- Boost-users - lists.boost.org

Re: [Boost-users] Strange boost::threads behaviour
by Gavin Lambert 20 Feb '15

20 Feb '15

On 19/02/2015 21:11, Lorenzo Trivelli wrote: >> >> Typically the way things work is that as long as a lock (or >> whatever it is you're using -- you didn't actually say) goes >> uncontended, or minimally contended such that a spin resolves it, >> then it will allocate no kernel handles. A kernel handle is only >> required when the lock is contended for sufficiently long that it >> needs to put the thread to sleep. > > thank you ver much for your response, > I'll try to detail the issue better. > The original software is an conccurrent transactional server which > preallocates a certain number of ftherads (say 100 ~ 130) and then > accepts and serves TCP/IP connections in a request-reply fashion. While thread-per-connection is a popular model for older or small-scale software (because it's easy to program), performance suffers quite significantly when a large number of connections are in use, especially when each connection needs to contend on the same locks. Especially if the workload is mostly I/O bound rather than CPU bound, you may want to consider a "properly" async design (eg. using Boost.Asio) where #threads == #cores (or thereabouts) and connection processing can shift between threads as needed. > Obviuosly, it heavily uses boost::shared_ptr to handle these connections, > but in the end I managed to use other shared pointers (for testing purposes only) > and the boost::thread object alone and a single shared static mutex object > > so, the sample program starrts, creates some threads which simply does some memory > allocations loops and s. In the main function I enclosed all the code between brackets, > and I print the handle count before and after, ans it's not the same, > something like > int main() { > GetProcessHandleCount().... > > { > [creates threads and waits for their termination through ->join] > > } > GetProcessHandleCount().... > > } > > The difference between the first and the last count is not always the same, it > depends on how much concccurent threads you create and how much iterations of the > same loop each thread does. Well, as I said that is not an error if some synchronization objects (eg. mutexes) still exist.

1 0

"xtime" ambiguous symbol in MSVC 12
by Kevin Williams 19 Feb '15

19 Feb '15

Hi I get a C2872 when including "chrono" (allegedly line 738) from MSVC 12, and also needing boost::mutex. I believe the boost mutex is a bit faster than std::mutex, and I need that. Many thanks, Kevin

2 1

Re: [Boost-users] Strange boost::threads behaviour
by Gavin Lambert 19 Feb '15

19 Feb '15

On 17/02/2015 04:01, Lorenzo Trivelli wrote: > While analyzing a resource exaustion problem on a production machine, > (Windows) I found out astrange behaviour of boost::threads regarding event object. > The problem is that after a certain number of iterations it semms that the > thread lib allocates an EventHandle which is never freed. > I managed to reproduce in some ways the behaviour, but I'm a bit clueless > about its reason. Typically the way things work is that as long as a lock (or whatever it is you're using -- you didn't actually say) goes uncontended, or minimally contended such that a spin resolves it, then it will allocate no kernel handles. A kernel handle is only required when the lock is contended for sufficiently long that it needs to put the thread to sleep. This handle should be released when the lock itself is deleted. If the lock is itself global or otherwise held by a global object, or if it's in an object that you are memory-leaking, then it won't go away until you close the process. > 3) if you want I can attach the sample code with a but if info > on how to reproduce the behaviour Without knowing at least *which* threading object you're using, I doubt a report will get anywhere.

2 1

Need help to build ( cross compile) Boost for ARM
by som jagadeeshan 18 Feb '15

18 Feb '15

Hello All, Currently I’m trying to get the boost_1_57_0 library built for ARM. But i don’t find any help on how to go about this. I'm using the poky hard float tool chain to get the library cross compiled(*arm-poky-linux-gnueabi-*) Any help on how to get this will be great. The below log is what i get when running the script file ( bootstrap.sh) $ ./bootstrap.sh *Building Boost.Build engine with toolset gcc... Failed to build Boost.Build build engineConsult 'bootstrap.log' for more details* Thanks, Mike

2 2

Boost MultiIndexContainer, how to do delayed indexing for fast insertion?
by HuMichael 17 Feb '15

17 Feb '15

I used below MultiIndexContainer typedef multi_index_container<PositionSummary*, indexed_by< ordered_unique< composite_key<PositionSummary, const_mem_fun<PositionSummary, int, &PositionSummary::positiondate>, const_mem_fun<PositionSummary, const std::string&, &PositionSummary::accountid>, const_mem_fun<PositionSummary, const std::string&, &PositionSummary::instid> > >, ordered_unique< composite_key<PositionSummary, const_mem_fun<PositionSummary, int, &PositionSummary::positiondate>, const_mem_fun<PositionSummary, const std::string&, &PositionSummary::instid>, const_mem_fun<PositionSummary, const std::string&, &PositionSummary::accountid> > > > > PositionSummaryContainer; And I inserted 10000 insts*36 accounts*100 days=36 million records //Begin testing of the multiIndexContainter std::cout << "Begin inserting data from array into the multiIndexContainter" << std::endl; timer.reset(); timer.begin(); for (int i = 0; i < numOfDays_; i++) { for (int j = 0; j < accountSize_; j++) { for (int k = 0; k < instSize_; k++) { PositionSummary* ps = psArray_[(i * accountSize_ + j) * instSize_ + k]; uniqueIndex.insert(ps); } } } printMemoryUsage(); timer.end(); std::cout << "Time take is " << timer.getInterval() << std::endl; And I found the speed of insertion is a little bit slow, about 20K+ records per second... Is there anyway to enhance this insertion speed? My data was in Oracle, properly indexed, so there should be no danger of corrupted data structure. I knew that in oracle you can first load then build index to save time, can I do the same with MultiIndexContainer, if there is a way? By the way, the parallel query speed is quite satisfactory, querying all the 36 m records on a 4 cpu(8kernal) machine takes only 2.8 seconds, code as below #pragma omp parallel for collapse(2) for (int i = 0; i < numOfDays_; i++) { for (int j = 0; j < accountSize_; j++) { const int& date = dates_[i]; const std::string& accountID = accountIDs_[j]; for (int k = 0; k < instSize_; k++) { const std::string& instID = instIDs_[i]; PositionSummaryContainer::iterator it = uniqueIndex.find(boost::make_tuple(date, accountID, instID)); if (it != uniqueIndex.end()) { #pragma omp atomic sum2 += (*it)->marketvalue(); } } //std::cout << "accountID: " << accountID << std::endl; } }

2 2

[gsoc15] Org application closes on Friday 20th
by Niall Douglas 17 Feb '15

17 Feb '15

Three days remain until the GSoC org application closes on Friday 20th of April. As can be seen at https://svn.boost.org/trac/boost/wiki/SoC2015, we have just SIX project proposals and FIVE mentors who have publicly made themselves available on the ideas list this year. This is significantly lower than this time last year. If you would like to attract students to work on your Boost library in 2015, please consider adding a formal project idea to the list above before Friday. If you would like to mentor in some topic or field, please make yourself known to us so we can add you to the list of available mentors for 2015. Oh, and don't forget to mention the fields you are interested in! Thanks, Niall and Boris --- Boost C++ Libraries Google Summer of Code 2015 admin https://svn.boost.org/trac/boost/wiki/SoC2015

1 0

generic vector
by U.Mutlu 17 Feb '15

17 Feb '15

Hi, I need to have a std::vector in my program, but it's type will be known only when it runs, ie. the type will be provided via a program param. The possible types are limited to these: int64_t, uint64_t, double, std::string. Is there a solution in boost or C++11 for this kind of problem? -- TIA Uenal

4 3

[regex] problem w/ regex::extended
by U.Mutlu 16 Feb '15

16 Feb '15

Hi, the following regex expression works in perl-mode, but does not work in extented-mode. What's the reason? I think I haven't used any perl-specific construct, or did I? Where? ... #define NUM "[+-]?(\\d+\\.?\\d*|\\.\\d+)([eE][+-]?\\d+)?" #define WORD "[\\w\\-\\.~/]+" #define EQU "\\s*\\=\\s*" const string sUserRE_VarAndVal = WORD EQU NUM; #if 0 #define REGEX_ENGINE boost::regex::extended #else #define REGEX_ENGINE boost::regex::perl #endif regex reNum(sUserRE_VarAndVal, REGEX_ENGINE); ...

2 4

[program options] option_groups.cpp example doesn't compile
by U.Mutlu 16 Feb '15

16 Feb '15

I've boost 1.55 from the debian repository. I'm trying to compile the below example of the 1.55 version (it's same as that of 1.57), but I get the following compile errors. Should I use some 'magic' compiler options, or what else is the reason for this grandious failure? $ g++ -Wall -std=gnu++11 option_groups.cpp -lboost_program_options option_groups.cpp: In function ‘int main(int, char**)’: option_groups.cpp:44:46: error: ‘class boost::program_options::typed_value<std::basic_string<char>, char>’ has no member named ‘implicit’ ("help-module", value<string>()->implicit(), ^ option_groups.cpp:96:11: error: reference to ‘exception’ is ambiguous catch(exception& e) { ^ In file included from /usr/include/c++/4.9/new:40:0, from /usr/include/c++/4.9/ext/new_allocator.h:33, from /usr/include/x86_64-linux-gnu/c++/4.9/bits/c++allocator.h:33, from /usr/include/c++/4.9/bits/allocator.h:46, from /usr/include/c++/4.9/string:41, from /usr/include/boost/program_options/errors.hpp:12, from /usr/include/boost/program_options/options_description.hpp:12, from option_groups.cpp:24: /usr/include/c++/4.9/exception:60:9: note: candidates are: class std::exception class exception ^ In file included from /usr/include/boost/throw_exception.hpp:43:0, from /usr/include/boost/any.hpp:26, from /usr/include/boost/program_options/value_semantic.hpp:12, from /usr/include/boost/program_options/options_description.hpp:13, from option_groups.cpp:24: /usr/include/boost/exception/exception.hpp:208:5: note: class boost::exception exception ^ option_groups.cpp:96:11: error: ‘exception’ does not name a type catch(exception& e) { ^ option_groups.cpp:97:17: error: ‘e’ was not declared in this scope cout << e.what() << "\n"; ^

2 4

[context] boost::coroutines::asymmetric_coroutine<T>::pull_type iterator does not properly model InputIterator
by Mueller-Roemer, Johannes Sebastian 16 Feb '15

16 Feb '15

I’m currently on Boost 1.56 so this might have been fixed by now, but in a simple test Boost.Coroutine fails to properly model the InputIterator concept, because *it++ does not conform to the operational semantics { T tmp = *it; ++it; return tmp; } required by the standard. The failing program: #include <boost/coroutine/asymmetric_coroutine.hpp> #include <iostream> template<typename Continuation> auto foo(Continuation && yield) -> void { yield("Hello"); yield(", "); yield("World"); yield("!\n"); } //using T = char const *; using T = std::string; auto foo() -> boost::coroutines::asymmetric_coroutine<T>::pull_type { using namespace boost::coroutines; return asymmetric_coroutine<T>::pull_type{ [](asymmetric_coroutine<T>::push_type& yield) { foo(yield); } }; } auto main(int, char **) -> int { auto f = foo(); for(auto it = boost::begin(f); it != boost::end(f);) std::cout << *it++; } Which (on my machine) outputs: , World! X☺à r2 ` , ` , Ç2 Ç Failing to model the InputIterator concept is bad, because then it isn’t any kind of standards-conforming iterator at all. Interestingly, if T is char const * and not std::string the program works as expected. Is there some other subtle bug at work here? -- Johannes S. Mueller-Roemer, MSc Wiss. Mitarbeiter - Interactive Engineering Technologies (IET) Fraunhofer-Institut für Graphische Datenverarbeitung IGD Fraunhoferstr. 5 | 64283 Darmstadt | Germany Tel +49 6151 155-606 | Fax +49 6151 155-139 johannes.mueller-roemer(a)igd.fraunhofer.de | www.igd.fraunhofer.de

2 8