice? If indeed the task is fairly intricate and requires re-use of foreach internals, would the library author consider including something in the library itself? Many thanks for reading, Pete Bartlett

4 6

boost python: linking with pythonxx_d.lib
by Christian Hoffmann 01 Feb '08

01 Feb '08

Hi, I tried to compile boost python so that it links against the debug version. I therefore run bjam toolset=msvc-9.0 threading=multi python-debugging=on link=shared variant=debug libs\python\build Running it with -n shows, that it sets -DBOOST_DEBUG_PYTHON, but it still links with python25.lib. It only picks "right" python25_d.lib, if I add something to my user-config.jam, like the following: using python : 2.5 : C:\\src\\Python-2.5\\PCBuild\\python ; using python : 2.5 : C:\\src\\Python-2.5\\PCBuild\\python_d : # includes : # libs : <python-debugging>on ; What is the difference between the two methods (as a bjam argument and in the user-config.jam), and is there a way to combine both on the command line? Thanks and regards, Chris

2 1

[serialization] hash_map declaration problem
by Jan Boehme 26 Jan '08

26 Jan '08

Hi, I encountered problems while trying to serialize hash_maps which use pool allocators ( gcc: no member 'serialize' ). I had a look into "boost/serialization/hash_map.hpp" and realized a problem in the declaration of hash_map in here: BOOST_STD_EXTENSION_NAMESPACE::hash_map<Key, HashFcn, EqualKey, Allocator> should be BOOST_STD_EXTENSION_NAMESPACE::hash_map<Key, Value, HashFcn, EqualKey, Allocator> It seems that nobody used the hash_map with more than 4 template parameters together with boost::serialization before. It would be nice, if the following changes for "boost/serialization/hash_map.hpp" would be adopted into the repos. Cheers, Jan. --- /boost-trunk/boost/serialization/hash_map.hpp 2007-08-01 09:03: 26.000000000 +0200 +++ /boost/serialization/hash_map.hpp 2007-12-17 11:36:27.000000000 +0100 @@ -33,3 +33,4 @@ class Archive, - class Key, + class Key, + class Value, class HashFcn, @@ -41,3 +42,3 @@ const BOOST_STD_EXTENSION_NAMESPACE::hash_map< - Key, HashFcn, EqualKey, Allocator + Key, Value, HashFcn, EqualKey, Allocator > &t, @@ -48,3 +49,3 @@ BOOST_STD_EXTENSION_NAMESPACE::hash_map< - Key, HashFcn, EqualKey, Allocator + Key, Value, HashFcn, EqualKey, Allocator > @@ -55,3 +56,4 @@ class Archive, - class Key, + class Key, + class Value, class HashFcn, @@ -63,3 +65,3 @@ BOOST_STD_EXTENSION_NAMESPACE::hash_map< - Key, HashFcn, EqualKey, Allocator + Key, Value, HashFcn, EqualKey, Allocator > &t, @@ -70,3 +72,3 @@ BOOST_STD_EXTENSION_NAMESPACE::hash_map< - Key, HashFcn, EqualKey, Allocator + Key, Value, HashFcn, EqualKey, Allocator >, @@ -75,3 +77,3 @@ BOOST_STD_EXTENSION_NAMESPACE::hash_map< - Key, HashFcn, EqualKey, Allocator + Key, Value, HashFcn, EqualKey, Allocator > @@ -85,3 +87,4 @@ class Archive, - class Key, + class Key, + class Value, class HashFcn, @@ -93,3 +96,3 @@ BOOST_STD_EXTENSION_NAMESPACE::hash_map< - Key, HashFcn, EqualKey, Allocator + Key, Value, HashFcn, EqualKey, Allocator > &t,

3 4

TestUnit - builing hierarchical tests
by Tomasz Kalkosiński 07 Jan '08

07 Jan '08

Hello group. I want to use Boost.UnitTest library for my code testing. I went thru examples and I've found it useful and easy. It's easy on simple one-method one-file examples. My problem begun when I wanted to build few hierarchical test suites splitted into directories and files. I failed. There is no example on boost page - how to write them splitted in a few files. Only cases are pointers to functions in one file. I had problems including unit test header - I had multiple definitions of boost classes. When i removed header - there were no macros etc. Can anyone share a good practice - how to wrap example1 and example2 to one mainsuite? And use 3 files for it? Examples from tutorial: http://www.pdc.kth.se/training/Talks/C++/boost/libs/test/doc/components/uni… On a side note I find myself disappointed with Boost.UnitTest documentation. It isn't easy to follow (like I did for example with Boost.Regex) . Links to examples are broken on main site : http://www.boost.org/libs/test/doc/components/utf/index.html. But there's a mirror on http://www.pdc.kth.se/training/Talks/C++/boost/libs/test/doc/components/uni…. That's nice, but their latest revision is more than 4 years old! But when you google enough you find another one - author's own - on http://www.patmedia.net/~rogeeff/html/ revised 2007 (at least copyright is 2007). Is it the way Boost library documentation is meant to be? On a second side note I had no "main ()" function problem on debian. There were changes on 1.34.0 so it won't work anymore as documentation says. So I had to dig thru bug reports, mailing lists to find out author's personal note on these changes. As a library user I don't want to waste time googling, I want to use documentation and get it working. I belive Boost.UnitTest is a good work so don't waste it on a poor support! Greetings, Tomasz Kalkosiński

3 7

statechart null in_state_reaction
by Sean Kelly 06 Jan '08

06 Jan '08

Hello, I would like to implement a null-op in_state_reaction - The in_state_reaction template requires a action argument whereas the transition to self doesn't - Which is the preferred method ? Thanks Sean

2 2

[Serialization] Can I load two object from one input file?
by sjdf 02 Jan '08

02 Jan '08

Hi all, Code like this: // Save std::ofstream ofs( "c.out" ); A testA1; B testB1; testA1.Save( ofs ); testB1.Save( ofs ); ofs.close(); // Load std::ifstream ifs( "c.out" ); A testA11; B testB11; testA11.Load( ifs ); testB11.Load( ifs ); The testA1 and testB1 can save successful, and the testA11 can load successful, but when loading testB11, it throw a exception "invalid_signature". I have trace into the code, I found when the testA11 loading, the "file_signature" is "serialization::archive", but when testB11 loading, the "file_signature" is empty. Why it like this? What should I do? May be there is the same question before I ask, but I don't know what keyword could be search, sorry. Thanks!

4 6

VS2005 with boost and /CLR
by Jason McGuidwin 02 Jan '08

02 Jan '08

I want to use the boost libraries with VS2005 and CLR option. how did you do this? how do you do this? I can get it to work without clr but I need to use managed and unmanaged code.

4 5

boost::thread passing reference of the data object to thread func does not work
by Xiaofan Li 31 Dec '07

31 Dec '07

Hi, Sorry to disturb the list but I've been stuck with this problem for a while. The idea is to pass the reference of some data object to the thread function, in which the data object (by which the ref is passed in) will be modified. However I have found that this would not work at run-time, as the program compiled, ran but the modifying code silently failed to execute. I've written a small sample, which is self-contained and compilable: // // File: thread_test.cc // Author: xli // // Created on December 26, 2007, 12:26 AM // #include <boost/thread/thread.hpp> #include <boost/bind.hpp> #include <iostream> #include <vector> using namespace std; int s_count = 0; boost::mutex mutex; struct some_data_t { some_data_t() : m_int(0) {} int m_int; vector<int> m_vec; }; class threaded { public: void increment_count( some_data_t& data ) { boost::mutex::scoped_lock lock(mutex); std::cout << "count = " << ++s_count << std::endl; data.m_int += s_count; data.m_vec.push_back(s_count); } }; some_data_t a; int main(int argc, char* argv[]) { threaded _thrd; boost::thread_group threads; for (int i = 0; i < 10; ++i) threads.create_thread( boost::bind( boost::mem_fn( &threaded::increment_count ), &_thrd, a ) ); threads.join_all(); cout <<"All done. " <<a.m_int <<" " <<a.m_vec.size() <<endl; } ----------------- The output (g++4.3/linux) is count = 1 count = 2 ... count = 10 All done. 0 0 However I am expecting the last line as "All done. 55 10" as indicated in increment_count(). I have no idea why the "some_data_t a" did not undergo any change even increment_count() had clearly been called 10 times. I also wish to note that the way I make a member func version of increment_count() is purely for testing and changing it back to a global/static function will not alter the un-designated behaviour I've got. Any opinions on this are more than welcome. Regards, Xiaofan -- Xiaofan Li

2 2

boost::shared_ptr question
by Michael Bradley Jr 31 Dec '07

31 Dec '07

Hi, i do have the following situation 3 independent classes A, B and C with the following requirements: 1. class A { private: A(int); ~A(); class deleter_t{ public: void operator()(A* p) { delete p; } }; friend class deleter_t; public: shared_ptr<A> create(int i) { return boost::shared_ptr<A>(new A(i),deleter_t()); } }; class B { boost::shared<A> obj_; }; class C { boost::shared_ptr<B> b_; boost::shared_ptr<A> a_; }; 2. Class C instantiates A and gets a boost::shared_ptr<A> through A::create Class B just need a boost::shared_ptr<A>. 3. C first release B and then A Now my question how can i achieve that goal, when A MUST obey to the singleton pattern? Thanks in advance for any help Cheers Michael

1 0