April 2011 - Boost - lists.boost.org

[vault] Boost Vault is down
by Edward Diener 27 Apr '11

27 Apr '11

It appears the link to the Boost vault is down. Attempting http://www.boostpro.com/vault/index.php gives the "Server not found" message. Can this be fixed fairly ASAP ? The distribution for the next review, starting tomorrow, is in the vault ( no surprise there ), and since I am the review manager I have already gotten an e-mail pointing out that the distribution can not be accessed.

4 4

[GSoC] Results
by Andrew Sutton 26 Apr '11

26 Apr '11

All, I'd like to thank all of the students who submitted proposals this year. We had a number of excellent proposals, many more than last year. Ultimately, we were only able to accept 10 proposals and so some deserving proposals could not be accepted. If your proposal was not accepted, and you would like to know why, please send me an email. I will be more than happy to discuss the results with you. For the students whose proposals were accepted, tomorrow starts the "Getting to know you" phase of GSoC. Assuming that you will be done with exams soon, you should be getting started with the Boost infrastructure, build system, and documentation tools. Questions about any of the above can be directed to your mentors or the mailing list --- hopefully both. I will send out some instructions this week for getting started with the Boost SVN repository. I'd also like to thank all of the mentors who contributed to the reviewing process. The reviewing seems to have gone (more or less) smoothly, and the quality of the reviews was excellent. Thanks again, Andrew

1 0

Re: [boost] [MSM] Interrupt anonymous transitions or blocking process_events?
by Christophe Henry 26 Apr '11

26 Apr '11

>> Depending on how you use it, yes. I'm a bit worried with race >> conditions in case your state machine is also used in another thread, >> which would also call process_event. > > Yes, I protect every call to process_event with a mutex. Since I don't > have that many events I have created a member function that wraps and > protects the process_event call. (One member function per event.) Usually, the destructor gets tricky. Not only, but this one is hard to get right. >> Alternatively (if you want to avoid locking/deadlocking/rc), you could >> also move the worker thread out of the fsm and have both communicate >> with synchronized message queues. You might want to have a look at >> boost::circular_buffer, which provides an example >> (http://www.boost.org/doc/libs/1_46_1/libs/circular_buffer/doc/circular_buff…) > > Yes, that is an interesting idea. But how do you solve the problem with > passing events on to the message queue? The events are all of different > types. Do you mean that there should be some parsing of strings or some > other identifier? You can make a circular buffer of, say, boost::function<void()>, this allows you to anonymize functors, for example this can be assigned into this boost::function: template <class Fsm, class Event> struct process_some_event { process_some_event(Fsm* fsm): fsm_(fsm){} void operator()() { fsm_->process_event(Event()); } Fsm* fsm_; }; Or as a variant, you can make a boost::function<void(my_fsm&)> with the corresponding operator(). Whatever fits better in your case. Regards, Christophe

2 1

Darwin Clang help needed
by John Maddock 26 Apr '11

26 Apr '11

There are some strange failures in Boost.Math for Clang on Darwin, for example: boost::math::binomial_coefficient<long double>(42, 15); Appears to be returning zero. The code is pretty simple, so is anyone able to step through it and see what's going wrong (it seems to be working OK for type double BTW). Thanks, John.

2 2

[asio] [date_time] boost::asio::deadline_timer performance
by Isidor Zeuner 25 Apr '11

25 Apr '11

Hi all, while profiling a boost::asio-based application with heavy use of timers, I noticed many CPU cycles being burnt inside a gregorian::date::date constructor, which I traced down as being called from asio::deadline_timer::expires_from_now. As asio::deadline_timer just specializes asio::basic_deadline_timer using posix_time::ptime, I assume this can be tuned by using another time implementation. Is there a particular implementation targeted at low overhead, ideally with high accuracy? Are there other ways of tuning timer performance? Best regards, Isidor -- Isidor Zeuner http://quidecco.de/cms/blog/

1 0

[accumulators]
by Edward Diener 25 Apr '11

25 Apr '11

Attempting to run the accumulators regression test on the trunk, using 'bjam toolset=gcc-4.5.2' ( MingW/gcc-4.5.2 ) is giving me a number of pop-up Windows in Vista saying that "cc1plus.exe has stopped working". Cc1plus.exe is the gcc compiler under MingW. In the output there are a number of failures. Does anybody know why this is occuring ?

1 0

[Pack Pointer] Review Request
by Phil Bouchard 25 Apr '11

25 Apr '11

Greetings, I've followed what users were dissatisfied with and I turned out renaming Shifted Pointer to Pack Pointer. Furthermore I've added a small tutorial that can be found at: https://svn.boost.org/svn/boost/sandbox/pack_ptr/libs/smart_ptr/doc/tutoria… Once again the main documentation is at: https://svn.boost.org/svn/boost/sandbox/pack_ptr/libs/smart_ptr/doc/index.h… I still need cleaning up some code but it works well with complex examples and I would like queuing the library up for a review if possible. Regards, -Phil

2 2

[build] bjam library path on gcc
by Lorenzo Caminiti 25 Apr '11

25 Apr '11

Hello all, How do I setup gcc library path (LD_LIBRARY_PATH) for bjam? For example I get this error when using bjam: lcaminiti@lcaminiti-PORTEGE-R700:~/sandbox/boost-sandbox-local/libs/local/example$ bjam -q profile_boost_local toolset=gcc ...patience... ...found 790 targets... ...updating 2 targets... gcc.compile.c++ bin/gcc-4.5.1/debug/profile_boost_local.o gcc.link bin/gcc-4.5.1/debug/profile_boost_local bin/gcc-4.5.1/debug/profile_boost_local.o: In function `main': /home/lcaminiti/sandbox/boost-sandbox-local/libs/local/example/profile_boost_local.cpp:17: undefined reference to `boost::chrono::system_clock::now()' /home/lcaminiti/sandbox/boost-sandbox-local/libs/local/example/profile_boost_local.cpp:23: undefined reference to `boost::chrono::system_clock::now()' /home/lcaminiti/sandbox/boost-sandbox-local/libs/local/example/profile_boost_local.cpp:31: undefined reference to `boost::chrono::system_clock::now()' /home/lcaminiti/sandbox/boost-sandbox-local/libs/local/example/profile_boost_local.cpp:33: undefined reference to `boost::chrono::system_clock::now()' bin/gcc-4.5.1/debug/profile_boost_local.o: In function `__static_initialization_and_destruction_0': /home/lcaminiti/sandbox/boost-trunk/boost/system/error_code.hpp:214: undefined reference to `boost::system::generic_category()' /home/lcaminiti/sandbox/boost-trunk/boost/system/error_code.hpp:215: undefined reference to `boost::system::generic_category()' /home/lcaminiti/sandbox/boost-trunk/boost/system/error_code.hpp:216: undefined reference to `boost::system::system_category()' collect2: ld returned 1 exit status "g++" -o "bin/gcc-4.5.1/debug/profile_boost_local" -Wl,--start-group "bin/gcc-4.5.1/debug/profile_boost_local.o" -Wl,-Bstatic -Wl,-Bdynamic -Wl,--end-group -g ...failed gcc.link bin/gcc-4.5.1/debug/profile_boost_local... ...failed updating 1 target... ...updated 1 target... But I can tell bjam to add -L and -l options as in the following it would work: lcaminiti@lcaminiti-PORTEGE-R700:~/sandbox/boost-sandbox-local/libs/local/example$ g++ -o "bin/gcc-4.5.1/debug/profile_boost_local" -Wl,--start-group "bin/gcc-4.5.1/debug/profile_boost_local.o" -Wl,-Bstatic -Wl,-Bdynamic -Wl,--end-group -g -L/home/lcaminiti/sandbox/boost-trunk/stage/lib -lboost_chrono -lboost_system Thank you. -- Lorenzo

2 1

How to use ASIO without SSLv2?
by Steve M. Robbins 25 Apr '11

25 Apr '11

Hi, Debian just turned off SSL v2 support in openssl due to security concerns [1]. Unfortunately, this broke compilation of several ASIO-using programs such as pion-net [2] and witty [3]. These programs all fail to compile with the following diagnostics: > In file included from /usr/include/boost/asio/ssl/context_service.hpp:30:0, > from /usr/include/boost/asio/ssl/context.hpp:22, > from /usr/include/boost/asio/ssl.hpp:19, > from /build/user-witty_3.1.8-2-amd64-wHJ4Kv/witty-3.1.8/src/http/Server.h:22, > from /build/user-witty_3.1.8-2-amd64-wHJ4Kv/witty-3.1.8/src/http/HTTPStream.C:12: > /usr/include/boost/asio/ssl/detail/openssl_context_service.hpp: In member function 'void boost::asio::ssl::detail::openssl_context_service::create(SSL_CTX*&, boost::asio::ssl::context_base::method)': > /usr/include/boost/asio/ssl/detail/openssl_context_service.hpp:73:28: error: '::SSLv2_method' has not been declared > /usr/include/boost/asio/ssl/detail/openssl_context_service.hpp:76:28: error: '::SSLv2_client_method' has not been declared > /usr/include/boost/asio/ssl/detail/openssl_context_service.hpp:79:28: error: '::SSLv2_server_method' has not been declared What does one have to do in order to use ASIO without SSLv2 support? One suggestion [3] is to simply enclose the related code within #ifndef OPENSSL_NO_SSL2. In the cse of openssl_context_service::create(), however, placing this around the affected switch cases would result in executing the default case default: impl = ::SSL_CTX_new(0); break; Would this be safe to do? Are there other places in the code that need to be addressed? Thanks, -Steve [1] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=589706 [2] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=621402 [3] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=622070

1 0

Re: [boost] [locale] Formal review of Boost.Locale library -- final days
by Artyom 24 Apr '11

24 Apr '11

> From: Gevorg Voskanyan <v_gevorg(a)yahoo.com> > Points I don't like most: > > - The boundary interface isn't attractive (in my opinion, shared by some other > > > reviewers) As it was common request that was shared by several reviewers I'll take a look on it. And I'll also take a look on regex like interface that may actually be relevant for this case. > - Operator overloading is sometimes used in unintuitive ways. For example > operator/ used to get the value for a specific period type > I'll also update a little date_time the interface to be consistent with other boost libraries, for example to get year(t) t=year(1990); In addition or instead of t = year * 1990 t / year > [snip] > > I noticed lack of exception safety in some places and places where RAII > should have been used. The implementation might benefit from a careful > examination intended to find and fix such kind of problems. I've seen several places that were pointed by reviewers and I had seem some problems with copy constructors in the code so I'll review them. > I also think that > dynamic memory management is used too freely - in many cases their usage is > unavoidable but in some cases they are. Having said that I appreciate that >some > > efforts have already been put in that direction: I noticed several places >where > > fixed-size buffers were used if data is small enough and dynamic allocation if > > > it's larger. > > More details at http://permalink.gmane.org/gmane.comp.lib.boost.devel/218352 > Noted. > > - What is your evaluation of the documentation? > > The documentation structure is good. It enables a new user to get started with > > > the library quickly. The well-organized references help to find documentation > for a specific facility in almost no time. I would however like to see more > links within the text of the pages. There are many spelling and grammar issues > > > in the documentation. Many of them have been already addressed to some extent >by > > me and other reviewers, but it feels like there are still more of them left > there. I agree that the documentation should be thoroughly scanned for such > language issues by a native English speaker. > I also think there should be more examples. > > More details at http://permalink.gmane.org/gmane.comp.lib.boost.devel/218117 >and > > http://permalink.gmane.org/gmane.comp.lib.boost.devel/218343 > I'll do my best, hopefully after third language review you did it would be much better. > > - What is your evaluation of the potential usefulness of the library? > > Very useful. > > > - Did you try to use the library? With what compiler? Did you have any > > problems? > > I have played with the library using number/ordinal formatting, date >formatting, > > currency formatting, spellout, case conversion and boundary analysis features. > >I > > have not used encoding conversion, normalization, timezone/calendar, (message) > > > formatting and collation features. I have used ICU backend only. > > Environment: > FreeBSD 8.0 > g++ 4.5.0 in C++03 mode > ICU 3.8.1 system-wide installation > > What I used was intuitive and easy, except for boundary analysis as noted >above. > > I had some incorrect results most of which I believe are ICU issues. > Some which I'm not so sure about though: > - output of as::local_time was wrong by 2 hours; as::gmt was correct. >Moreover, > > as::date printed the yesterday's date when the local time was 2:24 AM ! > /bin/date shows correct local time, so I believe my system time zone settings > are fine. This is problem with older ICU versions. ICU had included several bugs in the local time zone detection. I assume that the test test_icu_of_os_timezone fails. I've seen things things on FreeBSD: http://art-blog.no-ip.info/files/nightly-build-report.html (Even it is not tested under Boost namespace it still catches the problems) What you need is latest ICU. It had several fixes of timezone detection in 4.6 and this version works fine and passes all the tests. So I recommend you to use it with ICU 4.6 on FreeBSD. There is a some workaround I implemented for Linux for other problem of detecting timezone as "localtime" but it seems that on FreeBSD it is other issue as it has little bit different timezone handling. > - "Note: not all locales provide rules for spelling numbers. In such a case >the > > number would be displayed in decimal format." This was not the case: > std::cout << as::spellout << 117 << std::endl; printed "one hundred and > seventeen" under hy_AM.UTF-8 locale. I'll take a look on this, it may be either CLDR issue or ICU mis-detecting hy_AM locale. > > There were loads of "comma at end of enumerator list" warnings. > > > - How much effort did you put into your evaluation? A glance? A quick > > reading? In-depth study? > > Closer to in-depth study with more than 25 hours invested. > > > - Are you knowledgeable about the problem domain? > > I am well aware of localization problems and approaches that can be used to > tackle them. I have developed in-house solutions to address some localization > needs on as-necessary basis. > > > And finally, every review should answer this question: > > > > - Do you think the library should be accepted as a Boost library? > > Yes. Thank you very much for the review. Artyom

2 1