August 2011 - Boost-users - lists.preview.boost.org

adding elements to a std::container in a single expression
by Peter Foelsche 14 Aug '11

14 Aug '11

until now I've usually written my own template to do this: template<typename T> struct push_back_container:T { push_back_container &_push_back(const typename T::value_type &_r) { push_back(_r); return *this; } }; one can use this like this: push_back_container<std::vector<something> >()._push_back(something1)._push_back(_something2) Is there some more elegant way to do this? Peter

2 1

[C++0x] ISO/IEC approves C++0x standard!
by Beman Dawes 13 Aug '11

13 Aug '11

The C++0x standard has been approved by ISO/IEC. The vote was unanimous; all 21 National Bodies voting were in favor. That means the dozen or so Boost libraries that were in TR1 are now part of the standard itself. Good work, Boosters! --Beman

2 1

[serialization] unregistered class exception when using BOOST_CLASS_EXPORT_GUID
by Ron Hashimshony 13 Aug '11

13 Aug '11

Hi all boost users, We have been using boost.serialization for years. We want to move from 1.33 to 1.47 and encountered a problem. Many of our classes are abstract, and the derived are defined with BOOST_CLASS_EXPORT_GUID. At one point in time, a new class (B) was added that derives from a class that was defined with the BOOST_CLASS_EXPORT_GUID macro (A). It all worked well in 1.33. In 1.47, when we get to de-serialize class A, we get an exception of unregistered class. We traced it down, and found out that if we remove the derived class (B), it works fine, but this is not an option, and then the derived class (B) cannot be de-serialized from the file... If there any way to bypass this problem? was there a change made that broke this functionality? Many Thanks! Ron

2 1

Re: [Boost-users] [!! SPAM] [Boost.Wave] util::file_position<> and asserted !is_escaped_lit(..) requirement
by Hartmut Kaiser 13 Aug '11

13 Aug '11

> (Using namespace boost::wave here on.) > > It seems that util::file_position<> has an assertion that checks if a > given filename contains any following sequences: <\\>, <\"> or <\?> (using > util::debug::is_escaped_lit). > > I wonder why file_position<> does need to check this and if it's so > crucial why it's done using a mere BOOST_ASSERT? > > Now, here's comes the reason why I'm asking this: At least with MSVC9's > remote debugger (or some other lowlevel component; not really the > interesting part > here) translates all network shares on a remote computer to UNC-paths. > This is correct behaviour and actually a very neat feature. However, as > you might be already aware at this point this leads to the actual problem: > Any UNC-path triggers (consider <\\foo\path\file>) the assertion and > things go boom. > > It is not possible to easily work around this by refactoring user code > and/or input values as provided filenames are completed internally using > util::complete_path (after which <file> becomes <\\foo\path\file> etc. in > this case). > > I tried to disable this assertion (actually, there's two; one in the > constructor and one in the set_file member function) and didn't see any > strange behaviour. E.g. std::ifstream handling with UNC-paths in > iteration_context_policies::load_file_to_string::inner<> worked just fine, > as expected from a native STL. > > Goes without saying that I'm using the latest Boost version (1.47.0). I > would be very pleased if anyone could shed some light on this matter or > better still > - fix it permanently. Right, that's apparently a leftover from old boost.filesystem times, when it did not support UNC paths. I'll remove the checks asap. Regards Hartmut --------------- http://boost-spirit.com

1 0

[serialization] overloading load_construct_data: no accessible constructor at all
by Griffin Howlett 13 Aug '11

13 Aug '11

Hey, so I'm using the boost::serialization library, and I'm trying to override how a class is constructed, since it has no default constructor. This is demonstrated here<http://www.boost.org/doc/libs/1_47_0/libs/serialization/doc/serialization.h…>. To me it appears the function ONLY takes the `class* t` and then sets it to point to a newly constructed object. If I'm wrong, this is definitely the source of my error. However, the only way to construct my class is by using another classes `create()` function, meaning I need to stray from the code in the example (this is stated in the boost::serialization namespace): `::new(t)my_class(attribute);` I tried simply calling the create function and setting `t` equal to the returned pointer, but this doesn't seem to work because right after the `load_construct_data function`, and in the `serialization function`, the given `myClass&` is not the same as what I set 't' to. *How do I do whatever ::new(t) is doing so the object created using the create function follows through into the serialize/other functions?* [1]: http://www.boost.org/doc/libs/1_47_0/libs/serialization/doc/serialization.h…

1 0

JSON Property file reader compile problem
by Scott Smith 13 Aug '11

13 Aug '11

I'm trying to use the boost property tree to read a json file. I'm using 1.46 on gcc 4.6 (Fedora 15). I'm also running on Eclipse (Indigo) which is relevant for something I mention below. Here's the issue . I have a JSON file which has an array of parameters which looks something like: { "config": { "modules": [ { "name": "somename0", "param1": "someparam", "param2":"someotherparm" }, { "name": "somename1", "param1": "someparam", "param2":"someotherparm" }, { "name": "somename2", "param1": "someparam", "param2":"someotherparm" } ] } } So, I wrote the following code (Note that I can't use boost:FOR_EACH because of some bug in eclipse (#332278) which causes an error-or so I believe): ptree v = pt.get<ptree>("config.modules"); for (ptree::iterator it = v.begin() ; it != v.end ; ++it) { ptree module = ???; } The first two lines seem to compile fine. I'm assuming that what I get from v is the "root" ptree (the parent) for the various "modules" objects and that the iterator will walk through them. However, I've tried all sorts of things trying to get at the module data (e.g., it->second, (*it)->second, ...) and clearly I don't know what it is I need to tell the compiler I want. All I get are compile errors. The compiler is giving me no hints as to what I need to tell it. Can someone tell me what I'm missing here? Thanks Scott

2 1

(no subject)
by Hasanul Ferdaus 12 Aug '11

12 Aug '11

living paycheck to paycheck wasnt working for me I told myself to stay positive without this id still be completely distraught! <a href="http://salanowyswiat.home.pl/PhilipBennett68.html">http://salanowyswiat.home.pl/PhilipBennett68.html</a> I waited so long for my time to shine

1 0

[ASIO] Send data on a socket
by Claude 12 Aug '11

12 Aug '11

I created a socket with BOOST asio and I use it in a server process that accepts connections from "remote" client. It is possible from the process that created the socket, send data to the same socket simulating the arrival of data from a client process? -- View this message in context: http://boost.2283326.n4.nabble.com/ASIO-Send-data-on-a-socket-tp3738099p373… Sent from the Boost - Users mailing list archive at Nabble.com.

3 3

[Boost.Wave] util::file_position<> and asserted !is_escaped_lit(..) requirement
by Pekka Seppänen 12 Aug '11

12 Aug '11

Hello. (Using namespace boost::wave here on.) It seems that util::file_position<> has an assertion that checks if a given filename contains any following sequences: <\\>, <\"> or <\?> (using util::debug::is_escaped_lit). I wonder why file_position<> does need to check this and if it's so crucial why it's done using a mere BOOST_ASSERT? Now, here's comes the reason why I'm asking this: At least with MSVC9's remote debugger (or some other lowlevel component; not really the interesting part here) translates all network shares on a remote computer to UNC-paths. This is correct behaviour and actually a very neat feature. However, as you might be already aware at this point this leads to the actual problem: Any UNC-path triggers (consider <\\foo\path\file>) the assertion and things go boom. It is not possible to easily work around this by refactoring user code and/or input values as provided filenames are completed internally using util::complete_path (after which <file> becomes <\\foo\path\file> etc. in this case). I tried to disable this assertion (actually, there's two; one in the constructor and one in the set_file member function) and didn't see any strange behaviour. E.g. std::ifstream handling with UNC-paths in iteration_context_policies::load_file_to_string::inner<> worked just fine, as expected from a native STL. Goes without saying that I'm using the latest Boost version (1.47.0). I would be very pleased if anyone could shed some light on this matter or better still - fix it permanently. Br, Pekka

1 0

async_read_some() not found??
by Claude 12 Aug '11

12 Aug '11

I have this code (from async_read_until boost documentation): boost::asio::streambuf b; void handler(const boost::system::error_code& e, std::size_t size) { if (!e) { std::istream is(&b); std::string line; std::getline(is, line); // ... } } boost::asio::async_read_until(s, b, "\r\n", handler); But I obtain this error on Ubuntu 11.04 (boost 1.45.0): boost/asio/impl/read_until.hpp|388|*error: request for member ‘async_read_some’* in ‘((boost::asio::detail::read_until_delim_op<boost::asio::basic_stream_socket<boost::asio::ip::tcp>*, std::allocator<char>, void (*)(const boost::system::error_code&, unsigned int)>*)this)->boost::asio::detail::read_until_delim_op<boost::asio::basic_stream_socket<boost::asio::ip::tcp>*, std::allocator<char>, void (*)(const boost::system::error_code&, unsigned int)>::stream_’, which is of non-class type ‘boost::asio::basic_stream_socket<boost::asio::ip::tcp>*’ I Included in my project: libpthread libboost_system -- View this message in context: http://boost.2283326.n4.nabble.com/async-read-some-not-found-tp3732368p3732… Sent from the Boost - Users mailing list archive at Nabble.com.

2 2

[filesystem] misidentifies file type on Solaris x86 64-bit
by Will Mason 11 Aug '11

11 Aug '11

Hello, I'm using Boost 1.46.1 with filesystem v3. Code that I've been running successfully on Windows, Linux and FreeBSD started producing very mysterious results on Solaris. I boiled it down to this: boost::filesystem will not identify a directory as a directory. Here's a test program: #include <boost/filesystem.hpp> int main() { try { boost::filesystem::path d(boost::filesystem::absolute("dir")); if (boost::filesystem::exists(d)) { std::cout << d << " already exists" << std::endl; } else { boost::filesystem::create_directory(d); std::cout << d << " is " << (boost::filesystem::is_directory(d) ? "" : "not") << " a directory" << std::endl; } } catch (std::exception& e) { std::cout << "Exception: " << e.what() << std::endl; } return 0; } And here is a snippet of a terminal session compiling and running the program: $ CC -o fstest fstest.cpp -library=stlport4 -Iexternal/solaris/x86/64/suncc/5.11/include external/solaris/x86/64/suncc/5.11/lib/libboost_filesystem.a external/solaris/x86/64/suncc/5.11/lib/libboost_system.a $ ./fstest "/export/home/will/Projects/cap/dir" is not a directory $ file /export/home/will/Projects/cap/dir /export/home/will/Projects/cap/dir: directory $ Thanks for any help, Will

1 0

Boost Unit test on Eclipse
by Yong Qin 11 Aug '11

11 Aug '11

I have tried to use boost unit test on Eclipse CDT w/ MinGW and VC++. But I can not link to the static boost library: libboost_unit_test_framework-vc100-mt-1_47.lib. The code I used: #defineBOOST_TEST_MAIN #defineBOOST_TEST_MODULEMasterTestSuite #include<boost/test/unit_test.hpp> BOOST_AUTO_TEST_SUITE(FooTests) BOOST_AUTO_TEST_CASE(ctor_test1) { Fooa; BOOST_CHECK_EQUAL(a.name(),MY_PROJECT_NAMESPACE::Foo::DEF_NAME); } BOOST_AUTO_TEST_CASE(ctor_test2) { conststd::stringtheName("Blah"); Fooa(theName); BOOST_CHECK_EQUAL(a.name(),theName); } BOOST_AUTO_TEST_SUITE_END() What is wrong?

4 4

program_options: accepting NaNs for doubles
by Rhys Ulerich 11 Aug '11

11 Aug '11

Hi all, program_options does not accept NaN or Inf as valid double input. For example, the following sequence modifies the "Getting Started" first.cpp [1] to accept double input: cp $BOOST_SOURCE/libs/program_options/example/first.cpp . perl -pi -e 's/<int>/<double>/;' first.cpp LDFLAGS="-lboost_program_options" make first # ...or whatever The modified example indeed barfs on NaN or Inf input: $ ./first --compression=5.5 Compression level was set to 5.5. $ ./first --compression=Inf error: in option 'compression': invalid option value 'Inf' $ ./first --compression=NaN error: in option 'compression': invalid option value 'NaN' Two questions: 1) Does this count as a bug? 'std::cin >> foo' doesn't grok Inf or NaNs (on my system anyhow) but methods like strtod() do. Users providing explicit infinities or not-a-numbers as options parameters likely don't do so accidentally. This Inf/NaN behavior is not mentioned in the active tickets. 2) Can anyone suggest a workaround (shy of using string-based parameters and calling strtod myself)? Thanks, Rhys [1] http://www.boost.org/doc/libs/1_47_0/doc/html/program_options/tutorial.html…

2 2

Re: [Boost-users] Boost-users post acknowledgement
by Yong Qin 11 Aug '11

11 Aug '11

for VC, I try to link to: libboost_test_exec_monitor-vc100-mt-1_47.lib and got error: LNK1561: entry point must be defined VCApp_Test line 0 C/C++ Problem for MinGW, I try to link to: libboost_test_exec_monitor-mgw46-mt-1_47.a and got error: undefined reference to `__chkstk_ms' MinGWApp_Test line 0, external location: D:\boost_1_47_0-MinGW\stage\static\lib\libboost_unit_test_framework-mgw46-mt-1_47.a(unit_test_parameters.o):unit_test_parameters.cpp C/C++ Problem undefined reference to `std::__detail::_List_node_base::_M_hook(std::__detail::_List_node_base*)' MinGWApp_Test line 0, external location: D:\boost_1_47_0-MinGW\stage\static\lib\libboost_unit_test_framework-mgw46-mt-1_47.a(unit_test_suite.o):unit_test_suite.cpp C/C++ Problem undefined reference to `std::__detail::_List_node_base::_M_unhook()' MinGWApp_Test line 0, external location: D:\boost_1_47_0-MinGW\stage\static\lib\libboost_unit_test_framework-mgw46-mt-1_47.a(unit_test_suite.o):unit_test_suite.cpp C/C++ Problem

1 0

question about Boost naming
by Yong Qin 11 Aug '11

11 Aug '11

I am using Boost test library now and find there are several libraries there: libboost_unit_test_framework-mgw46-1_47.a libboost_unit_test_framework-mgw46-mt-1_47.a libboost_unit_test_framework-mgw46-s-1_47.a what are differences between the 3 libraies (with "s", or with "mt", or no "mt" and "s")? also, for VC++ compiler, there is: boost_unit_test_framework-vc100-mt-1_47.dll but MinGW have: libboost_unit_test_framework-mgw46-mt-1_47.dll libboost_unit_test_framework-mgw46-1_47.dll Why MinGW have more?

6 6

[boost-build] Dynamic linking dependency propagation on mac os x
by Tim Odenthal 11 Aug '11

11 Aug '11

Dear list, i'm trying to compile an existing project which compiles fine on Windows (where we link statically) and linux (where linking is default, i.e. dynamic) on Mac OS X (10.7). I'm essentially trying to use the same configuration as on linux, since darwin ~ gcc… The project makes use of (and therefore links to) several boost libraries including system, filesystem, serialization,… We create several libraries which depend upon each other: - lib1depends on serialization,filesystem,… - lib2 depends on lib1 - lib3 depends on lib1 and lib2 -> here's the first problem: the (implicit) dependence on system, filesystem and serialization is not automatically resolved anymore. At least, the linker complains about missing symbols, since the "g++"-command-line doesn't include the libraries. I could solve that by adding those libraries to the requirements with <use>/boost/serialization, etc. -lib4 depends on lib3 -> here's the real problem: again, the implicit dependence on (specifically) lib1 and the boost-libs is lost. I can use the same trick as before with the boost-libs, but whatever I do, I cannot make the "lib1.dylib" appear in the "g++" command-line. When I execute the command by hand - adding lib1.dylib - it executes fine & the linker does not complain. How/why does that happen? My user-config.jam simply consists of: using darwin : : : <compileflags>-DBOOST_WAVE_SUPPORT_THREADING=0 ; What else do I need to configure differently on Mac OS X? Thanks a lot in advance for any suggestions Tim

1 0

Windows Installer Build
by asif saeed 11 Aug '11

11 Aug '11

Hi, I installed Boost 1.47 using Windows Installer. I have Visual C++ 2010 Express installed and I installed Platform SDK for 2003 and Windows Driver Kit, that give me MFC/ATL, etc., separately. However, I have to set the directories in each project before I build anything otherwise the VC++ 2010 picks up the Windows SDK 7.0 that gets installed by default by Visual C++ 2010 Express. So Boost installer, I suspect, too picked up the wrong SDK, Windows SDK 7, directories instead of the those of the platform SDK. I suspect so because upon including ASIO header and just getting an IP Address string, I get build warnings like these: Please define _WIN32_WINNT or _WIN32_WINDOWS appropriately. For example: - add -D_WIN32_WINNT=0x0501 to the compiler command line; or - add _WIN32_WINNT=0x0501 to your project's Preprocessor Definitions. Assuming _WIN32_WINNT=0x0501 (i.e. Windows XP target). File0001.cpp Please define _WIN32_WINNT or _WIN32_WINDOWS appropriately. For example: - add -D_WIN32_WINNT=0x0501 to the compiler command line; or - add _WIN32_WINNT=0x0501 to your project's Preprocessor Definitions. Assuming _WIN32_WINNT=0x0501 (i.e. Windows XP target). File0002.cpp Second_test.cpp Please define _WIN32_WINNT or _WIN32_WINDOWS appropriately. For example: - add -D_WIN32_WINNT=0x0501 to the compiler command line; or - add _WIN32_WINNT=0x0501 to your project's Preprocessor Definitions. Assuming _WIN32_WINNT=0x0501 (i.e. Windows XP target). Any idea how I can fix this? Best, Asif

1 0

Re: [Boost-users] [asio] user defined composed operations and strands
by Marat Abrarov 11 Aug '11

11 Aug '11

Hi, Tacheon. > If async_write_data is called and io_service::run is called from multiple threads, strands are should > be necessary to avoid raced. > The question is: How do I support handlers wrapped in a strand and how do I get this strand forwarded > to async_read? > As far as I understand this should be somehow possible by using asio_handler_invoke but the > documentation is a bit too fuzzy on the topic for me to understand. There is no such info in Asio docs. You may look at asio-samples (http://sourceforge.net/projects/asio-samples) There is ma::context_wrapped_handler class (and ma::make_context_wrapped_handler helper function) that makes such a wrapper for intermediate handler. But you shouldn't use > typedef boost::function<void (error_t const &)> completion_handler_t; because it hides asio_hanlder_invoke defined for the original (user supplied) handler. You may write something like this: class io_port { public: //Ctors and so on... template<typename Handler> void async_read_data ( raw_buffer_t & buffer , Handler const & handler) { //uses a composed operation itself... boost::asio::async_read ( socket , boost::asio::buffer(header) , ma::make_context_wrapped_handler(handler, boost::bind ( &io_port::handle_header_received<Handler> , this , boost::ref(buffer) , handler , boost::asio::placeholders::error))); } //first intermediate handler template<typename Handler> void handle_header_received ( raw_buffer_t & data , Handler const & handler , boost::system::error_code const & error { if(error) { handler(error_t(error)); } else { prepare_buffer(data, header.perparation_data()); //pseudo-code //uses a composed operation itself... boost::asio::async_read ( socket , boost::asio::buffer(data) , ma::make_context_wrapped_handler(handler, boost::bind(&io_port::handle_data_received<Handler> , this , boost::ref(buffer) , handler , boost::asio::placeholders::error) ); } } //second intermediate handler template<typename Handler> void handle_data_received ( raw_buffer_t & data , Handler const & handler , boost::system::error_code const & error ) { if(error) { handler(error_t(error)); } else { handler(error_t()); } } }; Also, it will be very helpful to read Asio sources: /boost/asio/impl/read.hpp: 339 and 145 (Boost 1.47). May be it' will be better to implement io_port class similar to boost::asio::detail::read_op. Regards, Marat Abrarov.

2 1

question about changing permission with boost
by rudy b 11 Aug '11

11 Aug '11

Hi, I am trying to copy a file from one folder to another, but I am also trying to access to the last_write_time of the file, BEFORE copying it to the other folder, And I notice, whenever I am inside a Read-only directory, and trying to access the "last_write_time" of a Read-only file, I get the following error message: what(): boost::filesystem::last_write_time: Permission denied: "src/ temp1/methodologies.ppt" In this scenario "temp1" is a Read-only directory, also methodologies.ppt is a Read-only file as well. Does anyone know how to get around this? Is there a way to change the directory/file permission (to write) prior to copying, and then change the permission back again? I need to access the time-stamp before copying the file, because right after I copy the file, the date-stamp will be updated. I would really appreciate some help, please advise, thanks, RB

1 0

[BGL] bundled properties and forward declarations (changes from 1.42 -> 1.46.1 ?)
by Anders Wallin 10 Aug '11

10 Aug '11

I have code using bundled properties roughly like this that compiles with boost 1.42: struct EdgeProperty; typedef adjacency_list< ... , EdgeProperty , ...> MyGraph; typedef graph_traits<MyGraph>::edge_descriptor Edge; struct EdgeProperty { Edge next; }; When trying to compile this with boost 1.46.1 I get "error: forward declaration of EdgeProperty’". I am not sure why. The obvious solution would be to move the definition of EdgeProperty before the MyGraph typedef, but then I run into a circular dependency problem as I want to store an edge_descriptor in EdgeProperty. any ideas? Anders

2 3

[bimap] Problem with finding elements in a vector_of
by Teto 10 Aug '11

10 Aug '11

Hi, I've a peculiar bimap defined as such: bbi::bimap< bbi::unordered_set_of< TFullId>, bbi::vector_of<CBindDescriptor>> And I want to find all the TFullId matching a certain CBindDescriptor. but I can't succeed in compiling. Here are 10 lines of code: http://boost.codepad.org/99V38S9U Thx for any hint matt

2 3

[BGL] resize/reserve on graph?
by Geoff Hilton 10 Aug '11

10 Aug '11

If I know that during the execution of the algorithm the graph size must be grown by X number of vertices in one go, is there a way this can be done to avoid multiple allocations such as when using std::vector's reserve(..) or resize(..) member functions without subverting the BGL's documented interface? Currently I use a for loop which iterates over add_vertex(vertex_property, graph). Can a reserve and/or resize equivalent otherwise please be added somewhere? Thank you, Geoff

3 6

installing boost_1_47_0
by rudy b 10 Aug '11

10 Aug '11

Hi, I am newb in using boost. And, I am having so much trouble installing and running a simple example. I would REALLY appreciate some help. I tried to install boost_1_47_0.tar.gz on my unix machine, and I followed the stpes http://www.boost.org/doc/libs/1_47_0/more/getting_started/unix-variants.html to first build, and install the libraries. I had no problem running some the examples (provided on the link above, that was using some of its already built libraries), and they all worked. The problem is when I try to run an example that is using a boost_filesystem. I tried uninstalling/installing different boost versions, cause I thought the error may be version dependent, but it wasn't. I also tried to built the entire library using different versions of gcc (3.4.6, 4.3.5, 4.4.0, 4.5.0, 4.5.1, 4.5.2, 4.6.0), and I get the same error messages. I even made sure that when compile, I have the boost_1_4x_0 path in my compile script. I also made sure that I have appended boost_1_4x_0 path to my $path. I even tried to have -lboost_filesystem in my compile script, but this didn't help either !!! This is the type of error that I am getting: /tmp/cczDoDOU.o: In function `__static_initialization_and_destruction_0(int, int)': sync.cpp:(.text+0xe03): undefined reference to `boost::system::generic_category()' sync.cpp:(.text+0xe1e): undefined reference to `boost::system::generic_category()' sync.cpp:(.text+0xe39): undefined reference to `boost::system::system_category()' /tmp/cczDoDOU.o: In function `boost::filesystem3::last_write_time(boost::filesystem3::path const&, long)': And, when I change the version of compiler, I get something like: /tmp/ccw1Sepi.s: Assembler messages: /tmp/ccw1Sepi.s:408: Error: unknown pseudo-op: `.cfi_personality' /tmp/ccw1Sepi.s:409: Error: unknown pseudo-op: `.cfi_lsda' /tmp/ccw1Sepi.s:531: Error: unknown pseudo-op: `.cfi_personality' /tmp/ccw1Sepi.s:532: Error: unknown pseudo-op: `.cfi_lsda' /tmp/ccw1Sepi.s:722: Error: unknown pseudo-op: `.cfi_personality' /tmp/ccw1Sepi.s:723: Error: unknown pseudo-op: `.cfi_lsda' /tmp/ccw1Sepi.s:1046: Error: unknown pseudo-op: `.cfi_personality' Please advise, thanks, RB

3 3

Program_Options - Is there anyway to get the name of executable?
by asif saeed 10 Aug '11

10 Aug '11

Hi, Is it possible to get name of the executable that started the program using program_options? Thanks in advance, Asif

3 2

More problems with boost::asio and async_read
by Allan Nielsen 10 Aug '11

10 Aug '11

Hi I posted on this a couple of hours ago with some problems. It turned out that some of the problems I had might have been due to buggy hardware. But there is still some thing I do not understand... So, now I have reproduced some of of the problems I have with out any hardware (except from a PC). Basically I'm trying to read from a input source (in this case stdin) with a timeout. Due to the design of the existing application where this have to fit is it not possible to call run on my io_service. Here is my try so far: #include <boost/asio.hpp> #include <boost/bind.hpp> #include <boost/optional.hpp> using namespace std; void set_result( boost::optional<boost::system::error_code> * a, boost::system::error_code b ) { if( b == 0) a->reset( b ); } void receive(boost::asio::io_service & io, boost::asio::posix::stream_descriptor & stream, boost::asio::streambuf & result){ boost::optional<boost::system::error_code> timer_result; boost::optional<boost::system::error_code> read_result; boost::asio::deadline_timer timer( io ); timer.expires_from_now( boost::posix_time::milliseconds(5000) ); // allow up to 50ms of timeout for every char timer.async_wait( boost::bind(&set_result, &timer_result, _1) ); boost::asio::async_read( stream, result, boost::asio::transfer_at_least(1), boost::bind( &set_result, &read_result, _1 )); boost::system::error_code ec; while(1) { io.reset(); io.poll_one(ec); if ( read_result ) { timer.cancel(); return; } else if ( timer_result ) throw std::runtime_error("timeout"); } } void receive(boost::asio::io_service & io, boost::asio::posix::stream_descriptor & stream, size_t size, boost::asio::streambuf & result){ while( result.size() < size ) receive(io, stream, result); } int main(int argc, const char *argv[]) { boost::asio::io_service io; boost::asio::posix::stream_descriptor in(io, ::dup(STDIN_FILENO)); for(int i = 0; i < 5; i++){ std::cout < "; std::cout.flush(); try { boost::asio::streambuf buf; receive(io, in, 5, buf); std::cout << "out> "; std::copy(boost::asio::buffer_cast<const char *>(buf.data()), boost::asio::buffer_cast<const char *>(buf.data()) + buf.size(), ostream_iterator<char>(std::cout)); } catch (const std::exception & e) { std::cout << e.what() << endl; } } return 0; } At my first look, I thought that it was working, but after playing more around with the test app, I got some segmentations faults. I found that something bad happens if I wait for the first query to timeout, and then enter 5 chars the next. Here is what valgrind says: ==17848== Memcheck, a memory error detector ==17848== Copyright (C) 2002-2010, and GNU GPL'd, by Julian Seward et al. ==17848== Using Valgrind-3.6.1 and LibVEX; rerun with -h for copyright info ==17848== Command: ./ccTalkScan ==17848== ==17848== 0 Type in 4 chareters and press enter in> ==17848== Invalid read of size 8 ==17848== at 0x546EB4A: ??? (setcontext.S:60) ==17848== by 0x425847: receive(boost::asio::io_service&, boost::asio::posix::basic_stream_descriptor<boost::asio::posix::stream_descriptor_service>&, unsigned long, boost::asio::basic_streambuf<std::allocator<char> >&) (ccTalkScan.cxx:3 20) ==17848== by 0x425982: main (ccTalkScan.cxx:334) ==17848== Address 0x7feffdf68 is not stack'd, malloc'd or (recently) free'd ==17848== ==17848== Invalid read of size 8 ==17848== at 0x546EB52: ??? (setcontext.S:63) ==17848== by 0x425847: receive(boost::asio::io_service&, boost::asio::posix::basic_stream_descriptor<boost::asio::posix::stream_descriptor_service>&, unsigned long, boost::asio::basic_streambuf<std::allocator<char> >&) (ccTalkScan.cxx:3 20) ==17848== by 0x425982: main (ccTalkScan.cxx:334) ==17848== Address 0x7feffdf58 is not stack'd, malloc'd or (recently) free'd ==17848== ==17848== Invalid read of size 8 ==17848== at 0x546EB59: ??? (setcontext.S:64) ==17848== by 0x425847: receive(boost::asio::io_service&, boost::asio::posix::basic_stream_descriptor<boost::asio::posix::stream_descriptor_service>&, unsigned long, boost::asio::basic_streambuf<std::allocator<char> >&) (ccTalkScan.cxx:320) ==17848== by 0x425982: main (ccTalkScan.cxx:334) ==17848== Address 0x7feffdf28 is not stack'd, malloc'd or (recently) free'd ==17848== ==17848== Invalid read of size 8 ==17848== at 0x546EB4A: ??? (setcontext.S:60) ==17848== by 0x5E4BD2C: (below main) (in /lib64/libc-2.12.2.so) ==17848== Address 0x7feffdfd8 is not stack'd, malloc'd or (recently) free'd ==17848== ==17848== Invalid read of size 8 ==17848== at 0x546EB52: ??? (setcontext.S:63) ==17848== by 0x5E4BD2C: (below main) (in /lib64/libc-2.12.2.so) ==17848== Address 0x7feffdfc8 is not stack'd, malloc'd or (recently) free'd ==17848== ==17848== Invalid read of size 8 ==17848== at 0x546EB59: ??? (setcontext.S:64) ==17848== by 0x5E4BD2C: (below main) (in /lib64/libc-2.12.2.so) ==17848== Address 0x7feffdf98 is not stack'd, malloc'd or (recently) free'd ==17848== timeout 1 Type in 4 chareters and press enter in> asdf ==17848== Syscall param readv(vector[...]) points to unaddressable byte(s) ==17848== at 0x5EF7A81: readv (in /lib64/libc-2.12.2.so) ==17848== by 0x42AF1C: boost::asio::detail::descriptor_ops::non_blocking_read(int, iovec*, unsigned long, boost::system::error_code&, unsigned long&) (descriptor_ops.ipp:153) ==17848== by 0x4360BF: boost::asio::detail::descriptor_read_op_base<boost::asio::mutable_buffers_1>::do_perform(boost::asio::detail::reactor_op*) (descriptor_read_op.hpp:55) ==17848== by 0x427CE4: boost::asio::detail::reactor_op::perform() (reactor_op.hpp:40) ==17848== by 0x429003: boost::asio::detail::epoll_reactor::run(bool, boost::asio::detail::op_queue<boost::asio::detail::task_io_service_operation>&) (epoll_reactor.ipp:286) ==17848== by 0x429C81: boost::asio::detail::task_io_service::do_one(boost::asio::detail::scoped_lock<boost::asio::detail::posix_mutex>&, boost::asio::detail::task_io_service::idle_thread_info*) (task_io_service.ipp:264) ==17848== by 0x42986F: boost::asio::detail::task_io_service::poll_one(boost::system::error_code&) (task_io_service.ipp:188) ==17848== by 0x42A0C2: boost::asio::io_service::poll_one(boost::system::error_code&) (io_service.ipp:103) ==17848== by 0x4255E1: receive(boost::asio::io_service&, boost::asio::posix::basic_stream_descriptor<boost::asio::posix::stream_descriptor_service>&, boost::asio::basic_streambuf<std::allocator<char> >&) (ccTalkScan.cxx:307) ==17848== by 0x425847: receive(boost::asio::io_service&, boost::asio::posix::basic_stream_descriptor<boost::asio::posix::stream_descriptor_service>&, unsigned long, boost::asio::basic_streambuf<std::allocator<char> >&) (ccTalkScan.cxx:320) ==17848== by 0x425982: main (ccTalkScan.cxx:334) ==17848== Address 0x65ba4f0 is 0 bytes inside a block of size 512 free'd ==17848== at 0x4C25C4F: operator delete(void*) (vg_replace_malloc.c:387) ==17848== by 0x432D07: __gnu_cxx::new_allocator<char>::deallocate(char*, unsigned long) (new_allocator.h:95) ==17848== by 0x4316BF: std::_Vector_base<char, std::allocator<char> >::_M_deallocate(char*, unsigned long) (stl_vector.h:146) ==17848== by 0x42FD64: std::_Vector_base<char, std::allocator<char> >::~_Vector_base() (stl_vector.h:132) ==17848== by 0x42D6A4: std::vector<char, std::allocator<char> >::~vector() (stl_vector.h:313) ==17848== by 0x42B657: boost::asio::basic_streambuf<std::allocator<char> >::~basic_streambuf() (basic_streambuf.hpp:114) ==17848== by 0x425A77: main (ccTalkScan.cxx:336) ==17848== out> 2 Type in 4 chareters and press enter in> timeout 3 Type in 4 chareters and press enter in> timeout 4 Type in 4 chareters and press enter in> timeout ==17848== ==17848== HEAP SUMMARY: ==17848== in use at exit: 168 bytes in 3 blocks ==17848== total heap usage: 65 allocs, 62 frees, 22,246 bytes allocated ==17848== ==17848== LEAK SUMMARY: ==17848== definitely lost: 0 bytes in 0 blocks ==17848== indirectly lost: 0 bytes in 0 blocks ==17848== possibly lost: 0 bytes in 0 blocks ==17848== still reachable: 168 bytes in 3 blocks ==17848== suppressed: 0 bytes in 0 blocks ==17848== Rerun with --leak-check=full to see details of leaked memory ==17848== ==17848== For counts of detected and suppressed errors, rerun with: -v ==17848== Use --track-origins=yes to see where uninitialised values come from ==17848== ERROR SUMMARY: 27 errors from 9 contexts (suppressed: 4 from 4) I have tried different things to fix, but I that I might have mis understood a thing or two here. So some help would be most appreciated. -- Allan W. Nielsen

2 8