November 2012 - Boost-users - lists.preview.boost.org

boost asio: boost::asio::deadline_timer blocking ?
by Avi Bahra 09 Nov '12

09 Nov '12

Using boost asio 1.47, suse linux 11.3, gcc 4.5 I am using boost::asio::deadline_timer, where the timer fires every second. When the timer expires, I check responses from the clients. I use the same timer to check when I have reach the minute boundary, so that server can do some other work. Now 99.9% of the time this works perfectly. However occasionally the timer does not fire every second, in fact it can delay/block for several minutes! This is what I use: timer_.expires_from_now( boost::posix_time::seconds( 1 ) ); timer_.async_wait( server_->io_service_.wrap( boost::bind( &NodeTreeTraverser::traverse,this,boost::asio::placeholders::error ) ) ); When I run strace on the process I see: epoll_wait(4, {{EPOLLIN, {u32=15957244, u64=15957244}}}, 128, 4294967295) = 1 timerfd_settime(5, 0, {it_interval={0, 0}, it_value={16, 944525000}}, {it_interval={16048400, 15957984}, it_value={140733847085872, 6025966}}) = 0 timerfd_settime(5, 0, {it_interval={0, 0}, it_value={0, 999992000}}, {it_interval={15958092, 6011901}, it_value={17195917584, 140733847086024}}) = 0 I have several questions: o Are there known problems with boost::asio::deadline_timer accuracy I realise I need soft real time, with one minute resolution. o Should I have two separate timers, one 1 second & for 1 minute o Is using timer_.expires_at(...) more reliable, I assume this affects how it configures timerfd_settime o Now I suspect this issue may be related to slow disk system Could a slow disk system affect timerfd_settime system call, since it is based on file descriptors. o Are there any well known asio based design patterns/idioms for ensuring reliable timer resolutions Any help appreciated. Ta, Avi

2 1

[thread] thread.join() throws in 1.52.0
by Szymon Gatner 09 Nov '12

09 Nov '12

Hi, after updating from 1.51.0 to 1.52.0 call to a thread::join() throws "invalid_argument" / not joinable. I see that there is a change with regards to thread d-tor in this version but I don't really understand if and why this is related. Should joinable() be called now before every join()? -- Szymon Gatner The Lordz Games Studio www.thelordzgamesstudio.com

3 7

Unit Test Framework Static library compilation
by Edouard Tallent 08 Nov '12

08 Nov '12

Hi all. This is Boost 1.52.0 and Visual Studio 2012 Express (VC11). I am stuck with the static compilation of the Unit Test Framework, as documented here: http://www.boost.org/doc/libs/1_52_0/libs/test/doc/html/utf/compilation/sta… I get the message 'bjam' is not recognized as an internal or external command, operable program or batch file while using the command bjam [-sTOOLS=] {-sBUILD=boost_unit_test_framework} as read in the documentation. Boost is not installed on my computer (header-only). And, maybe the library Boost.build HAS to be installed first? (if it so, how do I do that?) That would be great if someone give a clear explanation of that. Thank you very much, indeed. Édouard -- http://quantcorner.wordpress.com

3 3

[Boost units] Derived units for specific_heat_capacity
by A Andrew 08 Nov '12

08 Nov '12

2 2

ptr_map<U, T> with polymorphic U
by Victor Snyder 08 Nov '12

08 Nov '12

ptr_map<U, T> lets me map simple types to pointers-to-T, preserving polymorphic behavior of the mapped-to objects, and clearly stating that the ptr_map solely owns the pointed-to objects. I'd like to map pointers-to-U to pointers-to-T and have the map solely own the key objects as well as the mapped-to objects. ptr_map doesn't do this. I could map shared_ptrs-to-U to pointers-to-T, but then I lose the clear ownership semantics. Does anyone know of an easy workaround to get the behavior and semantics that I want, or is this more properly a feature request? Thanks!

2 2

pattern support for traversing inheritance chain
by Brian Budge 07 Nov '12

07 Nov '12

Hi all - I have a case where I have a largeish inheritance hierarchy. I will be calling an initialization function on all objects before using them in some fashion. Basically what I need to happen is when I call init() on a Foo the following would occur: void Foo::init() { super::init(); // now with the knowledge that anything I inherited is set up, I can set myself up... } It is not necessary for every call to implement init, in which case the behavior should be that the call is propagated up the inheritance chain (which it is by default in C++). In order to make this work, I need to write different code for "super" in every class implementation, which is a bit error prone. Even if I don't copy/paste, my brain might be on autopilot and type something further up the inheritance chain than it should be. Alternatively, I could require that each class provides a typedef for super, but I'm still relying on myself to remember to call super's init function. Is there anything in boost that would support this kind of hierarchy traversal (semi-)automatically? If there is not, are there suggestions for ways to implement this in a less error-prone fashion? Thanks, Brian

3 5

[1.52] PDF Documentation build available
by John Maddock 07 Nov '12

07 Nov '12

The PDF build of Boost docs is now available for 1.52 from: https://sourceforge.net/projects/boost/files/boost-docs/1.52.0/ John.

1 0

Virtual memory exhausted compile error with too many Boost state machine transitions
by Adlai Shawareb 06 Nov '12

06 Nov '12

All, We are getting the virtual memory exhausted error when compiling a Boost state machine file. If we cut back the number of state transitions in this state machine, we do not get the error, and the compile completes without errors. We are using Boost 1.49. Here is what we have done: 1) Allocated 4GB of memory to our Virtual Machine(it runs Ubuntu), and created a 4.5GB swap file. When I compile I look at the System Monitor dialog, and can see that the RAM memory is not even used up when we get the compile error. The virtual memory is never used. 2) I saw online that the 3.80 version of make had a bug that would cause this type of error. I've confirmed that we are using 3.81. I also tried 3.82, but get the same error. 3) I've confirmed that "ulimit -v" returns "unlimited" 4) Looking online, I saw a recommendation to add the following two lines to /etc/security/limits.conf. I've tried this but I still get the error. * hard rss unlimited * hard cpu unlimited 1) 2) 3) 4) 5) We have set up precompiled headers, and I've added SystemHSM.hpp, and associated Boost include files to our list of precompiled headers. However, I still get the error. 6) FYI - In the past we ran up against Boost's limit of 20 state transitions per state machine. To get around that, I've added the following to SystemHSM.cpp: #define BOOST_MPL_CFG_NO_PREPROCESSED_HEADERS // Discard the pre-generated headers and use preprocessor metaprogramming techniques to generate the necessary versions of the library components on the fly. #define BOOST_MPL_LIMIT_VECTOR_SIZE 50 // Increase the default (transition table) vector size beyond the standard limit of 20. #include <boost/mpl/vector.hpp> 1) 2) 3) 4) 5) 6) 7) FYI - We recently ran into an issue with the set size in the SystemHSM #define BOOST_FUSION_DONT_USE_PREPROCESSED_FILES #define FUSION_MAX_VECTOR_SIZE 30 #define FUSION_MAX_SET_SIZE 30 #include <boost/mpl/set.hpp> In addition I have looked for this type of error with gcc, but have not found anything concrete online. Any help would be appreciated!

1 0

Known way to handle this?
by Lemay.Steve 06 Nov '12

06 Nov '12

Is there already a known [proper] way to correct his issue (shown below)? I had reported it during the beta - this occurs many times for various source. "QCC" -Wc,-ftemplate-depth-128 -DFD_SETSIZE=2048 -march=k8-sse3 -mtune=k8-sse3 -mmmx -msse3 -mfpmath=sse -mno-3dnow -mstackrealign -Y_cpp -Wc,-fPIC,-fnothrow-opt,-std=c++11,-std=gnu++11 -O3 -Wc,-finline-functions -Wc,-Wno-inline -DBOOST_ALL_NO_LIB=1 -DNDEBUG -I"." -c -o "D:\boostBuild_QCC_1_52_0\boost\bin.v2\libs\serialization\build\qcc\release\link-static\target-os-qnxnto\threadapi-pthread\threading-multi\xml_iarchive.o" "libs\serialization\src\xml_iarchive.cpp" ...failed qcc.compile.c++ D:\boostBuild_QCC_1_52_0\boost\bin.v2\libs\serialization\build\qcc\release\link-static\target-os-qnxnto\threadapi-pthread\threading-multi\xml_iarchive.o... qcc.compile.c++ D:\boostBuild_QCC_1_52_0\boost\bin.v2\libs\serialization\build\qcc\release\link-static\target-os-qnxnto\threadapi-pthread\threading-multi\xml_archive_exception.o In file included from libs\serialization\src\xml_archive_exception.cpp:21:0: ./boost/archive/xml_archive_exception.hpp:35:44: error: looser throw specifier for 'virtual boost::archive::xml_archive_exception::~xml_archive_exception()' In file included from ./boost/archive/xml_archive_exception.hpp:25:0, from libs\serialization\src\xml_archive_exception.cpp:21: ./boost/archive/archive_exception.hpp:86:5: error: overriding 'virtual boost::archive::archive_exception::~archive_exception() throw ()' cc: d:/qnxsdk/qnxsdk6.5-001/host/win32/x86/usr/lib/gcc/i486-pc-nto-qnx6.5.0/4.7.2/cc1plus caught signal 1 Steven LeMay Game Platform Engineering International Game Technology 9295 Prototype Way Reno, NV 89521 Office: +1 775.448.8089 Mobile: +1 775.813.6676 steven.lemay(a)igt.com<mailto:steven.lemay@igt.com> This email message is for the sole use of the intended recipient(s) and may contain confidential and privileged information. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message. Any humorous implications are purely the result of misinterpretations provided by the reader.

2 2

Question about boost::locale translations
by Bastien Montagne 06 Nov '12

06 Nov '12

Hi everybody! :) I’m currently trying to replace libintl (from gnu gettext) by boost::locale lib, in a (mostly) C program (blender, actually). To do so, I wrote a simplistic wrapper: #include <boost/locale.hpp> #include "boost_locale_wrapper.h" static boost::locale::generator gen; /* Called only once */ void boost_locale_init(const char *messages_path, const char *default_domain) { gen.add_messages_path(messages_path); gen.add_messages_domain(default_domain); gen.set_default_messages_domain(default_domain); } /* Called each time you change locale */ void boost_locale_set(const char *locale) { if (locale && locale[0]) { std::locale::global(gen(locale)); } else { std::locale::global(gen("")); } } const char* boost_locale_gettext(const char *msgid) { return boost::locale::gettext(msgid).c_str(); } (this wrapper lib is linked statically with the main app) But I'm facing a big problem: I would expect strings returned by boost::locale::gettext & co to be "const", i.e. that the char pointer gotten by boost::locale::gettext(msgid).c_str() would be const (and unique to a given locale/msgid/context/etc.), as it is with gnu gettext. But this is obviously not the case, as demonstrated by those prints (translated_msgid (address)): Serbe latin (Srpski latinica) (0x7fcaed0dac58) sr_RS@latin (0x7fcaee1e5628) Suédois (Svenska) (0x7fcaed0dac58) sv_SE (0x7fcaee1e5628) Am I missing something obvious here, or are returned string just not "const"? If so, how can a code like: MessageBoxW(0,pgettext <http://www.boost.org/doc/libs/1_49_0/libs/locale/doc/html/group__message.ht…>(L"File Dialog",L"Open?").c_str(),gettext <http://www.boost.org/doc/libs/1_49_0/libs/locale/doc/html/group__message.ht…>(L"Question").c_str(),MB_YESNO); (taken from boost doc) could work? I'm quite new to all this C++2C stuff (and to boost too), so please forgive me if this is a noob question... Best regards, Bastien

3 5