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

asio: transporting exceptions across threads
by Zachary Turner 21 Jul '09

21 Jul '09

Suppose I have some io_service object in which I call run() from thread A. Thread A is the main controller of the i/o operations and does things like issuing reads and, upon completion, issuing writes for those reads. The writing service might wish to spawn off some child services that runs in thread B. So when thread A says writer.async_write_data() all it really does is call child_writer_service.post() with the data. Supposing an exception occurs in thread B, what is the best way to have this terminate the entire asio pipeline and propagate the exception all the way up to outside of thread A's call to io_service::run()? I've looked at the sample code on the boost::exception tutorial but I'm unable to apply this to the situation in asio, either due to differences in the threading model or due to my own misunderstandings. Thanks

2 3

where is the pdf documentation?
by Ozgur Ozturk 21 Jul '09

21 Jul '09

Hi, I see mention of pdf documentation, but couldn't find it online. Can someone give the link to it? Thanks, Best regards, Ozgur (Oscar) Ozturk www.DrOzturk.com Phone: +1 (908) DROZGUR i.e, +1 (908) 376-9487

2 1

usage of boost::asio::async_write and boost::asio::buffer
by Ozgur Ozturk 21 Jul '09

21 Jul '09

Hi, I am modifying example echo server for my server. I started to get the attached error, and I suspect I am using buffer incorrectly since error started after a change there. I just changed a line from boost::asio::async_write(socket_, boost::asio::buffer("1^", 3, boost::bind(&session::handle_write, this, boost::asio::placeholders::error)); to boost::asio::async_write(socket_, boost::asio::buffer(replyString, replyString.size()), boost::bind(&session::handle_write, this, boost::asio::placeholders::error)); I generate the replyString to have results from my database query separated with ^ signs (which is used to tokenize the replyString on the client side...) string replyString="1^"; while(res->next()) { replyString+=res->getString("nickname")+"^"; } I checked that replyString.size() is really one more than the final filled position index in replyString. Any ideas, suggestions? TIA, Best regards, Ozgur (Oscar) Ozturk www.DrOzturk.com Phone: +1 (908) DROZGUR i.e, +1 (908) 376-9487

2 3

Docs missing info about which headers to include?
by Jason Cipriani 21 Jul '09

21 Jul '09

I'm new to much of boost. It is less than ideal to have to search through boost headers to find which ones to include for a class (e.g. boost::scoped_lock). Where can I find information about which headers the classes are in? It does not seem to be in the documentation, e.g.: http://www.boost.org/doc/libs/1_37_0/doc/html/boost/interprocess/scoped_loc… This is strange to me, since this type of info is pretty fundamental. In that particular case, after trying boost/thread/scoped_lock.hpp and boost/interprocess/scoped_lock.hpp, I resorted to searching through files and eventually found it in boost/interprocess/sync/scoped_lock.hpp (among other locations, but that one seems to be the most reasonable). Is there some place in the boost docs where I can find what headers to include to use the classes? Thanks, Jason

2 2

Re: [Boost-users] Spirit 1.39 compilation times (avi)
by Avi Bahra 21 Jul '09

21 Jul '09

Can someone explain why the compilation times are faster under Spirit2.1 I have noticed that even boost 1.39 references spirit qi, can I assume that faster compile times are due to moving from spirit classic to spirit qi ? More importantly can we intermix classic and qi grammars, if not are there any migration notes, to help users -- Best regards, Ta, Avi

3 2

Off Topic: Boost Book
by Etienne Philip Pretorius 21 Jul '09

21 Jul '09

Hello List, Any thoughts about this book: http://www.amazon.com/Beyond-Standard-Library-Introduction-Boost/dp/0321133… Also, any other recommendations (for boost) will be appreciated. Thank you, Etienne

6 6

[log] More than one module but static linking?
by Boris Schaeling 21 Jul '09

21 Jul '09

Andrey, the documentation of your very useful logging library says: "If your application consists of more than one module (e.g. an exe and one or several dll's) the library must be built as a shared object." As it turns out I want to use your library in such an application. However there is only one module writing to the log file. Do I still need to use the shared library? Or can I link statically if for example I don't use certain parts of the library? As far as I see there is no more information in the documentation why linking dynamically is forbidden? Boris

1 0

boost::mpl::inherit from a typelist?
by Zachary Turner 21 Jul '09

21 Jul '09

When I use typedef boost::mpl::inherit<boost::mpl::list<udt1, udt2, udt3> >::type derived_class; (at least in MSVC) then it inherits directly from boost::mpl::list, instead of inheriting from the individual members of the list. Is there a way to make it inherit from each type in the list instead? inherit_linearly appears to accept anything modelling the ForwardSequence concept as its template argument, so it seems like inherit should do the same unless I'm just missing something.

2 1

[lambda] instantiation in bind
by Igor R 21 Jul '09

21 Jul '09

In VC9.0 the following code gives the error message: 1>...\boost\tuple\detail\tuple_basic.hpp(419) : error C2259: 'Base' : cannot instantiate abstract class If the Base is not abstract, the code compiles and works as expected. Unfortunately, I'm unable to realise why it tries to instantiate the Base, and how to overcome this issue, so any enlightenment would be greatly appreciated! class Base { public: virtual void close() = 0; }; class Derived : public Base { public: void close() {} }; #include <boost/shared_ptr.hpp> typedef boost::shared_ptr<Base> BasePtr; typedef boost::shared_ptr<BasePtr> BasePtrPtr; #include <boost/lambda/lambda.hpp> #include <boost/lambda/bind.hpp> namespace bl = boost::lambda; BasePtrPtr makeHolder(Base *base) { return BasePtrPtr(new BasePtr(base), bl::bind(&Base::close, **bl::_1)); } int main() { Base *b = new Derived(); makeHolder(b); return 0; }

3 4

[boost][format] How does one determine the number of format arguments?
by Neil Groves 21 Jul '09

21 Jul '09

Dear all, Please accept my apologies if this ends up being a double post. I am having trouble with posting to the boost mailing lists. I have the need to obtain the number of 'arguments' (the %1% thingies!) in a boost::format. Is this functionality supported in a manner that I have failed to find in the documentation? If it is not supported, would it be a worthwhile addition since it seems easy to add assuming the num_args_ member variable is the information I require and that validation code of boost::format instances is commonly very useful. I have rarely executed error code that uses boost::format that is difficult to test, but by adding a StringTable that loads at module startup and wraps the boost format, I can provide compile-time validation of much of the formatting code, and runtime validation of the format strings at module startup. This would make it much easier to test. Thank you for your time and consideration. Regards, Neil Groves

3 2

[Math Toolkit] Gamma Distribution
by Christian Henning 21 Jul '09

21 Jul '09

Hi there, the documention of the gamma distribution talks about two different forms. How can can generate the second one with shape=1 and scale=1? I know how to get the first form: boost::math::gamma_distribution<> dist( 1, 1 ); I don't see how to get the second one? Thanks, Christian

5 13

Spirit 1.39 compilation times
by Avi Bahra 21 Jul '09

21 Jul '09

>First, you should use Boost Trunk, it compiles quite noticeably faster. >Second, look in the Spirit examples directly with Boost Trunk, in the >QI subdirectory, the mini_c directory has an example of just that. I cant use use boost trunk, in order to get faster compiles, since I have to use boost1.38/1.39. Also it looks like the grammar construction in the examples is different. I looked for a similar example in boost 1.39 since we use closures but none exists. Surely anyone who is using spirit and building on multiple platforms must be be affected with this bug ? Best regards. Avi ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// Message: 4 Date: Mon, 20 Jul 2009 10:37:57 +0100 From: Avi Bahra <avibahra(a)googlemail.com> Subject: [Boost-users] Spirit 1.39 compilation times To: boost-users(a)lists.boost.org Message-ID: <af1b101f0907200237ofdf91bbk98be95d0ab817208(a)mail.gmail.com> Content-Type: text/plain; charset="iso-8859-1" I am using boost spirit 1.39. On gcc the compilation times are acceptable. However on AIX with v10.1 compiler, it fails to even compile. After 24hrs waiting, I gave up. Ok I realise that the grammer needs to split. Is there any _small_ examples out there, that demonstrates this technique for a beginner ? In my case I have one grammer and ~70 rules. Take the following rule: rule<ScannerT, string_closure::context_t> identifier ; identifier = lexeme_d [ (alnum_p || ch_p('_')) >> *(alnum_p || ch_p('_')) ] [ identifier.name = construct_<std::string>(arg1, arg2) ] ; What is the technique for converting it into a grammer, that is separately compiled ? Also how would I reference it again, without including the header file, from another rule dotPath = ( str_p(".") >> +( str_p("/") >> identifier ) ) [ dotPath.name = construct_<std::string>(arg1, arg2) ] ; Any help would greatly appreciated. Best regards, Ta, Avi

3 2

[data_time] Are posix_time functions reentrant ?
by Ruediger Berlich 21 Jul '09

21 Jul '09

Hi there, here is a question I didn't find answers on in the documentation: Are the posix_time functions reentrant ? In particular: In my library a number of threads instantiate objects that internally measure the elapsed time like this: // Measure the start time ptime startTime = boost::posix_time::microsec_clock::local_time(); // [...] // Has a time-out been set and if so, have we exceeded it ? if(firstTimeOut_.total_microseconds() && ((microsec_clock::local_time()-startTime) > firstTimeOut_)) { // do something } Best Regards, Ruediger

2 1

A question on member functions in a class using lambda expression? Can anyone with kindness help me?
by fmingu 21 Jul '09

21 Jul '09

I am using Dev-C++ 4.9.8.0,with lambda expression: ...................... class A{ ............. int result; double function2(const int& n) ......................... double function3(const int& Q,const int& N) ...................... double function1(const int& q, int& n){ double result=0.0; int loopvalue=0, totaltime=n; n%=q; ++thenumbern; ( for_loop(var(loopvalue)=0,var(loopvalue)<=var(totaltime),++var(loopvalue), var(result)+=( bind(&A::function2,_2)*bind(&A::function3,_1,_2))))(q,n); return result; } But the complier told me that: ............................... \boost\lambda\detail\select_functions.hpp instantiated from `Arg::sig<boost::tuples::tuple<A&,................. instantiated from `RET boost::lambda::lambda_functor_base<boost::lambda::forloop_action, Args>::call(A&, B&, C.................... \boost\lambda\detail\lambda_functors.hpp instantiated from `T::sig<boost::tuples::tuple<A&, B&, boost::tuples::null_type, boost::tuples::null_type, ............... instantiated from here By the way, is the result in var(result) the double result in function1 or the int result in class A? Can any one with kindness help me?

3 3

[asio] async_resolve problems
by Roman Shmelev 21 Jul '09

21 Jul '09

Hi! I've written _test_ dummy service, working in single thread (prototype for more useful app). It accepts strings like "google.com" (and "gosdkfjslfjle.com"), resolves them and returns ip of the host. I noticed, that bad hostnames make it think too much, preventing other requests from being processed. So, I decided to use async_resolve approach, hoping that now I will be able to handle many requests concurrently. But it seems that async_resolve is not as async as desired - it feels like separate thread resolves in sync way all hosts. The question is - how to make async_resolve be really async? Am I missing something?

3 3

Suggestions/advice using boost::test
by Ahmed Badran 21 Jul '09

21 Jul '09

Hi, I want to use boost test in a particular way, I'm not sure if it's even possible, suggestions/advice are greatly appreciated. I have a shared library consisting of many classes. I would like to write my BOOST_AUTO_TEST_CASE at the end of the implementation of each class so as to encourage always writing tests. So this is what I've done, included <boost/test/unit_test.hpp> in the implementation file of each class, and wrote the test cases for each class at the end of the .cpp file for that class. In one of the .cpp files, I've stuck the following at the top to produce a canned 'main': #ifdef TEST #define BOOST_TEST_MAIN #include <boost/test/included/unit_test.hpp> #endif and compiled this particular file twice, so now I have a library without a main, and an executable with a main. The self-contained executable runs the tests perfectly as intended. The problem is that now the library is dependent on boost::test, any program that links against this library now needs to also link against the boost::test library (otherwise I get linker errors) which is _highly_ undesirable. It is perfectly fine however, if the tests get included/instantiated in the library but never get called, that's a perfectly valid solution. I also don't want to wrap the inclusion/code for the tests in ifdefs, since this will mean that the code (which is a lot) will have to be compiled twice and this process gets messy and tedious. Is there a way to accomplish this? Regards, Ahmed

1 0

[python][lambda] thin wrapper functions using boost.lambda
by Václav Šmilauer 21 Jul '09

21 Jul '09

Hi, I would like to use lambda functions to define thin wrapper functions in boost::python, like struct A{ void meth1(); }; [...] python::class_<A>("A").def("meth1",lambda::bind(&A::meth1,lambda::_1)); and I am getting "error: no matching function for call to 'get_signature [...]' " from the compiler. Reading boost::lambda documentation (http://www.boost.org/doc/libs/1_39_0/doc/html/lambda/using_library.html#lam…) """The basic rule is that a lambda function can be called with arguments of any types, as long as the lambda expression with substitutions performed is a valid C++ expression.""" Is there a way to enforce parameter types, so that boost::python can get the signature? Or, more generally, is it somehow possible to use such in-place functions to create the thin wrappers? Many thanks, Vaclav

3 3

variant<T&> with throwing initialization code.
by Zachary Turner 20 Jul '09

20 Jul '09

I have a couple of variants whose inner type are all references. For example, variant<a1&, a2&, ...> var1; variant<b1&, b2&, ...> var2; Because this variant will not be assignable due to the references, they need to be initialized in a constructor initializer list. However, it's possible for the function that creates these references in the first place to throw, and moreover the inner types might contain unmanaged resources that need to be properly cleaned up on a throw. Also the construction of var_k, might depend upon the successful construction of var_j where j<k (like for example I might need to use some member variable in one of the previously constructed objects to pass to one of the later objects to construct it). Is there a decent way to handle this? The problem is that if, say, something throws while trying to create var2, but var1 has already been created, then var1's unmanaged resources need to be properly freed. I tried to package all the variants together in a class such as typedef variant<a1&, a2&, ...> v1_type; typedef variant<b1&, b2&, ...> v2_type; struct package { package(v1_type& v1, v2_type& v2) : var1_(v1), var2_(v2) { } v1_type var1_; v2_type var2_; }; and then only have one instance of package in my class. Then, in the initializer list call package_(create_package()), and define create_package() as: package create_package() { optional<v1_type> v1; optional<v2_type> v2; try { v1 = create_v1(); v2 = create_v2(); return package(v1.get(), v2.get()); } catch(...) { } } but for some reason this fails to compile, saying that it's trying to call variant::operator= in the package constructor. Which doesn't make sense since I don't think it should be.

1 0

BGL: copy_graph does not work with bundled properties
by caligulaemperorofrome＠yahoo.com 20 Jul '09

20 Jul '09

Hi, copy_graph is not working with bundled property. See this minimal example program (40 lines) at http://tinyurl.com/lxh577. Just download make and execute - very easy. Thanks - Caligula

1 0

remove boost variable names from gcc compiled file
by ndgs＠lpemail.com 20 Jul '09

20 Jul '09

After I compile my file with g++ using the -s flag all that is striped out are my own variable names. I used xxd to look a the compiled output and "boost" is still all over the binary. How can I remove them? I used g++ with /usr/lib/librt.a and /usr/local/lib/libboost_thread-gcc43-mt.a I want to release this as a binary with no source code and I would prefer it if there wasn't boost this and boost that in the binary. Thanks, Piano

2 1

[phoenix] using dereference and increment operator
by Kim Kuen Tang 20 Jul '09

20 Jul '09

Hi all, the following piece of code compiled with boost 1.39: std::vector<double> u(11,1.0), w(11,3.0); std::vector<double>::const_iterator it=w.begin(); double value=(*ph::ref(it)++)(); But if i use the expression *ph::ref(it)++ as a functional in an algorithm, std::generate(u.begin(),u.end(),*phoenix::ref(it)++ ); the compilation fails with the following error: >c:\boost\boost_msvc_2008\include\boost-1_39\boost\fusion\container\vector\detail\vector_n.hpp(68) : error C2440: '<function-style-cast>' : cannot convert from 'const boost::phoenix::composite<EvalPolicy,EvalTuple>' to 'boost::fusion::vector_data1<Derived,T0>' 1> with 1> [ 1> EvalPolicy=boost::phoenix::post_increment_eval, 1> EvalTuple=boost::fusion::vector<boost::phoenix::reference<std::_Vector_const_iterator<double,std::allocator<double>>>,boost::fusion::void_,boost::fusion::void_,boost::fusion::void_,boost::fusion::void_,boost::fusion::void_,boost::fusion::void_,boost::fusion::void_,boost::fusion::void_,boost::fusion::void_> 1> ] 1> and 1> [ 1> Derived=boost::fusion::vector1<boost::phoenix::reference<std::_Vector_const_iterator<double,std::allocator<double>>>>, 1> T0=boost::phoenix::reference<std::_Vector_const_iterator<double,std::allocator<double>>> 1> ] 1> No constructor could take the source type, or constructor overload resolution was ambiguous I remember that the compilation was fine when using boost 1.38. ( I need to check this.) Here is the code that produces the errors: # include <algorithm> # include <vector> # include <boost/spirit/include/phoenix.hpp> using namespace boost; int main() { std::vector<double> u(11,1.0), w(11,3.0); std::vector<double>::const_iterator it=w.begin(); double value=(*phoenix::ref(it)++)(); it=w.begin(); std::generate(u.begin(),u.end(),*phoenix::ref(it)++ ); return 0; } With regards, Kim Tang

2 3

Definition of string in shared memory
by Samuel Gallard 20 Jul '09

20 Jul '09

Hello, I can't useful examples about how to store string in shared memory. I have to share a queue of Element objects (queue<Element> *) between processes. This is the signature of Element class : class Element { ... private : string dir; string file; ... } I used a shared_memory_object to use the queue. When I test it with thread, it is working fine. But when i test it with process, I obtained some segmentation fault because it can't retrieve the attributes for an Element object. it is normal because I am creating the Element object in one process the second can't access it. I have seen in boost documentation the template interprocess:basic_string. But I have no clue how to use it. I would like to have something like : typedef basic_string< string, segment> mystring class Element { ... private : mystring dir; mystring file; ... } Could you someone show me how to do that in the creation but also in the implementation. Thanks you in advance Regards Samuel

1 0

Locale and Boost.Serialization
by Luca Cappa 20 Jul '09

20 Jul '09

Hello, on the Windows platform, in my own application (built in unicode char econding) I am experiencing this problem with Boost 1.38: if I change the locale before loading an xml archive, then the application will hang in the loading of the archive file. I must add that this only occurs in a Release build on Windows Vista, but it never occurs on Windows XP. Also, if I run the application on Vista with the Compatibility Mode set to Windows XP SP2, it does work and does not hang while loading the archive. I tryed to get a stack trace with a ReleaseWithDebInfo on Windows Vista, here it is: Warning: stack unwind information unavailable. Following frames may be wrong. application!std::vector<double, std::allocator<double> > ::push_back+0xblabl1 application!std::vector<double, std::allocator<double> > ::push_back+0xblabl2 application!std::vector<double, std::allocator<double> > ::push_back+0xblabl3 application!boost::archive::detail::oserializer<boost::archive::xml_oarchive, boost::shared_ptr<TriggerUnit> > ::save_object_data+0x18b50 application!0xblabla Any hint what to look at to resolve this issue? Many thanks. Best Regards, Luca

1 0

Spirit 1.39 compilation times
by Avi Bahra 20 Jul '09

20 Jul '09

I am using boost spirit 1.39. On gcc the compilation times are acceptable. However on AIX with v10.1 compiler, it fails to even compile. After 24hrs waiting, I gave up. Ok I realise that the grammer needs to split. Is there any _small_ examples out there, that demonstrates this technique for a beginner ? In my case I have one grammer and ~70 rules. Take the following rule: rule<ScannerT, string_closure::context_t> identifier ; identifier = lexeme_d [ (alnum_p || ch_p('_')) >> *(alnum_p || ch_p('_')) ] [ identifier.name = construct_<std::string>(arg1, arg2) ] ; What is the technique for converting it into a grammer, that is separately compiled ? Also how would I reference it again, without including the header file, from another rule dotPath = ( str_p(".") >> +( str_p("/") >> identifier ) ) [ dotPath.name = construct_<std::string>(arg1, arg2) ] ; Any help would greatly appreciated. Best regards, Ta, Avi

2 1

[Asio] How do I dynamically select a port?
by Robert Jones 19 Jul '09

19 Jul '09

Hi All I'm new to Asio, and not a tcp/ip expert either, so if I'm approaching this all wrong then please tell me. I have two process on one machine, one of which streams a file to the other. In fact there may be several 'others', but one other will suffice to phrase the question. I need to dynamically select port numbers on which to stream the files to these other processes, based on whether that port number is available, and once the viability of using a particular port is established that port number can be communicated to the other process(es) by other means. So, the code sequence I have is this. At what point can I detect that use of this particular port number is ok? Is it appropriate to use from_string("127.0.0.1"), or should I use some other form for local sockets? Comments appreciated. using namespace boost; using asio :: ip :: tcp; unsigned short port = 2001; asio :: io_service io; tcp :: endpoint endpoint( asio :: ip :: address :: from_string( "127.0.0.1" ), port ); tcp :: acceptor acceptor( io, endpoint ); tcp :: iostream socket_stream; acceptor.accept( * socket_stream.rdbuf( ) ); Thanks, - Rob.

2 2