- Boost-users - lists.boost.org

Re: [Boost-users] Multi_indexcomposite_key
by isjk 26 Jul '10

26 Jul '10

there are one hundred million defferent manufacturer, all have "A4", "A8", "A12", "A16"... model car. manufacturer and model is many-to-many relationship.</:includetail> Particularly large data.</:includetail> </:includetail></:includetail> struct _Modify_price</:includetail> {</:includetail> _Modify_price(int price_):m_price_(price_){}</:includetail> void operator()(car_model& cm_)</:includetail> {</:includetail> cm_.price = m_price_;</:includetail> }</:includetail> private:</:includetail> int m_price_;</:includetail> }; ------------------ 原始邮件 ------------------ 发件人: "Joaquin M Lopez Munoz"<joaquin(a)tid.es>; 发送时间: 2010年7月26日(星期一) 凌晨2:49 收件人: "boost-users"<boost-users(a)lists.boost.org>; 主题: Re: [Boost-users] Multi_indexcomposite_key isjk <isjk <at> foxmail.com> writes: > Oh, sorry. Nothing to be sorry about. On another topic, please don't top-post and try to follow the discussion policies of the list as summarized at http://www.boost.org/community/policy.html > Assume "cadillac" have "A4" model, "bmw" have "A4" model too. > now, if I will querying "A4" 's price, I must kown the manufacturer. > > ordered_no_unique< > tag<model>,BOOST_MULTI_INDEX_MEMBER(car_model,std::string,model), Correct, you must change from the example's original ordered_unique to ordered_no_unique so as to allow cars with from different manufacturers yet the same model name. > ct.insert(car_model("A4",cadillac,78901)); > ct.insert(car_model("A4",bmw,65412)); > > index_iterator<car_table,model>::type im0,im1; > boost::tuples::tie(im0,im1)=get<model>(ct).equal_range(model); > > car_table_manufacturer_view ctmv; > copy(im0,im1,back_inserter(ctmv)); > car_table_manufacturer_view::iterator ictmv = ctmv.find(cm); This looks right (haven't compiled it), although a little contrived: unless you expect to have *many* cars with the same model name, you can just traverse [im0,im1) and locate the car you're after. > ctmv.modify(cctmv, _Modify_price(10000)); // right or wrong??? I don't know how you've defined _Modify_price. Care to show? Joaquín M López Muoz Telefónica, Investigación y Desarrollo _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users </:includetail>

2 1

[serialization] Reading back old archives
by David Raulo 26 Jul '10

26 Jul '10

Hello, I have a problem reading back serialization archives generated by older version of boost. I used to have a YYYYMMDD class versioning scheme. This broke with boost 1.42, which truncates the class version numbers to 16 bits, causing my code to generated corrupted archives. I fixed my code by implementing my own class versioning when compiling with boost 1.42, so I am again able to create archives that I can read back. Unfortunately this fixes only half of the problem. How can I read back archives that were saved using older releases of the serialization library? I saw that someone else opened ticket 3990 for the same problem. Here you suggested using the function get_library_version() in the following way: template<class Archive> void serialization(Archive & ar, unsigned int version){ library_version_type library_version = get_library_version if(library_version < 6) // the library version is a date my_date = version; else{ // don't use the boost serialization number, use ours instead ar & my_date; } ar && old_data_items if(ar.is_loading) if(my_date > "jan 10, 2010") ar & new_data_item; but I can't get this to work: the version argument is already truncated, and can't be interpreted as a date. As the bug reporter said, this is a case of data loss. Should I reopen ticket 3990, or is this code supposed to work with svn (I admit I did not test the trunk)? Thanks for your help, David.

2 11

[Serialization][BGL] problem serializing an edge
by Ireneusz Szcześniak 25 Jul '10

25 Jul '10

Hi, I have a problem serializing an edge in a graph. This is my test case: #include <iostream> #include <boost/archive/text_oarchive.hpp> #include <boost/graph/adj_list_serialize.hpp> using namespace std; using namespace boost; int main() { typedef adjacency_list<listS, vecS, undirectedS> Graph; typedef graph_traits<Graph>::edge_descriptor Edge; Graph g; Edge e; boost::archive::text_oarchive oa(cout); oa << g; oa << e; return 0; } Compilation with gcc version 4.4.1 returns this error: /usr/include/boost/serialization/access.hpp:109: error: ‘class boost::detail::edge_desc_impl<boost::undirected_tag, unsigned int>’ has no member named ‘serialize’ I would appreciate it if someone could point out what the problem is and how to fix it. Thanks, Irek -- Ireneusz (Irek) Szczesniak http://www.irkos.org

3 6

Re: [Boost-users] ¤Multi_indexcomposite_key~~
by isjk 25 Jul '10

25 Jul '10

Oh, sorry. Assume "cadillac" have "A4" model, "bmw" have "A4" model too. now, if I will querying "A4" 's price, I must kown the manufacturer. ordered_no_unique< tag<model>,BOOST_MULTI_INDEX_MEMBER(car_model,std::string,model) >,ct.insert(car_model("A4",cadillac,78901));ct.insert(car_model("A4",bmw,65412)); index_iterator<car_table,model>::type im0,im1; boost::tuples::tie(im0,im1)=get<model>(ct).equal_range(model);car_table_manufacturer_view ctmv;copy(im0,im1,back_inserter(ctmv));car_table_manufacturer_view::iterator ictmv = ctmv.find(cm); ctmv.modify(cctmv, _Modify_price(10000)); // right or wrong??? ------------------ 原始邮件 ------------------ 发件人: "Joaquin M Lopez Munoz"<joaquin(a)tid.es>; 发送时间: 2010年7月25日(星期天) 晚上7:17 收件人: "boost-users"<boost-users(a)lists.boost.org>; 主题: Re: [Boost-users]¤Multi_indexcomposite_key~~ isjk <isjk <at> foxmail.com> writes: > > > thanks a lot! > > now, I have a new problem: > > Multi_index example 6: complex searches and forergn keys, > struct car_manufacturer > { > std::string name; > }; > > struct car_model > { > std::string model; > car_manufacturer* manufacturer; > int price; > }; > > I want use "model, name" to query "price", and I maybe to modify "price". > > How to do it? thanks. Sorry but your question is not clear to me: what do you mean by querying "price" using "model, name"? If you know the model of a car then you just can use find() in the car_table to locate the element and access the price. As for modifying, there's a tutorial section on updating elements of a multi_index_container at http://www.boost.org/libs/multi_index/doc/tutorial/basics.html#ord_updating Joaquín M López Muoz Telefónica, Investigación y Desarrollo _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

2 1

Boost link error with mingw
by Eugen Dedu 25 Jul '10

25 Jul '10

Hi everybody, We compile ekiga for windows and boost signals is one of its dependencies, so we need to compile it for windows. I use gcc-mingw32 from debian unstable (gcc configured for mingw, version 4.4.4-0.1) and I do like this: - download boost 1.43.0 - echo "using gcc : : i586-mingw32msvc-gcc : ;" >$(BOOST_DIR)/user-config.jam - cd $(BOOST_DIR); ./bjam $(confboost) with: confboost := --user-config=user-config.jam --layout=system variant=release threading=multi link=shared runtime-link=shared target-os=windows threadapi=win32 --without-math --without-python --without-mpi --without-serialization --without-test --without-thread --without-wave --without-date_time --without-system --without-filesystem --without-graph --without-graph_parallel --without-iostreams --without-program_options --without-regex --without-random -q stage 1. I have a link error: Component configuration: - date_time : not building - filesystem : not building - graph : not building - graph_parallel : not building - iostreams : not building - math : not building - mpi : not building - program_options : not building - python : not building - random : not building - regex : not building - serialization : not building - signals : building - system : not building - test : not building - thread : not building - wave : not building ...patience... ...found 601 targets... ...updating 4 targets... gcc.link.dll bin.v2/libs/signals/build/gcc-mingw-4.4.4/release/target-os-windows/threadapi-win32/threading-multi/libboost_signals.dll.a Creating library file: bin.v2/libs/signals/build/gcc-mingw-4.4.4/release/target-os-windows/threadapi-win32/threading-multi/libboost_signals.dll.abin.v2/libs/signals/build/gcc-mingw-4.4.4/release/target-os-windows/threadapi-win32/threading-multi/trackable.o:trackable.cpp:(.text+0xc): undefined reference to `___gxx_personality_sj0' bin.v2/libs/signals/build/gcc-mingw-4.4.4/release/target-os-windows/threadapi-win32/threading-multi/trackable.o:trackable.cpp:(.text+0x6d): undefined reference to `operator delete(void*)' bin.v2/libs/signals/build/gcc-mingw-4.4.4/release/target-os-windows/threadapi-win32/threading-multi/trackable.o:trackable.cpp:(.text+0xac): undefined reference to `___gxx_personality_sj0' bin.v2/libs/signals/build/gcc-mingw-4.4.4/release/target-os-windows/threadapi-win32/threading-multi/trackable.o:trackable.cpp:(.text+0x10d): undefined reference to `operator delete(void*)' [...] I have discovered that the link can proceed if I do: - echo "using gcc : : i586-mingw32msvc-g++ : ;" >$(BOOST_DIR)/user-config.jam - cd $(BOOST_DIR); ./bjam $(confboost) i.e. compile again after changing gcc to g++, but of course this is a workaround. How to solve it properly? (here is the output after changing to g++: [..] - thread : not building - wave : not building ...patience... ...found 601 targets... ...updating 4 targets... gcc.link.dll bin.v2/libs/signals/build/gcc-mingw-4.4.4/release/target-os-windows/threadapi-win32/threading-multi/libboost_signals.dll.a Creating library file: bin.v2/libs/signals/build/gcc-mingw-4.4.4/release/target-os-windows/threadapi-win32/threading-multi/libboost_signals.dll.a common.copy stage/lib/libboost_signals.dll.a common.copy stage/lib/libboost_signals.dll ...updated 4 targets... ) Second, I execute afterwards: - cd $(BOOST_DIR); ./bjam $(confboost) install 2. Is $(confboost) still needed or not for install? 3. This command does nothing for about 1 minute 30 secs (I thought at the beginning that it simply does not work, fortunately I let it execute more than 10 seconds to see what happens), and only afterwards copy the files. Executing through strace shows many lines like this: brk(0x6116000) = 0x6116000 and these lines take almost all the installation time. Best regards, -- Eugen Dedu

2 2

回复： Multi_indexcomposite_key~~
by isjk 25 Jul '10

25 Jul '10

thanks a lot! now, I have a new problem: Multi_index example 6: complex searches and forergn keys, struct car_manufacturer { std::string name; }; struct car_model { std::string model; car_manufacturer* manufacturer; int price; }; I want use "model, name" to query "price", and I maybe to modify "price". How to do it? thanks. ------------------ 原始邮件 ------------------ 发件人: "Joaquin M Lopez Munoz"<joaquin(a)tid.es>; 发送时间: 2010年7月22日(星期四) 凌晨2:08 收件人: "boost-users"<boost-users(a)lists.boost.org>; 主题: Re: [Boost-users] Multi_indexcomposite_key~~ isjk <isjk <at> foxmail.com> writes: > > hi all, > > now,I have a composite_key like this, > struct leader_key > boost::multi_index::composite_key< > channel_info, > BOOST_MULTI_INDEX_MEMBER(channel_info, boost::uint64_t, st_md5_), > BOOST_MULTI_INDEX_MEMBER(channel_info, boost::uint32_t, st_rank_), > BOOST_MULTI_INDEX_MEMBER(channel_info, boost::uint8_t, st_degrees_), > BOOST_MULTI_INDEX_MEMBER( > channel_info, boost::uint8_t, st_lan_type_) > >{}; > > I used "md5, rank, degrees, lan_type" to query. > I want to the result like this, > equal to md5, > less than rank > degrees from small to large > equal to lan_type You can't do this by issuing a single lookup operation. A composite key on keys k1,...,kn allows you to do lookup operations of the form k1==a1, ... , kj==aj, kj+1==aj+1 k1==a1, ... , kj==aj, kj+1< aj+1 k1==a1, ... , kj==aj, kj+1<=aj+1 k1==a1, ... , kj==aj, kj+1> aj+1 k1==a1, ... , kj==aj, kj+1>=aj+1 For any j in the range [1,n], that is, you can query for the exact value of the first j keys and add any type of inequality for the last key being queried. In your particular case you can rearrange the key extractors of your composite key in the following order md5,lan_type,rank,degress then use lower_bound with elements equal to md5 equal to lan_type less than rank (note you've had to rearrange the extractors in leader_key to be able to do this) which gives you some result "it", and finally manually traverse the range [begin,it) to obtain those elements with degrees from small to large. Hope this helps, Joaquín M López Muoz Telefónica, Investigación y Desarrollo _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

2 1

boost::xpressive - Regex stack space exhausted
by Pavol Supa 25 Jul '10

25 Jul '10

Hi, I need to match a hex-written byte array, optionally separated with spaces. So i tried: boost::xpressive::sregex r = * ( * blank >> repeat<2,2> (xdigit)); smatch match; regex_match (input, match, r); when i use input of approx. 150 hex pairs, i get an exception "Regex stack space exhausted" (i use default stack size by Visual studio 2008) This pattern looks quite simple, so I'd like to know, if there is some fundamental problem with this expressions. I know i can enlarge the stack manually, but i expect the input to be cca 10kB long, depending on user input, so i consider it not being a good solution. Pavol

4 14

[units] inconvent printing of composite scaled units
by alfC 25 Jul '10

25 Jul '10

In the following program, a quantity of fempto*second unit is printed as 0 fs (correctly) but a quantity of composed units composed meter over femptosecond is printed as P( m s^-1) I was expecting: m fs^-1 At first I though it was a bug, where the P is coming from? where the parenthesis is coming from?, but then I realized that P stands for Peta (10e15). is there any workaround? For example an easy way to define fs_unit in some other way such that the f (fempto) keeps attached to the s (second). Below is the example code #include<boost/units/systems/si.hpp> #include<boost/units/make_scaled_unit.hpp> #include<iostream> using namespace boost::units; using namespace std; int main(){ typedef make_scaled_unit<si::time, scale<10, static_rational<-15> > >::type fs_unit; quantity<fs_unit> t(1.*si::second); quantity<si::length> l(1.*si::meter); quantity<divide_typeof_helper<si::length, fs_unit>::type > v = l/t; cout << t << endl; // prints 1e+15 fs cout << l << endl; // prints 1 m cout << v << endl; //prints 1e-15 P(m s^-1), notice the P return 0; }

2 1

Problem with list_of, make_shared and shared_ptr
by alapex0310＠gmail.com 24 Jul '10

24 Jul '10

Hi, I am using boost::list_of to pass a vector of pairs of string + "unsigned char" to SomeClass (as it can be seen in the typedef below - keys_t). This passes the whole vector to SomeClass which copies the vector in the keys_ private variable. The problem is: the copy ctor now copies the full vector in each SomeClass. To prevent this, as you can see, I have defined another data member which is a reference to const keys_t in my class and am initializing it in the member initialization list of the constructor with the value of keys_ and having the copy constructor copy just the reference of keys_ (rkeys_, that is - as you can see in the member initialization list of the copy ctor) instead of the full vector but the new SomeClass then holds an uninitialized vector (keys_) which is ugly and I hate seeing it in the class. Can I possible use Boost::make_shared with boost::list_of to return me a pointer to whatever (vector of type keys_t) boost::list_of returns and store that pointer instead in my class (e.g. keys_t typedef then reflects boost::shared_ptr instead of a traditional pointer) instead of the full vector. Specifying the arbitrary constants (apparently using boost::list_of) right in the constructor call is a requirement. Is there any way I can do it? Of course, I guess I could use boost::make_shared to return me a boost::shared_ptr to keys_t but my code is not even compiling and giving me errors in boost headers: std::vector<std::pair<std::string,unsigned char>> keys_t; class SomeClass { public: SomeClass(std::string nameStr, keys_t keys) : keys_(keys), rkeys_(keys_) {} SomeClass(const SomeClass &koa) : rkeys_(koa.keys_) {} private: keys_t keys_; const keys_t &rkeys_; }; SomeClass *psc = new SomeClass( "MyClass", list_of<pair<string,unsigned char>> ("one", 1) ("two", 2) ("three", 3) ("four", 4) ("five", 5) ("six", 6)); I read in boost::doc that I cannot pass shared_ptr directly as a function argument because of temporary problems. So, I cannot pass a shared_ptr to the constructor of SomeClass and that's the primary reason I hope if you can help me use make_shared or something similar to get the job done. The constraint is: I *must* pass the arbitrary number of pairs of string-unsigned-char in my constructor and that's why I am using list_of. Any way out? -Al

2 1

[asio] clang++ duplicate symbol error when compiling http server2 example
by Dane Springmeyer 24 Jul '10

24 Jul '10

I'm building boost trunk (r64316) using clang++ on mac os x 10.6. The standing library build/install went well, but I hit this error when trying to build one of the boost::asio examples: $ cd libs/asio/example/http/server2 $ ../../../../../bjam toolset=clang [...snip...] clang-darwin.compile.c++ ../../../../../bin.v2/libs/asio/example/http/server2/clang-darwin-2.8/debug/threading-multi/win_main.o clang: warning: argument unused during compilation: '-O0' In file included from win_main.cpp:13: In file included from ../../../../../boost/asio.hpp:20: In file included from ../../../../../boost/asio/basic_datagram_socket.hpp:20: In file included from ../../../../../boost/asio/basic_socket.hpp:19: In file included from ../../../../../boost/asio/basic_io_object.hpp:20: In file included from ../../../../../boost/asio/io_service.hpp:684: In file included from ../../../../../boost/asio/impl/io_service.hpp:18: ../../../../../boost/asio/detail/service_registry.hpp:38:15: warning: unknown pragma ignored [-Wunknown-pragmas] # pragma GCC visibility push (default) ^ ../../../../../boost/asio/detail/service_registry.hpp:47:15: warning: unknown pragma ignored [-Wunknown-pragmas] # pragma GCC visibility pop ^ 2 warnings generated. common.mkdir ../../../../../bin.v2/libs/thread/build/clang-darwin-2.8/debug common.mkdir ../../../../../bin.v2/libs/thread/build/clang-darwin-2.8/debug/threading-multi common.mkdir ../../../../../bin.v2/libs/thread/build/clang-darwin-2.8/debug/threading-multi/pthread clang-darwin.compile.c++ ../../../../../bin.v2/libs/thread/build/clang-darwin-2.8/debug/threading-multi/pthread/thread.o clang: warning: argument unused during compilation: '-O0' clang-darwin.compile.c++ ../../../../../bin.v2/libs/thread/build/clang-darwin-2.8/debug/threading-multi/pthread/once.o clang: warning: argument unused during compilation: '-O0' clang-darwin.link.dll ../../../../../bin.v2/libs/thread/build/clang-darwin-2.8/debug/threading-multi/libboost_thread.dylib common.mkdir ../../../../../bin.v2/libs/system/build/clang-darwin-2.8/debug common.mkdir ../../../../../bin.v2/libs/system/build/clang-darwin-2.8/debug/threading-multi clang-darwin.compile.c++ ../../../../../bin.v2/libs/system/build/clang-darwin-2.8/debug/threading-multi/error_code.o clang: warning: argument unused during compilation: '-O0' clang-darwin.link.dll ../../../../../bin.v2/libs/system/build/clang-darwin-2.8/debug/threading-multi/libboost_system.dylib clang-darwin.link ../../../../../bin.v2/libs/asio/example/http/server2/clang-darwin-2.8/debug/threading-multi/server ld: duplicate symbol __ZN5boost4asio6detail14do_throw_errorERKNS_6system10error_codeE in ../../../../../bin.v2/libs/asio/example/http/server2/clang-darwin-2.8/debug/threading-multi/io_service_pool.o and ../../../../../bin.v2/libs/asio/example/http/server2/clang-darwin-2.8/debug/threading-multi/connection.o clang: error: linker command failed with exit code 1 (use -v to see invocation) "clang++" -o "../../../../../bin.v2/libs/asio/example/http/server2/clang-darwin-2.8/debug/threading-multi/server" "../../../../../bin.v2/libs/asio/example/http/server2/clang-darwin-2.8/debug/threading-multi/connection.o" "../../../../../bin.v2/libs/asio/example/http/server2/clang-darwin-2.8/debug/threading-multi/io_service_pool.o" "../../../../../bin.v2/libs/asio/example/http/server2/clang-darwin-2.8/debug/threading-multi/mime_types.o" "../../../../../bin.v2/libs/asio/example/http/server2/clang-darwin-2.8/debug/threading-multi/posix_main.o" "../../../../../bin.v2/libs/asio/example/http/server2/clang-darwin-2.8/debug/threading-multi/reply.o" "../../../../../bin.v2/libs/asio/example/http/server2/clang-darwin-2.8/debug/threading-multi/request_handler.o" "../../../../../bin.v2/libs/asio/example/http/server2/clang-darwin-2.8/debug/threading-multi/request_parser.o" "../../../../../bin.v2/libs/asio/example/http/server2/clang-darwin-2.8/debug/threading-multi/server.o" "../../../../../bin.v2/libs/asio/example/http/server2/clang-darwin-2.8/debug/threading-multi/win_main.o" "../../../../../bin.v2/libs/thread/build/clang-darwin-2.8/debug/threading-multi/libboost_thread.dylib" "../../../../../bin.v2/libs/system/build/clang-darwin-2.8/debug/threading-multi/libboost_system.dylib" -g Does anyone have suggestions or workarounds? Thanks, Dane

1 0