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

Access violation when setting up a serial port
by Tangleman 11 Mar '11

11 Mar '11

This feels like it should be a very simple solution, but I'm very much a Boost newbie, and relearning C++. I'm trying to set up a serial port in a class constructor of a legacy program I am updating. The serial port and io_service definitions are in the header file. The class is MFC. Header file declarations: boost::asio::serial_port port; boost::asio::io_service io; Here is the class constructor: CCplusplus_TesterDlg::CCplusplus_TesterDlg(CWnd* pParent) : CDialog(CCplusplus_TesterDlg::IDD, pParent), port(io, "COM3") { m_hIcon=AfxGetApp()->LoadIcon(IDR_MAINFRAME); } If I comment out the port and io declaration and the port initialization from the constructor, I can pass the constructor fine in debug mode. However, if I leave them in, I get the following: First-chance exception at 0x771a6b90 in Cplusplus_Tester.exe: 0xC0000005: Access violation writing location 0xccccccd0. Unhandled exception at 0x771a6b90 in Cplusplus_Tester.exe: 0xC0000005: Access violation writing location 0xccccccd0. Debug mode breaks in win_mutex.hpp at line 56: ::EnterCriticalSection (&crit_section); Go ahead. Make me feel like an idiot. Any clues as to what I am doing wrong? Thanks! -- View this message in context: http://boost.2283326.n4.nabble.com/Access-violation-when-setting-up-a-seria… Sent from the Boost - Users mailing list archive at Nabble.com.

1 1

fork() example asio 1_52_0
by Marco Piacentini 11 Mar '11

11 Mar '11

Hi...do you know what it means the parameters "_1" and "_2"? socket_.async_read_some(boost::asio::buffer(data_),boost::bind(&server::handle_read, this, _1, _2)); the complete example is: http://think-async.com/Asio/boost_asio_1_5_2/doc/html/boost_asio/example/fo…

3 2

Re: [Boost-users] [ASIO] async_accept : Error m_val 1.
by Cyprien DIOT 11 Mar '11

11 Mar '11

Thank you for your help but it didn't change anything. After few hours working on this problem, i've fix it: In a new VS2010 project, with the same source code, this code works without any error in async_accept callback. Finally, i find out that it didn't work because i was using pragma lib in one of my file for managing windows services: #pragma comment (lib, "advapi32.lib") #pragma comment (lib, "user32.lib") When I removed this file, error disappears. Then, I put theses lib in project's configuration instead of using pragma and everything works! Although I don't understand the problem, it's now working. Thank you for your help Cyprien DIOT ----- Mail original ----- De: "Igor R" <boost.lists(a)gmail.com> À: boost-users(a)lists.boost.org Envoyé: Jeudi 10 Mars 2011 15:58:17 Objet: Re: [Boost-users] [ASIO] async_accept : Error m_val 1. > What is "already open" ? _client ? It has just been allocated before calling async_accept. > _io is properly initialized. > > If this error is about acceptor, how can I use it with async_accept ? I've seen http://www.boost.org/doc/libs/1_37_0/doc/html/boost_asio/reference/basic_so… but i don't see any line where the listen port is set to acceptor. Like Viatcheslav mentioned, set reuse_address option: http://think-async.com/Asio/boost_asio_1_5_2/doc/html/boost_asio/reference/… _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

1 0

Re: [Boost-users] [Context] Mac OS X support?
by Julian Gonggrijp 10 Mar '11

10 Mar '11

Nat Linden wrote: > Anyway -- it seems likely to me that Boost.Context will support Mac OS > X 10.5+ "out of the box." I intend to check that. > > If it's important for Boost.Context to support Mac OS X <= 10.4, it > might be fairly straightforward to implement the Boost.Context API on > top of Pth. Reference [2] contains my macro-based attempt to establish > equivalences between the ucontext API and the Pth API. > > However, at this moment in history, one must ask how important it > still is to support Mac OS X <= 10.4? I'm completely ignorant about today's marketshare of Mac OS X 10.4, but I can tell you I'd rather like to see it supported for a while. Personally, I own an i686 machine with Mac OS X 10.6 and a PPC machine with Mac OS X 10.4. The latter is quite old and won't be able to pull anything more modern than 10.4. Therefore my only option for PPC testing is under Mac OS X 10.4. Since Apple's switch to x86 is historically close to the release of 10.5 this is probably true for more developers. Another, perhaps more general reason to support 10.4 is that, compared to both the older and the newer versions of Mac OS X it's still quite usable. I can't very well imagine anyone still using 10.3 but 10.4 doesn't seem strange. My (vague) impression is that some software suppliers still attempt to support 10.4 for this reason. Of course, both of these "arguments" could be easily overridden by some decent marketshare statistics. (Completely unrelated: hi, I'm Julian. This is my first post to the Boost users list.) (About equally related: what is the best way to reply to an existing thread? The instructions in the digest header don't seem to care much as long as you make sure that the subject field is correct. The Boost mailing lists website on the other hand suggests that one should rather not reply to digests at all and rather use per-message mailing or reply through the GMane web interface. Any opinions on this matter would be greatly appreciated.)

6 6

Boost on MS Win 7 Mobile
by Hossein Haeri 10 Mar '11

10 Mar '11

Dear all, Is there any benchmarking regarding Boost material's performance on Win 7 Mobile? Which of the Boost libraries are known to be available basically on Win 7 Mobile and which aren't? What sort of porting effort do I need to undertake, if any? I would highly appreciate it if anyone on the list can provide advice/pointers regarding my above questions. TIA, --Hossein

4 5

[Filesystem] v3 using wide url with boost::iostreams::mapped_file
by Bibbo, Joe 10 Mar '11

10 Mar '11

We recently flipped the switch from v2 to v3 of the boost file system. Now that boost:filesystem::wpath is no longer a valid type, how do you work with the wide url and boost:iostreams::mapped_file? With v2 we would void doSomethingWithMapfile(std::wstring& url) { boost::iostreams::basic_mapped_file_params<boost::filesystem::wpath> attribs; // initialize attribs, including attribs.path = url; // create our mapped file boost::iostreams::mapped_file mapFile(attribs); // An then do something with the map file. } Regards, Joe

3 2

using bind with find_if
by Dennis Jones 10 Mar '11

10 Mar '11

Hello, I am currently using find_if with bind2nd to perform a search on a vector. My current code looks like this: struct OverrideMatches : public std::binary_function<const TItemOverride, const TItemOverride, bool> { bool operator()( const TItemOverride& OverrideSource, TItemOverride OverrideTarget ) const { return OverrideSource.ChildAssyItemRecNo == OverrideTarget.ChildAssyItemRecNo && OverrideSource.LinkItemRecNo == OverrideTarget.LinkItemRecNo; } }; TItemOverrideGateway::Container_t::iterator where = std::find_if( FOverridesContainer.begin(), FOverridesContainer.end(), std::bind2nd( OverrideMatches(), TItemOverride( "", 0, "", 0, ChildItemRecNo, LinkItemRecNo ) ) ); I would like to eliminate the need for that temporary object to hold the values I am searching for (and just specify the values instead), and to do that, I want to try using boost::bind instead of std::bind2nd, but I have not been able to determing exactly what that syntax would look like. I know I will need placeholders for the values to search on (ChildItemRecNo, LinkItemRecNo), but then how does the element from the vector get passed to the I think I want the functor to look something like this: struct OverrideMatches { bool operator()( const TItemOverride &OverrideSource, const RecNo_t ChildItemRecNo, const RecNo_t LinkItemRecNo ) { return OverrideSource.ChildAssyItemRecNo == ChildItemRecNo && OverrideSource.LinkItemRecNo == LinkItemRecNo; } }; If the above functor is correct (and maybe it's not, in which case, please correct me), how do I then rewrite std::find_if to use it with boost::bind? Thank you very much, Dennis Jones

3 13

function_base.hpp type-punned pointer warnings
by KSpam 10 Mar '11

10 Mar '11

I get a ton of warnings (well over 5000 in the code base I am compiling) from function_base.hpp when compiling with gcc. boost-1.46.0/include/boost/function/function_base.hpp:321: warning: dereferencing type-punned pointer will break strict-aliasing rules boost-1.46.0/include/boost/function/function_base.hpp:325: warning: dereferencing type-punned pointer will break strict-aliasing rules This has been a known problem for a very long time, and the work around is quite simple. Replace reinterpret_cast<functor_type*>(&in_buffer.data)->~Functor(); with functor_type* p = reinterpret_cast<functor_type*>(&in_buffer.data); p->~Functor(); and replace reinterpret_cast<functor_type*>(&out_buffer.data)->~Functor(); with functor_type* p = reinterpret_cast<functor_type*>(&out_buffer.data); p->~Functor(); Would it be a reasonable request to have this simple patch applied to the boost source code? Respectfully, Justin

2 4

min and max preferring NaN
by Rhys Ulerich 10 Mar '11

10 Mar '11

'Afternoon, The usual std::min and std::max prefer numbers over NaNs per http://en.wikipedia.org/wiki/IEEE_754_revision#min_and_max. POSIX's fmin(3) does as well. Can anyone suggest a better implementation (chained ternary complaints aside) for a NaN-preferring min or max than brute force things akin to template<class T> inline const T& minnan(const T& a, const T& b) { return UNLIKELY((boost::math::isnan)(a)) ? a : (a < b) ? a : b; } where UNLIKELY is a small macro telling the compiler to generate code expecting that the boost::math::isnan test fails? The need arises in a numerical simulation where I want to gradually accumulate a global minimum where throwing away NaNs ("windowing" in IEEE 754 revision-speak) is the unacceptable. Thanks, Rhys

3 4

async_read_some/async_write_some
by Marco Piacentini 10 Mar '11

10 Mar '11

Hi...this is my question: are there the concatenation of these functions be considered like a multithread approach on a socket? I try to explain myself better... I need to have a behaviour like the following: inside a process, one thread read over a scoket, and one thread write over the same socket... is it possible to make the same beahviour simply create a chain of async_read and Async_write? or I have to handle the socket with lock-unlock mechanism? thank you...

1 0