February 2010 - Boost-users

Boost File Locks Do not work as advertised?
by Michael Dehmlow 12 Feb '10

12 Feb '10

I'm following the example at: http://www.boost.org/doc/libs/1_35_0/doc/html/interprocess/synchronization_… for locking and writing to a file, but when a flush is called either implicitly due to a full buffer, or an explicit flush no data is written and the fstream bad() call returns true. If I write a small enough amount of data and don't explicitly flush it works because I unlock the file then call close (where an implicit flush occurs). However, if a flush occurs at any time the file lock is locked no data is written and bad() return true. [START OF CODE] #include <fstream> #include <iostream> #include <boost/interprocess/sync/file_lock.hpp> #include <boost/interprocess/sync/scoped_lock.hpp> int main() { std::string tocFile("C:\\temp\\myfile.bin"); std::fstream file; file.open(tocFile.c_str(), std::ios::out | std::ios::binary | std::ios::trunc); if (!file.is_open() || file.bad()) { std::cout << "Open failed" << std::endl; exit(-1); } boost::interprocess::file_lock f_lock(tocFile.c_str()); { boost::interprocess::scoped_lock<boost::interprocess::file_lock> e_lock(f_lock); file.write(tocFile.c_str(),tocFile.length()); if (!file.is_open() || file.bad()) { std::cout << "Write failed" << std::endl; exit(-1); } file.flush(); if (!file.is_open() || file.bad()) { std::cout << "Flush failed" << std::endl; exit(-1); } } file.close(); return 0; } [END OF CODE] -- View this message in context: http://old.nabble.com/Boost-File-Locks-Do-not-work-as-advertised--tp2757007… Sent from the Boost - Users mailing list archive at Nabble.com.

1 0

PP: is macro expansion cached?
by Hicham Mouline 12 Feb '10

12 Feb '10

Hello, Say I define a macro #define MYMACRO(x,y) .. MYMACRO because of various reasons takes very long to expand (1mn) (uses PP_FOR, PP_WHILE, PP_SEQ_FOR_EACH .). If later in the translation unit, I have a "call" to MYMACRO MYMACRO(a,b) and then a couple of lines later MYMACRO(a,b) Will it be faster to define line 15: #define MYMACRO_ab MYMACRO(a,b) and then use line 25: MYMACRO_ab . line 35: MYMACRO_ab When does MYMACRO expansion happen? Is it at line 15 once, or at lines 25 and 35? Is the answer compiler-dependent? Regards,

1 0

PP: are there known incompatibilities with vs2008?
by Hicham Mouline 12 Feb '10

12 Feb '10

Hi, I have some PP macros that expand correctly under g++4.4 but when I came to try them on vs2008sp1, it generated errors. I will dig inside the code now, but if anyone is aware of msvc-specific problems... Regards,

1 0

abandoned named_mutex
by Carla Strembicke 12 Feb '10

12 Feb '10

Hello, My question is with regard to named_mutexes that are abandoned by a process that dies abnormally. Essentially I have shared memory( utilizing boost shared memory libraries ) and am incorporating boost named mutexes to limit access to one process at a time. There is no concept of a server/client or parent/child process....the process have equal opportunity. This is Microsoft centric code that I am required to support cross-platforms( Unix-like). Microsoft's mutexes have the concept of a mutex object being abandoned if the thread terminates without releasing. Is there any support in the boost mutex to determine if the mutex has been abandoned? Option: What I am considering is perhaps before the mutex is locked, it check a file that has been created with the mutex name and the pid of the process that has it locked. If the process does not exisit the file the mutex is unlocked and the file entry is removed and the new process id is entered. I am not sure if there is a chance of a race condition I would have to scope it out. FYI - I am using 1.41. is there any changes in 1.42 ( i will be checking ). Any other suggestions? Carla Strembicke Software Developer [cid:imagea03ff0.jpg@eeae924c.570740bf] SUBNET Solutions Inc. www.SUBNET.com<http://www.subnet.com/> Making Substations More Intelligent T: 403.270.8885 F: 403.270.9631 #100, 4639 Manhattan Road SE, Calgary, Alberta, Canada; T2G 4B3 Follow SUBNET: Twitter<http://twitter.com/subnetsolutions> | Facebook<http://www.facebook.com/subnetsolutions> | Linkedln<http://www.linkedin.com/groups?gid=2733031&trk=hb_side_g> Visit the all new SUBNET website at http://www.SUBNET.com<http://www.subnet.com/> CONFIDENTIAL INFORMATION NOTICE: The information contained in this e-mail is privileged, confidential and intended solely for the use of the addressee named above. If the reader of this e-mail is not the intended recipient, you are hereby notified that any dissemination, distribution or copying of this e-mail is strictly prohibited. If you received this in error, please contact the sender and destroy all copies of this e-mail. Thank you.

2 2

Asio sharedptr heap corruption
by Lloyd 12 Feb '10

12 Feb '10

Hi, My program uses Asio to connect to a remote server. The connection class make use of sharedptr() as shown below... class Connect:public boost::enable_shared_from_this<Connect> { .... } All the async calls use sharedptr() And the object is created like this... boost::shared_ptr<Connect> Conn=boost::shared_ptr<Connect>(new Connect(IoService)); When the destructor of Conn is called the program crashes with the error... First-chance exception at 0x7c81eb33 in MYPROGRAM.exe: 0x000006D9: There are no more endpoints available from the endpoint mapper. HEAP[MYPROGRAM.exe]: Heap block at 00369710 modified at 00369918 past requested size of 200 Windows has triggered a breakpoint in MYPROGRAM.exe. What could be the reason? (If I avoid using sharedptr() and let it leak the memory, the program does not crash) Please give some hints to come out of this problem... Thanks a lot, Lloyd ______________________________________ Scanned and protected by Email scanner

2 1

[type_traits] remove_pointer from member function types?
by Rutger ter Borg 12 Feb '10

12 Feb '10

Hello, although mentioned as a tip on the function_traits documentation page, it seems that remove_pointer doesn't work for member functions. It does work for normal function pointers. See attached an example. Is this the described/correct behaviour? If so, how do I easily transform a member function pointer type to a function type? Thanks, Rutger

1 0

[proto] using proto with nvcc
by Eric Jourdanneau 12 Feb '10

12 Feb '10

Hi everybody, Nice to meet you :) I would like to test proto with nvcc, and when I compile the simple Hello world example of 1.42, I get the following error message: /usr/include/boost/proto/detail/decltype.hpp:314: error: invalid use of template-name ‘boost::type’ without an argument list. Of course it works with g++ I just don't know what to do, can somebody help me on that point? Thanks Eric

10 25

how to know when eof has been reached?
by eial＠cs.bgu.ac.il 12 Feb '10

12 Feb '10

hello I've implemented a base64 decoder using input_filter, now for some reason I can't seem to cause it to end when reaching EOF. I need to read the entire that but using >> returns what is in between spaces, so I've created a function that loops on the stream to get the data. using a bool flag haven't worked because for some reason, the value isn't saved between the functions so I've tried with a specific char, that didn't worked too, how can I know when I've reached EOF? the headers and source can be found at: http://gitorious.org/open-source-soccer-manager/ossm/blobs/master/src/Utili… http://gitorious.org/open-source-soccer-manager/ossm/blobs/master/src/Utili… thanks.

2 2

[boost-build] error: Unable to find file or target named '/qt//QtGui'
by Tim Odenthal 11 Feb '10

11 Feb '10

Hi! After successfully compiling our project dozens of times, I suddenly get the error: Unable to find file or target named '/qt//QtGui'. I have absolutely no clue to what I could possibly have changed; I reverted our whole jamtree to the date before the last successful compilation... The jamfile in which the error is thrown looks something like this: import qt4 ; lib foo : $(SOURCE_ROOT)/bar.cpp (and many more...) /qt//QtGui /q//QtXml /qt//QtCore /qt//QtOpenGL : ; install install_foo : foo : <variant>release:<location>$(INSTALL_ROOT) <variant>debug:<location>$(INSTALL_ROOT)/debug ; I thought that the build tools of bjam then find the appropriate stuff and build it according to /usr/local/boost_1_42_0/tools/build/v2/tools/qt.jam which I obviously didn't change at all. I get the same error both with boost version 1_40_0 and 1_42_0. The stuff not dependent on qt seems to work fine... I'm working on a Kubuntu 9.10 box with gcc 4.4.1 - and as far as I can see (the history of my package manager admits), there was no change to the qt libs of Kubuntu during the time in question - but maybe the history of this package manager is not so reliable (I checked again that all the -dev packages are installed as well). Has anybody an idea what could have gone wrong? Is there another way to specify that bjam should include, compile and link these libraries? Thanks in advance Tim

2 1

Seeding boost::mt19937 with bytes from CryptoGenRandom
by e m 11 Feb '10

11 Feb '10

I have a C++ program that uses /dev/urandom on Unix systems to seed boost::mt19937. I'm porting the program to Windows. I can get good, random seed data on Windows similar to random seed data produced from /dev/urandom on Unix systems like so: HCRYPTPROV hProvider = 0; BYTE randomBytes[8]; CryptAcquireContext(&hProvider, 0, 0, PROV_RSA_FULL, CRYPT_VERIFYCONTEXT); CryptGenRandom(hProvider, sizeof(randomBytes), randomBytes); CryptReleaseContext(hProvider, 0); However, I'm running into trouble attempting to pass these randomBytes to boost::mt19937. On Unix I do something similar to this: uint64_t GetSeed() { uint64_t seed; std::ifstream urandom; urandom.open("/dev/urandom"); urandom.read(reinterpret_cast<char*> (&seed), sizeof(seed)); urandom.close(); return seed; } boost::mt19937 rng(GetSeed()); That works OK on Unix I think. Could anyone suggest how to get the random data from the Windows CryptoAPI passed to boost::mt19937 as seed data? I don't normally program Windows systems, so this has been a bit of a challenge for me. Thanks for any tips, Ed

3 10