January 2010 - Boost - lists.preview.boost.org

asio::ip::tcp::socket and boost::thread
by xutm 06 Jan '10

06 Jan '10

Hi all, I am using 1_39_0 version boost in my application, and encounter two problems, one with tcp::socket and another with thread. The following is my codes: class transStream { public: transStream() { cb = NULL; thrd = NULL; } ~transStream() { if (sock) { sock->close; // would be segment fault delete sock; // would be segment fault } if (thrd) { cout << "ready to delete the thread ..." << endl; //delete thrd; thrd->interrupt(); // actually, the thread still exists and runs thrd->detach(); cout << "detach the thread ..." << endl; thrd = NULL; } cb = NULL; cout << "finish..." << endl; } tcp::socket* sock; TransDataCallBack cb; boost::thread* thrd; }; static void getDataThread(transStream* ts) { if (!ts) { return ; } cout << "in getDataThread method ...." << endl; try { boost::system::error_code error; char packet[PACKET_LENGTH] = {0}; memset(packet, 0, PACKET_LENGTH); for (;;) { ... usleep(10); } cout << " exit the for loop " << endl; } catch(std::exception& e) { std::cerr << "Exception: " << e.what() << "\n"; } if (ts) delete ts; return ; } int CreateTransStream(OUT void** pTsHandle, IN const char* sTransIp, IN unsigned short servTransPort, IN const CONVERT_STREAM_INFO* pInfo) { if (!sTransIp || !pInfo) { cout << "error parameters" << endl; return -1; } stringstream ss; string port; transStream* ts = new transStream(); ... ts->sock = sock; *pTsHandle = ts; cout << "finish sending the dahua device info to server and return ..." << endl; return 0; } int StartTransStream(IN void* tsHandle, IN TransDataCallBack cb, IN void* userData) { stringstream ss; transStream* ts = (transStream*)tsHandle; if (!tsHandle || !ts) { return -1; } ... ts->cb = cb; boost::thread* thrd = new boost::thread(boost::bind(getDataThread, ts)); ts->thrd = thrd; ... return 0; } int StopTransStream(IN void* tsHandle) { transStream* ts = (transStream*)tsHandle; if (!tsHandle || !ts) { return -1; } ... return 0; } int DestroyTransStream(IN void* tsHandle) { transStream* ts = (transStream*)tsHandle; if (!tsHandle || !ts) { return -1; } cout << "in method DestroyTransStream ..." << endl; if (ts->thrd) { cout << "destory the thread ..." << endl; if (ts) { delete ts; cout << "reset ..." << endl; ts = NULL; } } cout << "deleted the socket ..." << endl; return 0; } 1. the first problem: in the function CreateTransStream, i declare a transStream using "transStream* ts = new transStream();" and a socket using "tcp::socket* sock = new tcp::socket(io_service);", "ts->sock = sock;" , then i return the ts using "*pTsHandle = ts;" ; When i delete the ts in the function DestroyTransStream: "delete ts;", i know the system would call the deconstruct function ~transStream(), when i call sock->close() or delete sock, Segment Fault will occurs. Can somebody tell me why? Or how can i solve the problem? 2. the second problem: If i want to kill a thread which is no longer useful, I think i just need to call the two lines: " thrd->interrupt();" and " thrd->detach();" . But actully, the two lines cannot kill the thread. Can somebody tell me how to do? Thinks in advances!

3 2

Re: [boost] Boost library submission (poll for interest)
by Bob Walters 06 Jan '10

06 Jan '10

On Jan 5, 2010, at 9:08 AM, Ion Gazta?aga <igaztanaga(a)gmail.com> wrote: > I think most in-memory databases use T-trees as main containers. I plan > to add them as Interprocess containers for late 2010, but of course, > that will depend on my free time. With those containers, we could speed > up searches and waste less space. Would the interface remain identical to map? The current 'trans_map' can be revised slightly so that it would be a transactional wrapper for any map implementation, leaving the internals of storage to the implementation. Taking advantage of a T-tree variant would then be very easy to accomplish. One of the biggest challenges I'm seeing with high throughput on map is that operations like insert() and erase() cannot be done to concurrently by multiple threads, and consequently those operations become a bottlneck. I'm doing operations like find() and updates (to different entries) in parallel. A map implementation which internally segmented itself into sections under independent mutexes in order to permit highly concurrent modifications of any kind would be a big help. Anyone know of any such implementations? - Bob

3 3

[ASIO] Windows named pipes
by Fernando Lemos 06 Jan '10

06 Jan '10

Hello, I've implemented support for Windows named pipes in Asio. Windows named pipes are created with CreateFile, and you can use WriteFile/ReadFile with overlapped I/O to read from and/or write to them. It uses win_iocp_handle_service (just like serial_port_service), so it's really simple code. I've attached the implementation to this e-mail (hoping the attachments will get through) so that you guys can import it into Boost if you find it worthwhile (feel free to license it under the Boost license). Note, though, that it needs to be "Boostfied" because I had to adhere to the coding conventions of the current project I was working on (this includes variable naming conventions, code style conventions, namespacing, and the separation of named_pipe into basic_named_pipe and named_pipe, and so on). And since this is my first interaction with Boost "internals", I won't be surprised if you guys find rookie mistakes in it. I've tested this implementation under Vista and Windows 7, using Boost 1.40 (but 1.41 should work just fine too) and VC++ 2008 Express. I used pipes in message mode, but I believe both message and stream modes will work. Regards,

1 0

[io] Tickets #1414, #3588, #3725 (warning fixes)
by Richard Webb 06 Jan '10

06 Jan '10

Does anyone care to merge the fix for #1414 (https://svn.boost.org/trac/boost/ticket/1414) to the Release branch? It looks like that change should fix both #3588 and #3725 as well. Thanks, Richard Webb -- View this message in context: http://old.nabble.com/-io--Tickets--1414%2C--3588%2C--3725-%28warning-fixes… Sent from the Boost - Dev mailing list archive at Nabble.com.

2 1

Re: [boost] Boost library submission (poll for interest)
by Bob Walters 06 Jan '10

06 Jan '10

On Jan 5, 2010, at 14:08:18 +0000 Mathias Gaunard <mathias.gaunard(a)ens-lyon.org> wrote: >> Currently, I've only implemented map<> (probably the most useful >> container type in a database), but the plan is to continue to add >> containers as requested/needed until ACID compliant versions of a >> decent portion of the STL are available. > > I have several times used Boost.MultiIndex as a replacement for a database. > It would be great if there could be support for everything > Boost.MultiIndex can do. I intend to look at this (future consideration). A variant of MultiIndex which works with the transactional infrastructure would be a naturally good fit. Offering support for MultiIndex would provide a means of consistency control between primary containers and associated secondary indexes which is currently lacking. - Bob

1 0

Review Request: fiber
by Oliver Kowalke 05 Jan '10

05 Jan '10

I'd like to request a formal review of boost.fiber - available at boost vault. Thx and a happy new year, Oliver

2 3

ublas questions
by markus2004x 05 Jan '10

05 Jan '10

Hi. Happy new year! I have couple of questions on uBlas. 1) On the boost user guide I read: "However unbounded_array is aimed at optimal performance. Therefore unbounded_array does not model a Sequence like std::vector does". I do not quite understand this statement. An std::vector is memory contiguous and access happens by offset. How can an unbounded beat this? 2) In the matrix<> template class I cannot find the concept of leading dimension. E.g. in a row major matrix 3x3 matrix the lead dimension could be 4, meaning that in reality I have a 3x4 matrix, but I just use a 3x3 (i.e. I have some slack at the end of each row). Is there a way to specify it? This is the way BLAS and LAPACK are designed, and it becomes particularly important when using hi-performance implementation of BLAS on x86, like Intel MKL, which require 16-byte alignment of each row. I thought about allocating a larger matrix, and then taking a range on it, but that does not seem the natural thing to do. Thanks a lot. -- View this message in context: http://old.nabble.com/ublas-questions-tp27024487p27024487.html Sent from the Boost - Dev mailing list archive at Nabble.com.

3 2

Msm review
by Nelson, Erik - 2 05 Jan '10

05 Jan '10

Not nagging, but I'm curious about the status/outcome of the Boost MSM review. I know everyone was busy over the holidays. Thanks Erik

1 0

Re: [boost] [Boost 1.41: list_of] Boost compliation error, > Sun Studio 11 using standard C++ lib
by shing fan 04 Jan '10

04 Jan '10

I see, thank you very much for this helpful feedback. Message: 4 Date: Sun, 3 Jan 2010 22:49:01 -0500 From: Stefan Teleman <stefan.teleman(a)gmail.com> To: boost(a)lists.boost.org Subject: Re: [boost] [Boost 1.41: list_of] Boost compliation error, > Sun Studio 11 using standard C++ lib Message-ID: <1ccb59a21001031949x3fcacbb7yd6cf56cc13ae5b30(a)mail.gmail.com> Content-Type: text/plain; charset=UTF-8 On Sun, Jan 3, 2010 at 22:41, shing fan <shingfan(a)hotmail.com> wrote: > There really is no drawback in using GCC. If you are running on SPARC and are concerned about the default GCC SPARC optimizer, you can download GCCFSS from here: http://cooltools.sunsource.net/gcc/index.html The only restriction is: Studio- and GCC- generated binaries are ABI incompatible. So, you won't be able to build Boost with GCC, and the rest of the software with Studio, and link. Also, C++ shared libraries provided in Solaris are built with Sun Studio. The same restriction explained above applies. --Stefan -- Stefan Teleman KDE e.V. stefan.teleman(a)gmail.com _________________________________________________________________ Windows Live: Keep your friends up to date with what you do online. http://go.microsoft.com/?linkid=9691815

1 0

[test] va_copy
by Steven Watanabe 04 Jan '10

04 Jan '10

AMDG Boost.Test is still failing on some compilers that don't support va_copy. Is va_copy always a macro? Would it be possible to use #ifdef va_copy? In Christ, Steven Watanabe

4 5